ScentMatch · Instagram Growth OS · 2026

The Organic Growth Playbook

A ruthlessly practical, execution-first system built from deep analysis of ScentMatch's product, positioning, and audience. Every recommendation is measurable. Every action is tied to a KPI. Zero generic advice.

🔍
Brand Diagnostic
Business analysis + social presence audit — the foundation for everything below
Step 0
Business Model & Product
What ScentMatch Is
  • AI-powered fragrance advisory platform — olfactory profiling quiz → personalised scent recommendations
  • Database of 23,000+ fragrances across all price tiers (Drugstore → Ultra Niche)
  • Tools: collection tracker, side-by-side comparison, weather-aware recommendations, note pyramid explorer
  • Currently free, web-based — positioning for premium advisory / eventual monetisation
  • Operated by a single founder/CEO (34F, collector profile, deep olfactory expertise)
Core USP
  • "Precision Fragrance Advisory" — science + psychology + data, not hype or bestseller lists
  • Olfactory profile maps 8 axes (citrus, floral, woody, spicy, sweet, fresh, dark, musky) + psychological profiling (47-question depth)
  • Occasion, season, climate, skin type, and memory-trigger intelligence built in
  • Tone: expert, evidence-based, premium — "fragrance as identity, not product"
Social Presence Audit
Instagram · @scentmatch_official Primary Target
Gap No content data available — account appears to be early-stage or inactive. Zero established content cadence signals.
Gap No Reels presence identified. In 2026, accounts without a consistent Reels output are invisible to the algorithm.
Opportunity Clean slate advantage — no legacy content to contradict. Can establish authority positioning from Day 1 with correct strategy.
Opportunity Fragrance niche is underserved on Reels — most accounts are either luxury brand content (unrelatable) or basic "best drugstore fragrance" content (low authority). The expert-with-data position is open.
LinkedIn · ScentMatch Company Page Secondary
Strength 10 editorial posts planned/drafted — evidence-based, research-cited, authoritative. Topics: AI + olfaction, fragrance psychology, niche revolution, gender fluidity in scent.
Strength Strong written voice — long-form, intelligent, cites Chemical Senses, Nature, ACM RecSys. Positions founder as domain expert, not hobbyist.
Gap LinkedIn ≠ Instagram — this content cannot be repurposed directly. LinkedIn tone is B2B/thought-leadership; Instagram requires entertainment-first, educational-second.
Opportunity Cross-reference authority signals — the LinkedIn intellectual credibility can be referenced in IG bio/captions to elevate trust ("Fragrance advisor. 23K fragrances. Built by a collector.").
🚧 Specific Weaknesses Blocking Growth
  • No video content infrastructure — no established filming setup, style, or template means Week 1 output will be slow
  • Founder-led, no team — single person handling creation, editing, posting, and engagement is a bottleneck requiring strict time-boxing
  • Product is web-only — no app = harder to drive "try it now" conversion from IG; link in bio must do heavy lifting
  • Abstract value prop — "AI fragrance advisory" is not immediately visual or emotionally triggering; requires translation into concrete, compelling hooks
  • Niche credibility vs. mass reach tension — the expert positioning attracts a narrow audience; viral growth requires broader emotional hooks without diluting brand
  • No existing audience to seed engagement — early posts will have near-zero organic reach; must compensate with aggressive manual engagement and strong hook rates
  • Platform cold start — Instagram algorithm requires 4–6 weeks of consistent output before it learns what audience to serve content to
✅ Specific Advantages to Leverage
  • 23K fragrance database — no competitor has data depth like this; enables "this fragrance has X in common with Y" content nobody else can make
  • Real collector perspective — CEO is a genuine enthusiast (patchouli, labdanum, animalic profile, niche collector). Authenticity is detectable and algorithmically rewarded
  • Olfactory profiling system — interactive quiz creates a content hook: "What does your fragrance profile say about you?" (high share/save trigger)
  • Fragrance community is highly engaged — #perfumecommunity, #fragrancetok cross-platform audiences have 8–12% avg engagement (vs 2–3% industry average)
  • No-cost product demos — the web app can be screen-recorded for content; zero production budget needed for "product in action" Reels
  • Gender-neutral, psychology-first angle — aligns with 2025–26 algorithmic trends toward identity content and "what this says about you" formats
🗺
Growth Blueprint
4-phase roadmap from 0 to 100K+ with measurable weekly targets and core KPIs
Section 1
Reality check: 100K in 90 days is achievable but requires 1–2 viral Reels (1M+ views each) AND a consistent daily execution floor. The targets below are aggressive but based on real accounts that have achieved this in the fragrance and lifestyle niche. Deviation from the execution plan by more than 3 days resets the compounding effect.
Phase 1
0 → 1K
Days 1–14
Phase 2
1K → 10K
Days 15–45
Phase 3
10K → 100K
Days 46–90
Phase 4
100K+
Day 91+
Phase 1: 0 → 1K Days 1–14 · Foundation & Proof of Concept

Primary objective: Build the content machine. Establish visual identity. Test 14 different hook formats to find what works for this specific audience. 1K followers is not the goal — finding 3 content formats that hit >10K views is the goal.

14
Reels to post
5K+
Views per Reel target
8%+
Engagement rate
1K
Follower target
MetricWeek 1 BenchmarkWeek 2 BenchmarkKill Threshold
Views per Reel2K–5K5K–15K<500 = kill format
Watch time %35%+45%+<20% = new hook
Saves rate2%+4%+<1% = no value delivery
Follow conversion1–2%2–4%<0.5% = wrong CTA
Engagement rate6%+10%+<3% = topic mismatch
Weekly followers+200+800<50/week = pivot
Content volume
  • 1 Reel per day — no exceptions. Quantity unlocks algorithm learning.
  • 0 static posts this phase — all energy on video.
  • Each Reel tests a different hook format (listed in Content System section)
  • Duration: 7–15 seconds — short-form dominates discovery
Engagement floor (daily)
  • 50 targeted comments on accounts in #perfumecommunity, #fragranceoftheday
  • 20 follows of engaged fragrance accounts (not brands — real people)
  • 10 DMs to people who commented on competitor posts within 2h
  • Reply to every comment on your own posts within 1h of posting
Phase 2: 1K → 10K Days 15–45 · Compound Growth Engine

Primary objective: Scale the 3 proven formats from Phase 1. Introduce collaboration strategy. Begin DM funnel. Hit first 100K-view Reel. The algorithm now knows your audience — feed it relentlessly.

42
Reels (31 days)
30K+
Views per best Reel
5%+
Saves rate target
10K
Follower target
MetricWeek 3–4 BenchmarkWeek 5–6 BenchmarkScale Threshold
Views per Reel (avg)10K–30K30K–100K1 Reel hits 100K+ → double frequency of that format
Watch time %40%+50%+55%+ = push for explore page
Saves rate4%+6%+Saves > Likes = high-value content signal
Shares1%+3%+3%+ shares = viral loop trigger
Follow conversion3–5%5–8%Hit 5% on a format → it's a pillar
Weekly followers+500+2,000<200/week in week 5 = content pivot
New in Phase 2
  • Collaboration outreach begins — 5 DMs/week to micro-influencers (5K–50K) in fragrance niche
  • Duet/stitch response strategy — respond to viral fragrance content with expert commentary
  • Series content — introduce recurring formats ("Fragrance of the Week", "This vs That")
  • First Story polls & Q&As — "What's your signature scent?" drives comment volume
Engagement scale-up
  • 100 comments/day on trending fragrance posts (not just niche — fashion, luxury, travel cross-posts)
  • DM funnel active: 20 DMs/day to recent followers of competitor accounts
  • Engage in Instagram Q&A Sessions — run 1 story Q&A per week
  • Start Broadcast Channel — daily fragrance tip (seed 200 subscribers)
Phase 3: 10K → 100K Days 46–90 · Viral Amplification

Primary objective: Engineer at least 2 viral Reels (>500K views). These 2 videos will drive 60–70% of total follower growth in this phase. Simultaneously: deepen community, activate collaboration pipeline, and make the account feel like the definitive fragrance authority.

2+
Viral Reels needed
500K+
Views per viral Reel
3–5%
Avg follow conversion
100K
Follower target
MetricWeek 7–9 TargetWeek 10–13 TargetDecision Rule
Views per Reel (avg)50K–200K200K–500K+Any Reel 200K+ → immediately create 3 variations
Profile visits per Reel5K+20K+High visits + low follows = bio/CTA problem, not content
Shares per Reel500+2K+Shares are the #1 viral signal — prioritise above likes
Collab reach2 collabs done5+ collabs totalEach collab should bring minimum +300 followers
Weekly followers+5,000+20,000Miss 2 consecutive weeks → execute "viral attempt" protocol
The 500K formula: Your highest-performing format from Phase 2 + a broadly relatable hook + a trending audio + a clear share trigger = the anatomy of your first viral Reel. Do not try to engineer virality from scratch — amplify what already works at 30K.
Phase 4: 100K+ Day 91+ · Authority & Monetisation

Primary objective: Convert audience into product users, brand partners, and community members. Begin paid collaboration conversations. Launch affiliate programme. Position ScentMatch as the #1 fragrance advisory platform on Instagram.

5K+
Platform signups/month
€5K+
Collab revenue/month
4%+
Profile-to-signup rate
1M
Next milestone
  • Monetisation: affiliate links (Notino, Sephora, niche retailers) embedded in Reel descriptions + stories
  • Paid brand partnerships with niche houses (Amouage, Xerjoff, Nishane) — audience quality > size for these brands
  • Launch ScentMatch Pro (premium profiling, personal consultations)
  • Introduce Community Challenge — "Share your ScentMatch profile" UGC campaign
  • Begin cross-platform expansion to TikTok using proven IG content templates
🎬
Content System
5 pillars · Viral Reels framework · 10 hooks · 5 storytelling formulas · performance rules
Section 2
A. Content Pillars (5)
Pillar 1: "What's in your bottle?" Viral
  • Purpose: Decode famous fragrances using data — note pyramids, performance scores, audience ratings
  • Audience benefit: "I finally understand why I love/hate this scent"
  • Formats: Screen-record ScentMatch app showing note breakdown, 7–12s fast cuts
  • Example: "Why Sauvage smells like everyone's ex (the data)" → mass relatable hook
  • KPI: 15%+ share rate (people tag their ex or friends)
Pillar 2: "Your scent personality" Engage
  • Purpose: Make olfactory profiling tangible and personal — "What your fragrance choice reveals about you"
  • Audience benefit: Identity affirmation + curiosity about their own profile
  • Formats: "If you wear [X] you are..." listicle Reels + quiz prompt Stories
  • Example: "Wearing Oud? Here's your psychological profile" → high save + share
  • KPI: 8%+ saves, 500+ comments per Reel
Pillar 3: Niche Education Trust
  • Purpose: Establish ScentMatch as the expert source — nobody else teaches this on IG
  • Audience benefit: Real knowledge that makes them look smart to friends
  • Formats: "3 things the fragrance industry doesn't want you to know", note chemistry explainers
  • Example: "The ISO E Super molecule in 40% of designer fragrances — explained in 10s"
  • KPI: 6%+ saves, profile visits spike after post
Pillar 4: "Fragrance for [occasion]" Convert
  • Purpose: Drive platform usage — "find the right scent" is the product promise
  • Audience benefit: Practical, actionable — they can buy something today
  • Formats: "Best fragrances for a first date", "What to wear in summer heat", occasion-specific picks
  • Example: "I ran 850 fragrances through an algorithm to find the best office scent" → data authority + practical
  • KPI: 4%+ link-in-bio clicks, 10%+ saves
Pillar 5: Founder POV / Behind the Build Trust
  • Purpose: Humanise the brand — show the obsessed collector behind the algorithm
  • Audience benefit: Trust + parasocial connection — people follow people, not platforms
  • Formats: "My collection of 47 fragrances ranked", "How I built the algorithm that decoded my nose", day-in-the-life of a fragrance obsessive
  • Example: "I spent 3 years building an AI that finally told me what I actually like" → story arc + product demonstration
  • KPI: 12%+ engagement rate, highest comment volume of all pillars, highest follow conversion
  • Frequency: 1× per week maximum — scarcity keeps it premium
B. Viral Reels Framework

Every Reel must follow this 5-part structure. Deviation from the hook formula is the #1 reason content fails.

1
HOOK — First 1.5 seconds
Must stop the scroll before the audio registers. Visual + text overlay must create a question or tension that demands resolution. Never start with your logo, a greeting, or a slow pan.
↓
2
PATTERN INTERRUPT — Seconds 1.5–4
A visual cut, unexpected data point, or counterintuitive claim that resets attention. Example: cut to a data visualisation from ScentMatch, or a "wait, what?" statistic about the fragrance industry.
↓
3
VALUE DELIVERY — Seconds 4–12
The actual insight, ranking, or reveal. Delivered fast with text overlays reinforcing every spoken word. Must be genuinely useful or genuinely surprising — no filler. Each second must earn the next.
↓
4
RETENTION LOOP — Throughout
Tease the ending at the beginning ("the #1 is going to surprise you"). Use list formats ("5 things... #3 is the most underrated") — viewers stay to see if they agree. Never reveal the best point first.
↓
5
CTA — Final 1–2 seconds
One action only. Never ask for follow + like + comment simultaneously. Choose: Save this (for ranked/list content) · Comment yours (for opinion/personality content) · Find yours [link in bio] (for platform conversion). Match CTA to pillar type.
C. 10 Proven Hook Templates for ScentMatch
"Why everyone's wearing the same 3 fragrances" — Mass social observation → shame + curiosity loop. Audience already has an opinion before you speak.
"I analysed 850 fragrances to find the best [occasion] scent" — Data authority hook. The number does the work. "850" is specific enough to be credible.
"The fragrance industry doesn't want you to know this" — Conspiracy/insider hook. Works for note chemistry, markup secrets, reformulation history.
"Your fragrance choice reveals your personality type" — Identity/psychology hook. Highest share rate because people immediately tag friends.
"POV: Your nose has 400 types of receptors and you're still buying Sauvage" — Humour + superiority gap. Fragrance community inside joke with mass recognition.
"Rating fragrances by how many DMs they'd get" — Social proof framing. Highly shareable — people want to validate/debate the rankings.
"The €35 fragrance that outperforms a €350 niche" — Price disruption hook. Universally relevant — everyone wants to win on value.
"I wore [X] for 30 days and here's what happened" — Time-commitment authority. Signals obsession + expertise. High saves because it reads as research.
"Nobody talks about this note but it's in 80% of fragrances you own" — Revelation format. "80%" is the specific-number hook. Forces self-check → saves.
"If your perfume smells like this after 30 minutes, you made a mistake" — Fear of loss + expertise signal. Drydown chemistry is rarely explained on IG — instant authority.
D. 5 Storytelling Formulas
Formula 1: The Contrast
"Everyone thinks X → but the data says Y."
Example: "Everyone thinks longevity = quality. I tested 200 fragrances. The highest-rated ones had shorter sillage." → data-backed contrarianism. Works in every pillar.
Formula 2: The Rank Reveal
"I ranked [5 things] and #1 surprised everyone."
Tease position 5→1. Never start at 1. The reveal structure keeps watch time above 70%. Best for Pillar 1 (fragrance breakdowns) and Pillar 4 (occasion picks).
Formula 3: The Identity Mirror
"If you wear [X], here's what it says about you."
Uses olfactory psychology data from the profiling system. High saves + shares because people screenshot to send to friends. Drives quiz traffic.
Formula 4: The Journey
"I spent [time/effort] to discover [insight]."
Founder story format. "I built an algorithm over 3 years" validates expertise without bragging. Specific time/effort = perceived value. Reserve for Pillar 5.
Formula 5: The Before/After
"Before I knew about [concept] I did [wrong thing]. After: [transformation]."
Example: "Before I understood note pyramids, I was buying blind. After: my hit rate went from 30% to 90%." Highly relatable — everyone has bought a fragrance they regretted. Drives platform trial (the solution is ScentMatch).
E. 5 Share / Save Triggers
Trigger 1: "Send this to someone"
Explicit in-video prompt: "tag the person whose ex this reminds you of." Fragrance memory is deeply social — leverage it. Creates DM chains which signal virality to the algorithm.
Trigger 2: The Ranking People Disagree With
Take a mildly controversial position (e.g., "Aventus is overrated"). Comment volume 10× normal. Algorithm reads comments as engagement = more distribution. Disagree with a brand, never with a person.
Trigger 3: Reference Content
"Save this for your next fragrance purchase." List-format educational content people want to reference later. Saves rate directly influences how often Instagram recommends your content to non-followers.
Trigger 4: "Take the test"
End the Reel with "Find your actual scent profile — link in bio." The quiz CTA converts watch-time into platform traffic. This is the monetisation funnel entry point.
Trigger 5: Identity Labelling
"If you're a [Woody/Oriental/Gourmand] person, you'll understand this." Calling out a specific identity group creates belonging. People share content that represents who they are. Use family names from the ScentMatch taxonomy — it subtly educates on the platform vocabulary.
F. Content Performance Rules
MetricMinimum ViableWinning ThresholdAction
Views (first 48h)1,000+10,000+Under 500 after 48h → analyse hook, not content quality
Watch time %30%50%+Under 25% → hook failed, kill format
Saves rate2%5%+Saves > Likes = algorithm push signal. Target this metric first.
Shares1%3%+Shares are weighted 4× likes. If low: add explicit "send this to" prompt
Comments10+50+Controversial or identity hook needed if comments < 5
Follow conversion1%4%+High views + low follows = content is entertainment, not authority. Fix: strengthen profile bio, add more Pillar 5 content
Profile visits500+3,000+High visits + low follows = bio/CTA problem. Rewrite bio to lead with result, not description.
Kill rule: Any content format that produces <500 views AND <1% saves in its first 5 posts is dead. Do not optimise it. Kill it and move to the next hook template. Speed of experimentation beats perfection of execution.
📅
90-Day Daily Execution Program
Strict daily playbook — every day has a task, time estimate, output, and KPI target
Section 3
How to use this plan: Each week is collapsed by default — click to expand. Day types: CREATE = filming/editing day · POST = publishing day · ENGAGE = community day · REVIEW = analytics day. Most days combine types.
🎬 Reel Calendar
Legend: Bottle Decode Scent Personality Education Occasion Founder
Programme progress calculating…
26 May 2026 23 Aug 2026
90
Total Reels
0
Published
0
Upcoming
0
Overdue
Phase 1
Foundation · Days 1–14
Phase 2
Momentum · Days 15–42
Phase 3
Authority & Viral · Days 43–90

Checkboxes are saved in your browser (localStorage). Click a Reel title to open the script file.

Week 1

Days 1–7 · Infrastructure & First Content (Phase 1)

Target: +200 followers · 7 Reels posted ▼
DAY 1Create + Setup
  • Optimise bio: "Fragrance advisor. 23K fragrances decoded. Find yours ↓" — link to ScentMatch quiz
  • Profile photo: Clean, high-contrast, no text — face or bottle close-up
  • Film Reel #1: Hook 2 — "I analysed 850 fragrances to find the best daily scent" (use ScentMatch screen-record + voiceover)
  • Edit to 10–12s: Fast cuts, text overlays on every line, trending audio
  • Research: Find 10 accounts in #perfumecommunity with 5K–50K followers — note their top posts
3h total Output: 1 Reel ready to post KPI: Bio complete, link active
DAY 2Post + Engage
  • Post Reel #1 at 7–9pm local time. Caption: 1 line + 3 questions to drive comments. 5 targeted hashtags only.
  • First 60 minutes after posting: Reply to every comment within 5 min. React to 30 posts in #fragranceoftheday
  • Film Reel #2: Hook 4 — "Your fragrance choice reveals your personality" (use olfactory profile axes from ScentMatch)
  • DMs: 10 people who posted in #fragrancecommunity today — genuine compliment on their collection, no pitch
2.5h total Output: Reel #1 live, Reel #2 ready KPI: Watch time >30%, Engagement >5%
DAY 3Post + Engage
  • Post Reel #2. Add Story immediately after: "Which personality type are you? Comment below ↓"
  • Engage 60min: 50 comments on posts tagged #perfumecommunity (real opinions, not "great post!")
  • Film Reel #3: Hook 7 — "The €35 fragrance that outperforms a €350 niche" — pick actual data from DB
  • Check metrics: Reel #1 at 24h — note views, watch time%, save rate. Record in tracker.
2.5h total Output: Reel #2 live, Reel #3 ready KPI: Reel #1 should have 500+ views by now
DAY 4Post + Engage
  • Post Reel #3. CTA: "Save this before your next fragrance purchase"
  • Story: Poll — "Do you prefer long-lasting or soft skin-scents?" — builds engagement history
  • Follow 20 accounts that engage on competitor posts (people, not brands)
  • Film Reel #4: Hook 3 — "The fragrance industry doesn't want you to know this" — ISO E Super, reformulations, price markup
2h total Output: Reel #3 live, Story active KPI: 3 Reels live, total 1,500+ views
DAY 5Post + Engage
  • Post Reel #4 — insider knowledge format. CTA: "Comment 'NOTES' and I'll tell you what's in your favourite fragrance"
  • DM 15 people who commented on a competitor's top post this week — lead with value ("saw your comment on X, have you tried Y?"). No pitch.
  • Film Reel #5: Storytelling Formula 1 (The Contrast) — "Everyone thinks [popular fragrance] is masculine/feminine. The chemistry says otherwise."
  • Reply to all DMs received. First DMs coming in = high priority.
2.5h total Output: Reel #4 live, 15 DMs sent KPI: DM reply rate >30%
DAY 6Post + Engage
  • Post Reel #5. Add to Story with a sticker "What do you think?" — drives story views which boost Reel distribution
  • Engage 90min: Comment on 30 posts, reply to every comment on your 5 posts so far
  • Film Reel #6: Hook 9 — "Nobody talks about this note but it's in 80% of fragrances you own" — Iso E Super / Ambroxan / Hedione
  • Research: Identify 5 micro-influencers (5K–30K) for potential collab outreach in Week 3
3h total Output: Reel #5 live, Reel #6 ready KPI: 5 Reels live, follower count +100
DAY 7Post + Weekly Review
  • Post Reel #6. CTA: "Follow for more data-backed fragrance picks"
  • Weekly review (30 min): Which of the 6 Reels got the most saves? The most comments? The best watch time? That format gets repeated next week — twice.
  • Decision: Any format under 500 views after 48h → retired. Move to next hook template.
  • Plan Week 2: Write the hooks for Reels 7–13 based on Week 1 data
2h total Output: Reel #6 live, Week 2 plan written KPI: Total followers +200. Best Reel: 3K+ views.
Weekly Review System
Review ItemWhat to measureDecision if off-track
Top Reel of the weekViews, saves rate, follow conversion, watch time %Double its format next week
Bottom Reel of the weekSame metricsRetire that hook template if it's failed 3× in a row
Follower trajectoryWeek-over-week growth rate %If growth <15% vs prior week for 2 weeks → engagement volume ×1.5
Engagement rate avgTotal engagements ÷ total reach across all postsUnder 5% → more controversial/identity content needed
Platform traffic from IGReferral sessions from Instagram in analyticsUnder 50/week → update link in bio CTA, test different landing page
DM conversation rateDMs sent → replies received %Under 20% reply rate → change DM opener (see Growth Systems)
Pivot Rules
IF
Week 2 and no Reel has hit 1K views → hook problem. Replace the opening 2 seconds of all future Reels. The content is fine; the entry point isn't working.
IF
High views but followers not growing → bio/profile problem. People watch but don't convert. Rewrite bio to lead with "what you get" not "what we are."
IF
Good followers but low platform signups → CTA problem. Add a dedicated Reel showing the ScentMatch quiz in action. Make the output (personalised results) visually compelling.
IF
Engagement drops week 3+ → content fatigue. Introduce a new pillar or format. Audit if you're posting too much of the same hook type.
⚙️
Growth Systems
Repeatable engines: viral loop · comment funnel · DM scripts · collaboration system
Section 4
A. Viral Loop System
1
Hook triggers watch → Watch triggers share
Mechanism: Every Reel must have 1 explicit "share this to someone" moment — a joke, a revelation, a personality label, or a ranking they disagree with. Shares are weighted 4× likes by the algorithm. Volume target: 3%+ shares per Reel to trigger distribution loop.
↓
2
New viewer → Profile visit → Follow
Mechanism: Viral Reel sends 10K+ profile visits. Bio must immediately answer "why should I follow?" Bio template: "[Result you deliver] · [Proof point] · [Action]". Example: "Your perfect fragrance exists. 23K options decoded. Take the quiz ↓" — 3 lines, no emojis, no fluff. Volume target: 4%+ profile visit → follow conversion.
↓
3
Follower → Platform user → Brand advocate
Mechanism: Welcome DM → quiz link → completed profile → they post their result → UGC. Every person who completes the ScentMatch quiz becomes a potential content creator if prompted. "Share your profile" CTA in the quiz result screen creates organic UGC. Volume target: 10% of new followers take the quiz. 10% of quiz takers post their result.
↓
4
UGC → New viral surface
Mechanism: When followers post "I just took the ScentMatch quiz and my profile is…", repost it. Every repost is free reach + social proof + a signal to new viewers that the product is worth trying. Volume target: 5 UGC reposts per week by Month 2. Offer featured users a shoutout as incentive.
B. Comment Funnel System
The comment funnel converts views into conversations, conversations into followers, and followers into platform users. This is the highest-ROI manual activity in the entire growth system.
Outbound Comments (50–100/day)
  • Target accounts: Posts in #perfumecommunity, #fragranceoftheday, #nichebeauty tagged in last 2 hours — not older. Recency matters for visibility.
  • Comment structure: Add knowledge, not compliments. "This is exactly the Ambroxan signature you're smelling — it's in 60% of designer fragrances" is worth 10× "amazing collection!"
  • Strategic targets: Comment on posts by accounts with 50K–500K followers — your comment appears alongside theirs and gets exposure to their audience
  • Reply to comments on competitors' posts: When someone asks "what should I wear for summer?" on a large fragrance account — answer it better than anyone. That person often visits your profile.
  • KPI: 5–10% of people you comment on visit your profile. 10–20% of those follow. At 100 comments/day: 5–20 new followers/day from comments alone.
Inbound Comment Management
  • Reply within 60 min of posting — algorithm reads early reply velocity as a quality signal
  • Reply to every comment for the first 4 weeks — after that, prioritise questions and detailed comments
  • Drive comment threads: Reply with a follow-up question — "what family does your signature scent fall into?" Thread depth increases distribution
  • "Pin" the best comment — pick the most interesting comment and pin it. This shows curation and invites others to top it.
  • Convert commenters to DMs: When someone writes a detailed fragrance opinion, reply: "This is a great point — I DM'd you something relevant." Then do it. These are your highest-value followers.
  • KPI: Average comment thread depth >2 replies. Comments per Reel: 20+ by Week 4.
C. DM Funnel — Scripts & Goals
DM Script #1: Cold Opener (Phase 1–2)
Template: "Hey [name] — saw your post about [specific fragrance]. I'm working on a fragrance profiling tool and [that fragrance] has a really interesting [note/characteristic]. Are you more drawn to the [X] or the [Y] side of it?"
Goal: Start a genuine fragrance conversation — not pitch. Volume: 20/day. KPI: 25%+ reply rate. Why it works: Shows you actually looked at their post. Asks a question they have an opinion on. Zero ask from them.
DM Script #2: Platform Introduction (Phase 2, after 2 exchanges)
Template: "By the way — I built something that might actually map your fragrance preferences into a proper profile. It goes way deeper than the usual 'try Chanel No.5 if you like florals' stuff. Your current collection suggests [X] — curious if the algorithm lands the same way?"
Goal: Introduce platform without a hard pitch. Frame as curiosity. Volume: Send to every DM contact after 2+ exchanges. KPI: 40%+ click the link. Why it works: Personal reference to their specific collection. Positioned as expert prediction, not product promotion.
DM Script #3: Collab Outreach (Phase 2–3)
Template: "Hey [name] — love your [specific post/series]. I have a fragrance data tool that mapped [their aesthetic/niche] — I found 3 fragrances that match your profile exactly that I'd bet you haven't tried. Would you be interested in doing a 'vs.' Reel? I can do the data side, you do the review side."
Goal: Propose a collab that gives them value (content + novelty + data angle) and you get audience exposure. Volume: 5/week to accounts 5K–50K. KPI: 20% acceptance rate = 1 collab/week by Month 2. Why it works: Specific offer. Clear format. They don't need to do anything complex.
DM Script #4: Welcome DM (all new followers after viral Reel)
Template: "Welcome! Since you followed — your scent profile is probably waiting to be decoded. Most people who find this account turn out to be [Woody/Oriental/Chypre] types. Takes 2 min: [link]"
Goal: Convert new followers to platform users within 24h of follow. Volume: Every new follower during viral period. KPI: 15%+ quiz completion from DM traffic. Why it works: Immediate, personalised, makes them curious about their "type."
D. Collaboration System
PhaseTarget Account SizeCollab FormatExpected ROIVolume
Phase 10–5K (peer accounts)Comment exchange, mutual engagement — no formal collab yet+50–100 followers per peer5 peer accounts/week
Phase 25K–30K (micro)"This vs. That" fragrance Reel — shared topic, each posts own version and cross-tags+200–500 followers per collab1 collab/week
Phase 330K–200K (mid)Guest expert Reel, duet/stitch, or joint quiz — full shared Reel with both accounts tagged+1K–5K followers per collab1–2 collabs/week
Phase 4200K+ (large)Sponsored collab — provide free advisory + data content in exchange for post+5K–20K followers, potential paid deal2 per month
Who to target specifically: Fragrance review accounts, beauty editors, "niche perfume collector" personas, men's lifestyle creators who cover grooming. Avoid: mass beauty influencers — their audience doesn't convert for a precision advisory tool. Avoid collabing with anyone whose audience is "gifted/PR" — their engagement is performative, not genuine.
🛠
Tools & Workflow
Specific tools, exact daily usage, and the efficiency gain each delivers
Section 5
✂️
CapCut (Mobile + Desktop)
Reels Editing
Daily use: Film raw footage on phone → edit in CapCut → export 9:16, 1080p. Use Auto Captions (accuracy 90%+ for English) to add text overlays in 10 seconds. Use "Beat Sync" to cut to audio automatically. Templates allow consistent visual style without graphic design skills. Efficiency gain: Reduces edit time from 45 min to 12 min per Reel once templates are set. Critical for 1-Reel/day cadence.
📱
iPhone native camera + Ring Light
Content Capture
Daily use: Cinematic mode for face-to-camera content. Screen record (built-in) for ScentMatch app demos — this is your most unique content surface. Ring light + neutral background = consistent brand aesthetic at zero cost. Key rule: Film in 4K, export from CapCut at 1080p — keeps file size manageable without quality loss. Efficiency gain: No need for external camera or videographer. Batch film 3 Reels in one 45-min session.
📆
Later or Buffer
Scheduling & Planning
Daily use: Schedule 3–4 Reels in advance on Sunday. Drag-and-drop visual calendar shows content gaps. Best posting times (auto-calculated from your audience data) are applied automatically. Key use case: When you batch-film on Sunday, you don't have to think about content for the next 3 days — it's scheduled and posted without manual action. Efficiency gain: Eliminates daily "what do I post?" decision fatigue. Saves 30 min/day.
📊
Instagram Insights (Native)
Analytics — Daily + Weekly
Daily use (5 min): Check yesterday's Reel: views, watch time%, saves, shares. Note the number — that's your benchmark to beat. Weekly use (20 min): Export last 7 days of Reels data. Rank by saves rate. Identify top format. Make one decision. What to ignore: Likes count, follower count hour-to-hour — these are vanity metrics. Watch time% and saves rate are the only leading indicators of algorithmic distribution. Efficiency gain: Free, native, actionable — no paid tool needed until 50K+ followers.
🎵
Instagram Reels Audio Library + TokBoard
Trend Tracking — Audio
Daily use (5 min): Check Instagram's "Trending" audio section before filming — filter by "Reels" and look for sounds with 50K–500K uses (not 5M+ — those are already saturated). TokBoard (free web tool) shows which TikTok sounds are crossing to Instagram 48h before they peak. Being early on a trending sound gives 3–5× normal distribution. Efficiency gain: One early trending audio can add 50K views to an otherwise average Reel. This is the single highest-leverage technical insight in the entire system.
🔍
Notion or Obsidian (Content Tracker)
Content Ideation & Performance Tracking
Daily use: One table: Reel # · Date · Hook type · Pillar · Views · Watch time% · Saves · Shares · Follows gained · Decision (scale/retire/test again). Weekly use: Sort by "Saves rate" descending — this is your content strategy in one view. Idea bank: Add hook ideas whenever they occur. Aim for 30+ ideas in backlog at all times — never film from a blank slate. Efficiency gain: Eliminates guesswork. Data-driven decisions replace intuition by Week 4.
🤖
Claude / ChatGPT
Hook Writing & Caption Drafts
Daily use: Use to generate 5 hook variations for any planned Reel topic in 2 min. Prompt: "Write 5 Instagram Reel hooks for this topic: [topic]. The audience is fragrance enthusiasts who value data and expert knowledge over hype. Each hook must be under 8 words and create immediate curiosity." Pick the best one, adapt to your voice. Caption drafting: "Write a 2-sentence Instagram caption for a Reel about [topic]. Include a question to drive comments. No hashtag suggestions." Efficiency gain: Replaces 20 min of staring at a blank screen. You go from idea to final hook in under 5 min.
📊
Analytics & Decision System
Daily 5-min check · Weekly 20-min review · IF/THEN decision rules
Section 6
Daily Metrics (5 min — check every morning)
Views
Yesterday's Reel at 24h. Target: 2× your current average
Watch %
Under 30%? Hook failed. Over 50%? Replicate immediately.
Saves
Saves ÷ views = save rate. This is your #1 distribution signal.
Shares
Every share = 4 algorithm points. Track separately from other engagements.
Weekly Metrics (20 min — every Sunday)
MetricHow to calculateTargetBenchmark source
Average views/ReelTotal views ÷ Reels posted this weekGrowing 20%+ week-over-weekInstagram Insights → Reels tab
Average save rateTotal saves ÷ total views × 100>3% = good · >6% = excellentInstagram Insights → individual Reel
Average share rateTotal shares ÷ total views × 100>1% = good · >3% = viral signalInstagram Insights → individual Reel
Follower growth rate(New followers ÷ start of week followers) × 100Phase 1: 20%+/wk · Phase 3: 30%+/wkInstagram Insights → Overview
Follow conversion rateNew followers ÷ total profile visits × 100>3% = bio working · <1% = rewrite bioInstagram Insights → Profile Activity
Engagement rate(Likes + Comments + Saves + Shares) ÷ Reach × 1006–10% = strong for niche accountCalculate manually from Insights data
Platform trafficSessions from Instagram in web analytics50+ per week by Week 4Google Analytics / Plausible → Referrers
Winning Content Identification Rule
The 3-Signal Test — a Reel is "winning" if it passes ALL 3
Signal 1
Watch time >40% — people are watching past the hook. The content earns attention.
Signal 2
Saves rate >3% — people are saving it for later. The content has utility value.
Signal 3
Shares rate >1% — people are sending it to others. The content has social currency.
A Reel that passes all 3 signals gets immediately replicated 3× — same structure, different topic/fragrance. A Reel that passes only 1–2 signals gets one more attempt with a revised hook.
IF / THEN Decision Framework
IF
Watch time >50% AND shares >2% → replicate immediately. Create 3 variations of this exact structure within 72 hours. Do not wait for the next weekly review.
THEN
When replicating: keep the hook formula identical, change only the fragrance example or specific claim. Confirm the format — not the topic — is the winning variable.
IF
Views spike 3–5× on a single Reel AND watch time >40% → this is a viral window. Post a follow-up Reel on the same topic within 24 hours while distribution is hot.
IF
3 consecutive Reels under 1K views → hook audit. Watch your own Reels with the sound off. If the first 1.5 seconds don't make you want to keep watching, redo it.
IF
Saves > 5% but views are low (<2K) → distribution problem, not content problem. The content is valuable but not reaching enough people. Fix: add trending audio, post at a different time, or cross-post to Stories with a "swipe to save" prompt.
IF
High views (>20K) but follow rate <1% → profile problem. They watch but don't care enough to follow. Rewrite bio. Change profile photo. Add a pinned Reel that functions as a "why follow me" explanation.
IF
Engagement rate drops below 4% for 2 consecutive weeks → audience mismatch growing. Your content is attracting passive viewers, not engaged followers. Shift to more comment-bait formats (rankings, controversial opinions, personality labels).
IF
Platform traffic from IG exceeds 200 sessions/week → amplify the conversion CTA. More people are ready to try the product than you're currently converting. Add a dedicated Reel showing the quiz output. Update the link-in-bio landing page.
⚠️
Failure Prevention
10 main failure risks — early warning signs + prevention action for each
Section 7
🎭
Risk 1: Inconsistent Posting — The #1 Killer
Warning signs: Skipping 2+ consecutive days in Week 1–3. "I'll post 2 tomorrow to make up for it." Any gap over 3 days in the first 60 days.
Prevention: Batch film 3–4 Reels every Sunday. Keep a buffer of 3 ready-to-post Reels at all times. The algorithm's audience model resets after 5–7 days of inactivity — you lose the distribution advantage built up during active periods. There is no shortcut around consistency.
🪝
Risk 2: Weak Hooks — Content That Never Gets Watched
Warning signs: Watch time consistently under 25%. Views plateau at 500–1,000 despite daily posting for 2+ weeks.
Prevention: Test your own hook by watching the first 2 seconds with sound off. If you wouldn't stop scrolling, your audience won't either. Use the 10 hook templates from Section 2. A good hook is a promise or a question. Never start with your name, a greeting, or a slow visual. If your first Reel didn't work, the hook was the problem — not the topic.
🚫
Risk 3: Generic Content — "Just Another Fragrance Account"
Warning signs: You're posting content that 50 other fragrance accounts already post. Flat engagement despite views. Followers don't save or share — they just scroll past.
Prevention: Every Reel must have at least one element nobody else can produce: ScentMatch data (note breakdowns, performance scores, cross-database comparisons), your olfactory profiling system, or your specific collector expertise. If a faceless stock-footage account could make your Reel, delete it and start over.
🔇
Risk 4: No Engagement Activity — "Post and Ghost"
Warning signs: Posting daily but not spending time in comments, not sending DMs, not engaging with other accounts. Growth stalls at 200–400 followers despite consistent content.
Prevention: The content creates reach; the engagement creates community and signals intent to the algorithm. 60–90 min/day of outbound engagement is non-negotiable in Phase 1–2. This drops to 30–45 min after 10K, when inbound engagement becomes substantial enough to replace it.
🎯
Risk 5: Wrong Audience — Watchers, Not Followers
Warning signs: High Reel views but very low follow conversion (<0.5%). Content getting shared outside the fragrance community (meme accounts, general beauty).
Prevention: Viral reach with the wrong audience is worthless. Check your Insights audience demographics (Instagram Insights → Audience → Top Locations, Age, Gender). If your audience doesn't match your target (fragrance enthusiast, 25–45, engaged), the hook is too broad. Narrow it slightly — use fragrance-specific vocabulary in hooks to filter for quality audience.
🔁
Risk 6: Content Repetition Fatigue
Warning signs: Engagement rate drops week-over-week despite consistent posting. Comments become fewer and less engaged. Existing followers stop watching new content.
Prevention: Never post the exact same format more than 2× in a row. Rotate through all 5 pillars within each 7-day period. If a format is "winning," vary the topic not the structure — personality profiles for different fragrance families, data reveals for different occasions. Introduce a new content series every 3–4 weeks to refresh the feed.
📉
Risk 7: Platform Algorithm Change Mid-Campaign
Warning signs: Sudden unexplained drop in reach (30%+ overnight). Previously high-performing format suddenly underperforming. Industry-wide reports of algorithm update.
Prevention: Diversify across formats — don't become a single-format account. Always maintain an email list / broadcast channel / platform presence so algorithm changes don't wipe your audience access. When a format stops working: don't chase it. Test 3 new hook types within 72 hours and let the data tell you what the algorithm is currently rewarding.
⚡
Risk 8: Burnout — Solo Operator Ceiling
Warning signs: Posting quality declining. Skipping engagement sessions. "I'll do it tomorrow" becoming a pattern. Energy going into perfecting individual posts instead of volume.
Prevention: Batch creation on Sundays covers Mon–Wed. Batch creation on Wednesday covers Thu–Sat. This means you film twice per week, not daily. Engagement can be done in 2× 30-min focused sessions rather than scattered throughout the day. Done is better than perfect. A 70% quality Reel posted is infinitely more valuable than a 100% quality Reel that never gets made.
🏪
Risk 9: Product Not Converting — Traffic Without Signups
Warning signs: Good Instagram metrics but flat or low platform signups. High link-in-bio clicks but low quiz completion rate.
Prevention: The link-in-bio landing page must promise the same outcome the Reel teases. If the Reel says "find out your fragrance personality" — the landing page must immediately show the quiz, not a homepage with multiple options. Test: show the quiz result screen in a Reel — make the output look desirable. The closer the content is to the actual product experience, the higher the conversion rate.
🧊
Risk 10: Cold Start Failure — Giving Up Before the Compounding Starts
Warning signs: Week 1–2 numbers are low and discouraging. Comparing daily follower gains to established accounts. Posting slows because "it's not working."
Prevention: Instagram's algorithm takes 4–6 weeks to understand what audience to serve your content to. The first 2 weeks are almost always slow. The accounts that hit 100K in 90 days are not the ones who went viral on Day 1 — they're the ones who kept posting on Days 10, 20, and 30 when nothing seemed to be happening. The compounding does not start until you've proven consistency. Consistency is the product, not the content.
The Only 3 Metrics That Actually Matter
Saves %
The single strongest predictor of algorithmic distribution. Every 1% improvement in save rate doubles your potential reach.
Watch %
Tells you if the hook and content are working. Under 30%: people left. Over 50%: Instagram will push this harder.
Shares
Each share exposes your content to an entirely new network. This is the only organic "paid amplification" you have.
Everything else — likes, comments, follower count — are outputs of these three inputs. If you optimise only saves rate, watch time, and shares across every Reel, the followers will follow. The reverse is not true.
ScentMatch · LinkedIn Growth OS · 2026

The Thought Leadership Playbook

A data-driven LinkedIn growth system calibrated to ScentMatch's existing editorial voice, 10-post content library, and founder expertise. Every action tied to a measurable outcome. Built for a solo operator.

🔍
Platform Audit
LinkedIn vs Instagram — different algorithm, different audience, different success metrics
Step 0
Why LinkedIn is a Different Game
LinkedIn Algorithm (2026)
  • Dwell time is the #1 signal — how long someone reads your post, not just if they click
  • Comments outweigh likes 10× — a post with 5 comments outperforms one with 50 likes
  • Connection-first distribution — post first goes to your 1st-degree connections; their engagement unlocks 2nd-degree reach
  • Posts with external links are suppressed — LinkedIn penalises posts that take users off-platform. Never put a link in the post body — put it in the first comment.
  • Posting window: Tue–Thu 7–9am or 12–1pm. Weekends are dead for B2B/professional content.
ScentMatch's LinkedIn Advantage
  • 10 editorial posts already drafted — expert voice, research-cited, differentiated. This is a major head start.
  • Fragrance × AI × Data angle — no other fragrance account on LinkedIn owns this position. It's completely open.
  • Target audience exists on LinkedIn: luxury brand managers, fragrance industry professionals, beauty editors, tech-meets-luxury investors, sophisticated consumers
  • Founder credibility — built the platform, genuine collector, cites Chemical Senses / Nature / ACM RecSys. LinkedIn audiences reward verifiable expertise.
Current State — Gaps
GapNo consistent posting cadence — 3 posts over 3 weeks with irregular gaps. LinkedIn requires 3–5 posts/week to build algorithmic momentum.
GapLong-form only — all 10 drafted posts are essay-length. LinkedIn rewards variety: carousels, short punchy posts, polls, and listicles alongside long-form.
GapNo engagement strategy — posting without commenting on others' posts means the algorithm treats the account as broadcast-only. Low distribution.
GapCompany page vs. personal page — company pages get ~60% less organic reach than personal profiles on LinkedIn. Founder personal page should be the primary growth vehicle.
Current State — Strengths
StrengthExceptional written voice — the drafted posts are among the best fragrance content on LinkedIn. Research-backed, confident, not sycophantic. Rare quality.
StrengthTopic breadth already mapped — 10 topics cover the full content surface: AI, olfaction science, gender fluidity, niche revolution, memory, wellness, brand critique.
StrengthCross-platform authority amplifier — Instagram scale + LinkedIn credibility creates a compound trust signal. B2B partners check LinkedIn; consumers check Instagram.
OpportunityLinkedIn fragrance niche is empty — search "fragrance" on LinkedIn and you find brand PR, not real expertise. First-mover advantage is real and available now.
🗺
Growth Blueprint
4 phases from 0 to 10K followers with weekly targets and KPIs
Section 1
LinkedIn reality check: 10K followers in 90 days is aggressive but achievable with 3–5 posts/week + active engagement. The key lever is one post going viral (10K+ impressions) — this happens when a post triggers high comment velocity in the first 2 hours. Unlike Instagram, LinkedIn virality is driven by debate and professional disagreement, not entertainment.
Phase 1
0 → 500
Days 1–14
Phase 2
500 → 2K
Days 15–45
Phase 3
2K → 10K
Days 46–90
Phase 4
10K+
Day 91+
Phase 1: 0 → 500 · Days 1–14 · Foundation

Objective: Post the 10 existing drafted posts (2 per week) + 3 short-form posts per week. Build a 200-connection base of relevant professionals. Identify which of the 10 topics generates the most comments.

14
Posts in 2 weeks
500+
Impressions per post
3%+
Engagement rate
500
Follower target
MetricWeek 1Week 2Kill signal
Impressions/post300–800500–2K<100 = wrong posting time or topic
Comments/post3+8+<1 = hook/opening line failed
Connection requests sent50/week50/week—
Connection accept rate30%+40%+<20% = wrong targeting
Weekly follower growth+100+300<50/week = increase outreach
Phase 2: 500 → 2K · Days 15–45 · Momentum

Objective: Introduce carousel posts (highest reach format on LinkedIn). Hit first viral post (5K+ impressions). Start commenting on large industry accounts to get name recognition. Begin building the "fragrance × tech" narrative that no one else owns.

3–5
Posts/week
5K+
Target: 1 viral post
20+
Comments target/week
2K
Follower target
  • Carousel posts (2/week): "5 things the fragrance industry doesn't tell you" — slide format stays in feed longer, gets more swipe interactions, boosted by algorithm
  • Poll post (1/week): "What drives your fragrance choice? A) Occasion B) Mood C) Season D) Brand" — polls get 3× normal reach, comments driven by results
  • Comment 20 posts/day on accounts followed by your target audience (beauty editors, luxury brand executives, fragrance houses, tech/AI journalists)
  • Connection targeting: Fragrance industry professionals, luxury brand managers, beauty tech investors — 50 requests/week with personalised note
Phase 3: 2K → 10K · Days 46–90 · Authority

Objective: Engineer 2–3 viral posts (10K+ impressions each). These drive 70% of follower growth. Key mechanic: take a mildly controversial professional position on the fragrance industry and defend it with data. This drives comments from both sides — algorithm gold.

  • Controversy-driven posts: "The niche fragrance industry has a transparency problem" / "AI will replace the fragrance consultant — here's why I built one anyway" / "Why gender-neutral perfume marketing is still not working"
  • Data posts: Use ScentMatch database — "I analysed 23,000 fragrances. Here's what the data says about [topic]." No other LinkedIn account can write this sentence credibly.
  • Newsletter launch: LinkedIn Newsletter (built-in feature) — "The Olfactory Intelligence Brief" — weekly. Subscribers get notified automatically, boosting weekly reach.
  • Collaboration posts: Co-write a post with a fragrance brand founder or industry figure — each tags the other, combining audiences
2–3
Viral posts needed (>10K impressions)
500+
Newsletter subscribers by Day 90
5K+
Weekly impressions average
10K
Follower target
Phase 4: 10K+ · Day 91+ · Monetisation
  • B2B advisory pipeline: Inbound enquiries from fragrance brands wanting to understand their audience data — ScentMatch becomes a B2B data product
  • Speaking / media: LinkedIn authority → podcast invitations, trade press features, conference panels (Esxence, Fragrance Foundation)
  • LinkedIn Newsletter monetisation: Sponsored issues from niche fragrance houses at 500+ subscribers
  • Recruitment pipeline: 10K LinkedIn followers makes talent acquisition and partnership formation dramatically easier
✍️
Content System
5 pillars · 6 post formats · 10 opening line hooks · performance rules
Section 2
A. Content Pillars (5)
Pillar 1: Fragrance × Technology Authority
  • Purpose: Own the "AI + olfaction" intersection — nobody else does this on LinkedIn
  • Topics: AI olfaction research (Google DeepMind), OR gene diversity, algorithmic recommendation systems applied to scent
  • Source material: The 10 drafted posts are mostly this pillar — already done
  • KPI: Saves + shares from tech/AI audience. Target 2%+ share rate.
Pillar 2: Industry Critique Debate
  • Purpose: Drive comments from industry insiders who agree or disagree — both outcomes are wins algorithmically
  • Topics: Reformulation opacity, gender-neutral marketing failures, markup vs. quality correlation, niche vs. designer narrative
  • Rule: Critique the system, never attack a specific brand. Data-backed opinions only.
  • KPI: Comment volume. Target 20+ comments per post in this pillar.
Pillar 3: Olfactory Science Education Trust
  • Topics: Note pyramid mechanics, skin chemistry and drydown, olfactory memory (Proust effect), the 400-receptor anatomy — all from drafted posts
  • Format: Mix of long-form essays (drafted posts) + short carousel "did you know" slides
  • KPI: Saves. Educational content has the highest save rate. Target 5%+.
Pillar 4: Founder / Build Story Personal
  • Purpose: Humanise the brand, build parasocial trust, attract co-founders/investors/partners
  • Topics: "Why I built ScentMatch", "what I learned analysing 23K fragrances", "the hardest decision in building a solo platform"
  • Frequency: 1×/week max — scarcity maintains premium feel
  • KPI: Follow conversion. Founder stories drive the highest follow rate of any format.
Pillar 5: Data Insights from ScentMatch Unique
  • Purpose: No one else has this data. Use it. "I analysed 23,000 fragrances and found..." is a sentence that opens every door on LinkedIn.
  • Topics: Most common top notes by family, rating vs. price correlation, seasonality patterns across the DB, longevity by fragrance house
  • Format: Data post + carousel visualisation. Screenshot ScentGraph to show the network.
  • KPI: Reshares from industry accounts. This pillar generates inbound B2B interest. Target 1 industry repost per data post.
B. 6 Post Formats (with frequency)
FormatFrequencyWhy it works on LinkedInKPI
Long-form essay (the 10 drafted posts)2×/weekDwell time signal — LinkedIn rewards posts people read for 60+ seconds. Your drafted posts are this.Comments + saves
Short punchy post (3–5 lines)2×/weekFast distribution test. Low effort. High frequency. If it gets 50+ likes in 2 hours → expand into long-form.Likes + shares
Carousel / document post1×/weekHighest reach format on LinkedIn (2× normal). Each slide keeps user in feed. 5–8 slides optimal.Impressions + swipe-throughs
Poll1×/week3× normal reach. All votes are engagement signals. Best for audience research + algorithm boost simultaneously.Votes + comments
Image + data point1×/weekA single striking statistic as an image stops the scroll. "400 olfactory receptor types. 800 OR genes. And you bought the same scent as everyone else." Shares + saves
LinkedIn Newsletter issue1×/week (from Day 30)Subscribers get push notifications — free guaranteed reach to your most engaged audience. Compounds over time.Subscriber growth + open rate
C. 10 Opening Line Hooks for LinkedIn
LinkedIn hook rule: The first 2 lines are all most people see before clicking "see more." They must create a knowledge gap or a professional provocation. Never start with "I'm excited to share…" or "In today's post…"
"The fragrance industry has a €30B problem that nobody in it wants to admit." — Industry critique opener. Forces "what problem?" response. High comment rate from insiders.
"I built an algorithm that knows your fragrance better than you do. Here's what it found." — Data authority + personal story. Founder credibility combined with curiosity gap.
"After analysing 23,000 fragrances, I can tell you what the data says about the niche revolution — and it's not what the brands are claiming." — Contrarian data hook. Signals unique access.
"Your nose has 400 types of olfactory receptors. Your genome encodes 800 OR genes. So why does fragrance advice still sound like a horoscope?" — Science + frustration. Validates audience's unspoken irritation.
"3 things I wish someone had told me before I built a fragrance AI:" — Listicle opener. LinkedIn users know the format delivers value. High click-through on "see more."
"The most underrated luxury is wearing a fragrance nobody else is wearing. Here's how to find it." — Identity + aspiration. Resonates with the LinkedIn professional-status mindset.
"We've known since 2002 (Chemical Senses) that scent perception is genetically unique. The industry still sells one-size-fits-all. Why?" — Research citation + industry critique. Signals authority, invites debate.
"I had a conversation with a perfumer last week that changed how I think about personalisation. Here's what they said." — Social proof + narrative tease. "Here's what they said" is the strongest "see more" trigger on LinkedIn.
"Hot take: The €400 niche fragrance and the €40 designer have the same base ingredient in 60% of cases. The rest is marketing." — Controversial claim with implied data backing. Fragrance insiders will comment to confirm or deny.
"What would happen if you applied the Netflix recommendation model to fragrance? We tried it. The results were unexpected." — Experiment format. "Unexpected results" is the highest-performing curiosity trigger on LinkedIn (tested across 10K+ posts in platform studies).
D. Performance Rules
MetricMinimum viableWinning thresholdAction
Impressions (first 24h)200+2,000+Under 200 → posting time wrong OR opening line failed. Change both and repost next week.
Comments (first 24h)3+15+Under 3 → no controversy, no question, no curiosity gap. Add a direct question at the end.
Engagement rate2%5%+Under 2% for 3 consecutive posts → content format pivot (switch from long-form to carousel)
Saves5+30+Saves = people want to reference later = genuine value. Under 5 on educational content → simplify the insight.
Follower adds per post5+30+Under 5 → profile is weak or content isn't converting visits. Add a clear "Follow for more [specific value]" at post end.
Kill rule: Any post topic that gets <2% engagement in 3 consecutive attempts is dead on LinkedIn. Move to the next pillar. Speed of experimentation beats optimisation of a dead format.
📅
90-Day Daily Execution Program
Week-by-week LinkedIn playbook with daily tasks, time estimates and KPI targets
Section 3
Time investment: LinkedIn requires 45–60 min/day — less than Instagram, but more concentrated. Split: 20 min writing/editing, 25 min commenting/engaging, 10 min metrics. Batch write on Sunday for the week.
📅 Post Calendar
Legend: Fragrance × Tech Science Education Industry Critique Founder Story Data Insights
Programme progress calculating…
26 May 2026 7 Sep 2026
30
Total Posts
0
Published
0
Upcoming
0
Overdue
Phase 1
Foundation · Days 1–14
Phase 2
Momentum · Days 15–45
Phase 3
Authority & Viral · Days 43–90

Checkboxes are saved in your browser (localStorage). Click a post title to open the file.

Week 1–2

Days 1–14 · Launch + Foundation

Target: +500 followers · 14 posts · 200 connection requests ▼
DAYS 1–3Setup
  • Optimise LinkedIn profile: Headline = "Building the fragrance intelligence platform · Founder @ScentMatch" — not job title, but value + proof
  • Featured section: Pin ScentMatch URL + 1 best LinkedIn post
  • About section: 3 paragraphs max — problem you solve, how you solve it, who follows you for. End with a CTA.
  • Post #1: Use LinkedIn Post 01 (Fragrance and Identity) — already drafted. Add a direct question at the end.
  • Send 50 connection requests to fragrance industry professionals with personalised note
2h setup + 45min/day Output: Profile complete, Post #1 live KPI: Profile completeness 100%, 1 post live
DAYS 4–7Post + Engage
  • Post #2: Short punchy version of LinkedIn Post 02 (Expert Nose Behind the Algorithm) — 5 lines, end with "What's your fragrance personality?"
  • Post #3: Poll — "How do you choose a new fragrance? A) Recommendation B) Nose test C) Algorithm D) Brand loyalty"
  • Comment 20× per day on posts by fragrance/luxury/beauty accounts with 5K+ followers — add expert knowledge, not just reaction
  • Reply to every comment on your posts within 2 hours
45min/day Output: 3 posts live, 80 comments sent KPI: Post #1 has 200+ impressions, 3+ comments
DAYS 8–14Scale + Review
  • Posts #4–7: LinkedIn Post 03 (What Your Fragrance Says) + 1 carousel (convert post 02 into 6 slides) + 1 short punchy + 1 data point image
  • Week 2 review: Which post got the most comments? That topic gets 2× coverage in Week 3.
  • Continue connection requests: 50/week. Accept all inbound. Follow back relevant profiles.
  • First Newsletter: Set up LinkedIn Newsletter "The Olfactory Intelligence Brief" — don't publish yet, just configure and announce it in a post
45–60min/day Output: 7 total posts, newsletter configured KPI: Followers 300–500. Best post 500+ impressions.
⚙️
Growth Systems
Comment funnel · connection strategy · DM scripts · collaboration system
Section 4
Comment Funnel (20–30/day)
  • Target: Posts by accounts with 5K–500K followers in fragrance, luxury, beauty-tech, AI × consumer spaces
  • Comment structure: Add a genuinely new data point or perspective — never "great post!" Your comment represents your brand. One expert comment on a 50K-follower post is worth 10 posts of your own.
  • Timing: Comment within the first 60 minutes of a post going live — early comments get the most exposure as they sit at the top of the feed
  • KPI: 5–15% of people who see your comment visit your profile. At 25 comments/day: 20–50 profile visits → 5–15 connection requests → 2–5 followers/day from comments alone
Connection Strategy (50 requests/week)
  • Priority 1: Fragrance house employees (marketing, product, R&D) — they are your B2B customers
  • Priority 2: Beauty/luxury journalists and editors — they are your media pipeline
  • Priority 3: Fragrance collectors and enthusiasts with professional profiles — they are your most engaged audience
  • Personalised note template: "Hi [name] — I saw your post/article about [specific thing]. I'm building ScentMatch, a data-driven fragrance advisory platform. The [specific topic they cover] is directly relevant to what we're doing — would love to connect."
  • KPI: 30%+ accept rate. Under 20% = note is too generic.
DM Scripts
DM Script #1: Post-connection opener
Template: "Thanks for connecting [name]. I saw you work in [area] — I'm curious what your take is on [specific trend / their recent post topic]. I'm building ScentMatch partly because [genuine reason related to their area] — would value your perspective."
Goal: Start a conversation, not a pitch. KPI: 40%+ reply rate. No pitch until message 3+.
DM Script #2: Collaboration proposal
Template: "Hi [name] — I've been following your work on [topic]. I have something that might be genuinely useful for a post you could write: access to ScentMatch's dataset of 23K fragrances with performance scores and note analysis. If you ever want to write a data-backed piece on [their area], I'd be happy to run the numbers. No ask, just think it could make for a great post."
Goal: Create a data partnership that earns media mentions. KPI: 1 collaboration per month by Month 2.
🛠
Tools & Workflow
LinkedIn-specific tool stack for a solo operator
Section 5
📝
Notion — Weekly Content Calendar
Planning + Drafting
One database: Post # · Pillar · Format · Hook · Draft · Status · Impressions (filled after posting). Write all 5 posts for the week on Sunday in one session (60 min). Copy-paste to LinkedIn on posting days. Efficiency gain: Eliminates daily writing decisions. 5 posts in 60 min batch vs. 20 min each = saves 40 min/week.
🎨
Canva — Carousels + Data Images
Visual Content
Use the LinkedIn carousel template (1080×1350px). Set brand colours once (gold #c9a96e + dark #0a0a0f). 6-slide carousels take 20 min once the template is set. Key: Slide 1 must be a provocative question or bold claim — it's the thumbnail. Efficiency gain: One carousel template reused weekly saves 30 min per carousel vs. designing from scratch.
📊
LinkedIn Analytics (Native)
Daily + Weekly Metrics
Daily (3 min): Check yesterday's post — impressions, comments, reposts. Record in Notion tracker. Weekly (15 min): Sort last 7 posts by impressions. Top format → replicate. Bottom format → retire or change hook. Key metric to watch: Comments ÷ Impressions. Over 1% = the post triggered genuine engagement. Under 0.3% = broadcast-only content.
🔔
LinkedIn Notifications + Shield App
Engagement Speed
Enable mobile push notifications for comments on your posts. The first 60 minutes after posting are critical — every comment you reply to within that window signals high engagement to the algorithm and keeps the post in feeds. Shield (paid, ~€15/mo) gives deeper analytics than native LinkedIn — post performance over time, best posting times for your specific audience. Worth it from Month 2 onwards.
🤖
Claude — Hook + Post Drafting
Writing Acceleration
Prompt: "Rewrite this LinkedIn post opening line. Current: [your line]. Make it more provocative, under 20 words, and create a knowledge gap that makes professionals click 'see more'. Audience: fragrance industry, luxury, tech." Also use for carousel slide copy: "Turn this paragraph into 6 LinkedIn carousel slides. Each slide = 1 bold claim + 1 supporting sentence. Make slide 1 a question."
📊
Analytics & Decision System
Daily 3-min check · weekly 15-min review · IF/THEN rules
Section 6
Impressions
Total reach. Target: growing 20%/week in Phase 1–2.
Comments
Most important engagement signal. Target: 1%+ of impressions.
Reposts
Viral amplifier. Each repost = 2nd-degree network exposure.
Followers/post
Direct conversion. Target: 10+ new followers per post by Week 6.
IF / THEN Decision Rules
IF
Post gets 20+ comments in first 2 hours → you're in a viral window. Reply to every comment. Write a follow-up post on the same topic within 48 hours. Do not post anything else that day.
IF
3 consecutive posts under 300 impressions → posting time or hook problem. Shift posting to Tuesday 8am. Change opening line to a direct provocative question.
IF
High impressions (>2K) but under 5 comments → content is interesting but not debatable. End every future post with a direct question that has no obvious answer.
IF
Newsletter open rate under 30% → subject line problem. Test: "The fragrance industry won't like this" vs. "3 things your perfume says about your personality." A/B two subject line styles back-to-back.
IF
Connection accept rate under 20% → targeting too broad. Narrow to only first-degree connections of accounts you interact with. More relevant = higher acceptance.
IF
Followers plateau for 2+ weeks → add a new format. If you've only been doing long-form, introduce a carousel. Format novelty alone can break a growth plateau.
⚠️
Failure Prevention
LinkedIn-specific risks with early warning signs and prevention actions
Section 7
🔗
Risk 1: Links in Post Body — The Silent Suppressor
Warning sign: Posts with links get 3–5× fewer impressions than posts without. Prevention: Never put URLs in the post body. Always add "Link in first comment ↓" — then post the link as the first comment immediately after publishing. This is the #1 most commonly broken LinkedIn rule.
📉
Risk 2: Company Page as Primary Channel
Warning sign: Company page posts getting 60–80% fewer impressions than expected. Prevention: The founder personal profile must be the primary posting account. The company page should repost the best personal posts. LinkedIn's algorithm weights personal profiles significantly higher than company pages for organic distribution.
🗣
Risk 3: Broadcast Without Engagement — "Post and Ghost"
Warning sign: Posts get views but comment threads stay short. No inbound connection requests from strangers. Prevention: 20 outbound comments per day is non-negotiable. LinkedIn is a professional network — people follow those they've already interacted with. If you're not visible in other people's comment sections, you don't exist to their network.
🎭
Risk 4: Staying in Safe Territory — No Debate, No Growth
Warning sign: Posts consistently get low comments despite decent impressions. Everyone agrees with you. Prevention: LinkedIn rewards professional disagreement. You don't need to be provocative — you need to take a clear, data-backed position that not everyone shares. "The niche perfume premium is mostly marketing" will generate 10× more comments than "personalisation in fragrance is important." Say something specific and defensible.
⏰
Risk 5: Inconsistent Posting Cadence
Warning sign: Skipping a week, then posting 5 times the next week. Uneven weekly follower growth. Prevention: 3 posts/week minimum, every week, for 90 days. Batch write on Sunday. The LinkedIn algorithm does not reward burst posting — it rewards reliability. A 4-week gap resets your distribution priority almost completely.
🎯
Risk 6: Wrong Audience Building — Vanity Followers
Warning sign: High follower count but low engagement rate. Comments from accounts unrelated to fragrance/luxury/tech. Prevention: Do not accept every connection request — be selective about who you follow back. Quality audience (fragrance professionals, beauty industry, luxury consumers) drives B2B inbound. A 2K engaged professional audience is worth more than 20K passive consumers on LinkedIn.
The 3 Metrics That Matter on LinkedIn
Comments
The single strongest distribution signal. 1 comment = 10 likes in algorithmic weight. Engineer every post to provoke a response.
Reposts
Each repost exposes your content to a fully new network. The viral mechanism on LinkedIn. Target: 2+ reposts per post by Week 6.
Dwell Time
How long people read. Long-form posts that are actually read outperform short posts that are scrolled past. Write for depth, not length.
LinkedIn vs Instagram in one line: Instagram rewards entertainment and emotion. LinkedIn rewards expertise and debate. Your content strategy for each must be built from the ground up — the tone, format, and hooks are completely different, even when covering the same topic.
ScentMatch · AI Avatar Studio · Complete Build Guide

Build Your AI Avatar — Zero to Automated

A complete beginner-to-advanced guide: photorealistic avatar creation, voice cloning, lip-sync animation, multilingual script-to-video pipeline, and full batch automation — optimised for Instagram Reels.

9
Core Modules
$0
Min. Budget to Start
∞
Languages Supported
Batch
Automated Output
🤖
Overview & Tool Stack
The full pipeline at a glance — choose your budget tier, then follow the relevant modules
Module 0
How the Pipeline Works
1
Photo Capture Session
30–60 photos under controlled lighting → feed into avatar training model
↓
2
Avatar Creation
AI generates a photorealistic 3D/2D digital twin trained on your face
↓
3
Voice Cloning
3–10 min audio sample → AI voice model with your tone, emotion, multilingual
↓
4
Script → TTS Audio
Any script (any language) → cloned voice audio file via TTS engine
↓
5
Lip-Sync Animation
Avatar photo/video + audio → perfectly animated talking-head video
↓
6
Instagram Post-Production
9:16 crop, captions, hook overlay, music → upload-ready Reel
Tool Stack by Budget
Free / $0
  • Avatar: HeyGen (free tier) or D-ID free
  • Voice: ElevenLabs free (10K chars/mo)
  • Lip-sync: SadTalker (open-source, local)
  • TTS: ElevenLabs / Coqui TTS (open-source)
  • Edit: CapCut (free), DaVinci Resolve
  • Captions: CapCut auto-captions
Best for: testing the workflow before spending money
$30–80/mo
  • Avatar: HeyGen Creator ($29/mo) — custom avatar training
  • Voice: ElevenLabs Starter ($5/mo) — 30K chars, 10 voices
  • Lip-sync: Hedra ($8/mo) or HeyGen built-in
  • TTS: ElevenLabs API
  • Edit: CapCut Pro ($10/mo)
  • Automation: Make (Integromat) free tier
Best for: consistent weekly content creation
$150–400/mo Pro
  • Avatar: Synthesia Enterprise or HeyGen Business
  • Voice: ElevenLabs Creator ($22/mo) — 100K chars, instant clone
  • Lip-sync: Wav2Lip (local GPU) for max quality
  • TTS: ElevenLabs + OpenAI TTS (gpt-4o-mini-tts)
  • Edit: Premiere Pro or DaVinci Studio
  • Automation: Make Pro + n8n self-hosted
Best for: daily batch output, brand-grade quality
StepFree ToolMid TierPro ToolTime Investment
Photo CaptureSmartphone + natural lightRing light + tripodPro photo session2 hours once
Avatar CreationD-ID / HeyGen freeHeyGen Custom AvatarSynthesia / Unreal MetaHuman30 min setup
Voice CloneElevenLabs freeElevenLabs StarterElevenLabs Creator + OpenAI1 hour once
Lip-SyncSadTalker (local)Hedra / HeyGenWav2Lip GPU / Synthesia2–10 min/video
Post-ProductionCapCut freeCapCut ProPremiere Pro5–15 min/video
AutomationManual + scriptsMake.com freeMake Pro + n8n4 hours setup
📸
Photo Capture Session
The quality of your photos is the single biggest factor in avatar realism — this is not a step to rush
Module 1
What You Need to Capture
Required Angles (minimum 30 photos)
  • Front-facing, eyes open: 10 photos with slight head rotations (±5°, ±10°, ±15°)
  • Left profile (90°): 4 photos — natural, smile, mouth slightly open, neutral
  • Right profile (90°): 4 photos — same set
  • 3/4 left (45°): 4 photos — this is the most used angle in video
  • 3/4 right (45°): 4 photos
  • Slightly down-looking: 2 photos (natural talking position)
  • Slightly up-looking: 2 photos
Expressions to Capture
  • Neutral / resting: lips slightly apart, relaxed jaw — most important
  • Speaking position: mouth slightly open as if mid-sentence
  • Genuine smile: with teeth visible (eyes engaged — not posed)
  • Eyebrow raised: questioning expression
  • Eyes wide: emphasis expression
  • Nodding: slight chin-down agreement
Lighting Setup (Critical)
Option A — Natural Light (Free)
  • Sit facing a large window (north or east facing)
  • Shoot 1–2 hours after sunrise or before sunset
  • Use a white sheet or foam board on the shadow side as a reflector
  • Overcast day = perfect diffuse light — best scenario
  • Avoid direct sunlight — creates harsh shadows in eye sockets
Option B — Ring Light ($25–60)
  • 18" ring light, mounted at eye level, 60–80 cm away
  • Set colour temperature to 5500K (daylight)
  • Place a softbox or diffusion panel if available
  • Add a secondary fill light from one side to avoid flat "ring light look"
  • Keep background plain: white, grey, or solid dark colour
Option C — 3-Point Lighting
  • Key light: 45° left at eye level (main source)
  • Fill light: 45° right, 50% power (reduces shadows)
  • Back/rim light: behind head for hair separation from background
  • All lights: 5500K daylight balanced
  • Result: most professional avatar training data
Technical Camera Settings
  • Camera: Any modern smartphone (iPhone 13+ or Samsung S21+) is sufficient for HeyGen/D-ID avatars
  • Resolution: Shoot in highest resolution available — minimum 1080p, ideal 4K
  • Focal length: Portrait mode or 2× zoom — avoids wide-angle facial distortion
  • Focus lock: Tap face on screen and lock focus before each shot
  • Stabilisation: Use a tripod or prop phone — no handheld wobble
  • Background: Plain, uniform, high contrast to your skin tone
  • Hair: Styled consistently as you want it in videos — pulled back initially shows face best
  • Clothes: Solid colour top, no busy patterns — keeps attention on face
Common Mistakes to Avoid: Glasses (creates reflections + occludes eyes — train without, add digitally later), strong directional shadows across face, photos taken in different lighting conditions (keep all photos in one session), filters or heavy beauty mode applied (train on your real face).
🎭
Avatar Creation
Step-by-step setup for HeyGen (recommended), D-ID, and open-source alternatives
Module 2
Path 1 — HeyGen Custom Avatar (Recommended, $29+/mo)
  1. Sign up at heygen.com → upgrade to Creator plan ($29/mo)
  2. Navigate to Avatars → Create Custom Avatar → Upload Photos
  3. Upload your 30+ photos in the exact angles from Module 1
  4. Set avatar type: Photo Avatar (for still-photo source) or Video Avatar (requires a 2-min consent video — higher quality)
  5. For Video Avatar: record a 2-minute video of yourself speaking naturally to camera — good lighting, clear audio, varied expressions
  6. Submit training — processing takes 4–24 hours
  7. Once approved, your avatar appears in the Avatar Library — test with a short script first
  8. Fine-tune: adjust camera angle, background, framing in HeyGen's editor
Consent requirement: HeyGen requires a recorded consent statement reading a specific phrase — this is a one-time verification step, not ongoing. Required by their Terms of Service.
Path 2 — D-ID (Good Free Tier)
  1. Go to d-id.com → Create Free Account (20 free video credits/month)
  2. Click Create Video → Add a Presenter → Upload
  3. Upload your best front-facing photo (single photo, well-lit, neutral expression)
  4. Add your script text or upload audio file
  5. Select voice (use ElevenLabs integration for your cloned voice)
  6. Generate — D-ID produces lip-synced talking head videos in minutes
  7. D-ID is not trained — it works from a single photo, so consistency is automatic
Path 3 — Hedra (Best Free Lip-Sync)
  • Go to hedra.com — free tier available
  • Upload a portrait photo + audio file
  • Hedra generates highly realistic talking-head video with natural head movement
  • Outputs MP4 — direct Instagram ready
  • No training required — works from any photo
  • Best quality free option as of 2026
Path 4 — SadTalker (Open-Source, Local)
  • Completely free — runs on your local machine
  • Requires Python + CUDA GPU (or runs slowly on CPU)
  • Install via GitHub: OpenTalker/SadTalker
  • Input: portrait photo + audio WAV file → output: MP4
  • Quality is lower than HeyGen but fully unlimited and private
  • Best for: batch generation without API costs
Advanced: Unreal Engine MetaHuman (Hyper-Realistic)
  • Go to metahuman.unrealengine.com — free tool
  • Use MetaHuman Creator to sculpt a digital twin from your reference photos
  • Export to Unreal Engine 5 — animate with audio2face (NVIDIA) or LiveLink
  • Steep learning curve (2–4 weeks) but produces cinematic quality output
  • Renders take significant time/GPU — not suitable for daily batch production
  • Best use case: hero brand video or launch content, not daily Reels
🎙️
Voice Cloning
Clone your voice with natural tone, emotion, and multilingual support
Module 3
Recording Your Voice Sample (Critical Step)
Recording Requirements
  • Duration: Minimum 3 minutes; ideal is 10–30 minutes for premium quality
  • Microphone: Any USB mic ($30+) or even AirPods Pro in a quiet room — no laptop mic
  • Room: Soft furnishings (bedroom), no echo, no background noise — closet full of clothes is ideal
  • Format: WAV or MP3, 44.1kHz, 16-bit minimum — no compression artefacts
  • Consistency: Same distance from mic throughout, stable volume
What to Record
  • Read naturally — don't perform, speak like you do in casual conversation
  • Include varied sentence lengths: short punchy statements + longer explanations
  • Vary your energy: enthusiastic, calm, authoritative, warm
  • Pronounce fragrance-specific vocabulary naturally (oud, vetiver, bergamot, etc.)
  • Record both English AND your target language(s) if multilingual support is needed
  • Read a mix of question sentences, declarative statements, and list-style content
Voice Cloning Tools — Step by Step
🔊
ElevenLabs — Recommended Primary Voice Engine
Best quality · Multilingual · API Access
  1. Create account at elevenlabs.io
  2. Go to Voices → Add a New Voice → Voice Cloning
  3. Upload your audio file(s) — you can upload multiple clips, combined total 10+ minutes ideal
  4. Name the voice (e.g. "My Avatar — EN") and set language
  5. Generate — takes 30–60 seconds
  6. Test with a sample text — adjust stability (0.4–0.6 is natural) and similarity boost (0.75–0.85)
  7. For multilingual: create separate voice clones trained on audio in each language OR use the multilingual v2 model with a single English clone
  8. API key: Settings → API → copy your key for automation scripts
🌍
Multilingual Voice Support
Arabic, German, French, Spanish + 28 more languages
  • ElevenLabs Multilingual v2 model: Takes your English voice clone and auto-generates any of 29 languages — accent is preserved. No separate recording needed. Select the model in the TTS generation step.
  • For Arabic specifically: ElevenLabs supports Arabic with diacritics. Record a 3-minute Arabic sample separately and create a dedicated Arabic voice clone for maximum naturalness.
  • Alternative — OpenAI TTS (gpt-4o-mini-tts): Does not clone your voice but produces highly natural speech for 57 languages. Use as fallback for languages where ElevenLabs quality is inconsistent.
  • Alternative — Coqui TTS (open-source): Free, local voice cloning with multilingual support. Quality is lower but has no usage caps. Install from coqui.ai
Emotion Control in ElevenLabs
  • Add SSML tags in your script to control emphasis: <emphasis level="strong">word</emphasis>
  • Use breaks: <break time="500ms"/> for natural pauses
  • Adjust speaking rate per script segment via API parameter speaking_rate
  • Stability slider: lower = more expressive/varied; higher = more consistent/monotone
  • Add emotional stage directions in brackets in your script: [excited] — ElevenLabs v3 models respond to these prompts natively
Audio Quality Checklist Before Submission
  • ✅ No background noise, fans, traffic, keyboard clicks
  • ✅ No room echo or reverb
  • ✅ No clipping (audio peaking red in editor)
  • ✅ Consistent volume throughout — normalise to -3dB peak
  • ✅ No long silences (trim to remove gaps >2 seconds)
  • ✅ Single continuous take or seamlessly edited clips
  • Use Audacity (free) to clean: Noise Reduction + Normalize before uploading
🎬
Lip-Sync Animation
Animate your avatar to speak with perfect lip-sync — tool-by-tool breakdown
Module 4
HeyGen — Full Integrated Workflow
  1. Open HeyGen → Create Video → New Video
  2. Select your custom avatar from the library
  3. Paste your script into the text field OR upload audio file
  4. If using text: select your cloned ElevenLabs voice (HeyGen integrates via API key in settings)
  5. If using audio: switch to "Audio Upload" mode — your pre-generated ElevenLabs file syncs automatically
  6. Preview at 50% to check sync before generating full video
  7. Export at 1080p → download MP4
Best for: Integrated end-to-end solution with no separate tools needed
Hedra — Best Free Quality
  1. Go to hedra.com → Upload portrait photo
  2. Upload your ElevenLabs-generated audio file (MP3 or WAV)
  3. Select aspect ratio: 9:16 for Reels directly
  4. Generate — Hedra produces natural head movements automatically (not just mouth movement)
  5. Download MP4 — typically 2–4 minutes processing per 60 seconds of audio
  6. No login required for basic free tier
Best for: Free high-quality output, no custom avatar training needed
Wav2Lip — Open Source Maximum Quality
  1. Install from GitHub: Rudrabha/Wav2Lip
  2. Requires Python 3.7, PyTorch, CUDA (GPU recommended — CPU is 10–30× slower)
  3. Prepare: base video (yourself speaking any neutral content, looped) + audio file
  4. Run: python inference.py --checkpoint_path wav2lip_gan.pth --face input.mp4 --audio input.wav
  5. Output: MP4 with replaced lip movement synced to audio
  6. Use GFPGAN post-processing for face enhancement
D-ID — Quick API-Based
  1. D-ID API accepts: image URL + audio URL → returns video URL
  2. REST API call — integrates easily into any automation
  3. Best for: automation pipelines where you want API-first workflow
  4. Free tier: 20 credits/month; paid from $5.9/mo
  5. Image: your face photo hosted on any URL
  6. Audio: ElevenLabs output URL or uploaded file
API endpoint: POST /talks — accepts source_url + audio_url, returns video_url when complete
ToolInputQualitySpeedCostAPI?
HeyGenPhoto/Video + text/audio⭐⭐⭐⭐⭐2–5 min$29/mo✅
HedraPhoto + audio⭐⭐⭐⭐2–4 minFree / $8/moLimited
D-IDPhoto + audio/text⭐⭐⭐1–2 minFree / $5.9/mo✅
Wav2LipVideo + audio⭐⭐⭐⭐Slow (local GPU)FreeLocal
SadTalkerPhoto + audio⭐⭐⭐Slow (local)FreeLocal
⚙️
Script → Finished Video Pipeline
The full end-to-end workflow from a blank text file to an upload-ready Instagram Reel
Module 5
Standard Pipeline (HeyGen + ElevenLabs)
1
Write Script (5–10 min)
Write your script in a Google Doc or Notion. Format: [Hook line] → [3 core points] → [CTA]. Keep to 45–60 seconds for Reels = 100–150 words. Add SSML pause tags where you want natural breaths. Include the language tag if multilingual.
↓
2
Generate Audio in ElevenLabs (2 min)
Paste script → select your cloned voice → select Multilingual v2 model → adjust stability (0.5) and similarity (0.8) → Generate → Download MP3. Listen once to check for mispronunciations — regenerate any sentences that sound off.
↓
3
Generate Avatar Video in HeyGen or Hedra (5 min)
Upload audio to your preferred lip-sync tool. Select your avatar. Set background (remove BG or use custom branded background). Set aspect ratio 9:16. Generate. Download MP4. Review: check sync quality in the first 5 seconds — if off, re-upload audio.
↓
4
Post-Production in CapCut (5–10 min)
Import avatar MP4. Add auto-captions (CapCut → Captions → Auto-caption). Style text: white bold, black stroke, bottom third position — ScentMatch brand font if set. Add hook text overlay for first 1.5 seconds. Add background music at 15–20% volume. Trim start/end. Export 1080×1920.
↓
5
Upload to Instagram (2 min)
Write caption (hook + 3 hashtags max). Add cover frame: first 3 seconds auto-selected or choose manually. Post as Reel. Schedule via Instagram native scheduler or Buffer for optimal time.
Total time per video: 20–30 minutes manual. 3–5 minutes once automated (Module 6).
Script Templates for Fragrance Content
[Hook] "Most people have been buying the wrong fragrance their entire life. Here's why — and what to do instead."
[Structure] Point 1 (10–15 sec) → Point 2 (10–15 sec) → Point 3 (10–15 sec) → CTA: "Save this, link in bio for your full profile."
[Multilingual variant] Write the script in English first. Run through ElevenLabs Multilingual v2 for Arabic/German/French audio — the AI translates and reads in your cloned voice. Review translation separately before publishing.
🔄
Batch Automation Pipeline
Automate the entire script-to-video workflow for batch content creation
Module 6
Architecture Overview
A
Google Sheets Script Queue
One row per video: Script text | Language | Voice settings | Background | Title | Target date. This is your content calendar + pipeline input.
↓
B
Make.com Trigger: New Row in Sheet
Make watches the sheet. When status = "Ready", it triggers the pipeline automatically.
↓
C
ElevenLabs API → Generate Audio
Make sends script text + voice_id to ElevenLabs API → receives MP3 URL → saves to Google Drive.
↓
D
D-ID or HeyGen API → Generate Video
Make sends avatar image URL + audio file URL to D-ID /talks endpoint → polls until video_url is returned → downloads MP4 to Drive.
↓
E
Notification + Review Queue
Make sends Telegram/Email notification with video link for your 60-second quality review. Approved → manually uploaded, or further automated via Buffer API.
ElevenLabs API Call (Python)
import requests XI_API_KEY = "your_elevenlabs_api_key" VOICE_ID = "your_cloned_voice_id" def text_to_speech(text, output_path="audio.mp3", language="en"): url = f"https://api.elevenlabs.io/v1/text-to-speech/{VOICE_ID}" headers = {"xi-api-key": XI_API_KEY, "Content-Type": "application/json"} payload = { "text": text, "model_id": "eleven_multilingual_v2", "voice_settings": {"stability": 0.5, "similarity_boost": 0.8} } r = requests.post(url, json=payload, headers=headers) with open(output_path, "wb") as f: f.write(r.content) return output_path
D-ID API Call (Python)
import requests, time DID_API_KEY = "Basic your_did_api_key" AVATAR_IMG = "https://your-cdn.com/avatar-photo.jpg" def generate_video(audio_url, output_path="video.mp4"): headers = {"Authorization": DID_API_KEY, "Content-Type": "application/json"} # Create talk r = requests.post("https://api.d-id.com/talks", headers=headers, json={ "source_url": AVATAR_IMG, "script": {"type": "audio", "audio_url": audio_url} }) talk_id = r.json()["id"] # Poll until ready while True: status = requests.get(f"https://api.d-id.com/talks/{talk_id}", headers=headers).json() if status["status"] == "done": video_url = status["result_url"] break time.sleep(8) # Download with open(output_path, "wb") as f: f.write(requests.get(video_url).content) return output_path
Make.com Setup (No-Code Automation)
  1. Create free account at make.com
  2. New Scenario → Google Sheets trigger: "Watch New Rows"
  3. Add HTTP module: POST to ElevenLabs API (map script column)
  4. Add Google Drive module: Save audio file from response
  5. Add HTTP module: POST to D-ID API with audio URL
  6. Add Loop + HTTP GET: Poll D-ID until status = "done"
  7. Add Google Drive: Save final MP4
  8. Add Email/Telegram: Notify with video link
  9. Schedule: Run every hour to check for new "Ready" rows
Batch Processing: 30 Videos in One Session
  1. Prepare 30 scripts in Google Sheet (one Sunday session)
  2. Set all rows to status = "Ready"
  3. Make triggers automatically — processes one at a time
  4. ElevenLabs + D-ID can handle concurrent requests on paid tiers
  5. Typical throughput: 6–8 videos/hour on mid tier
  6. 30 videos completed overnight with zero manual work
  7. Review queue next morning: approve/reject in 30 min total
  8. Schedule approved videos via Buffer (one week ahead)
🔒
Consistency System
Maintain identical appearance and voice across every video — brand trust through uniformity
Module 7
Visual Consistency Checklist
  • One canonical avatar photo: Store a single master portrait file — use this exact file for every video generation, never swap it out
  • Fixed background: Choose one of: (a) solid dark background matching ScentMatch palette, (b) custom branded set, or (c) transparent BG + consistent overlay in CapCut
  • Framing template: In CapCut, create a template project with fixed crop, position, and scale — always start from this template
  • Text overlay style: Lock one font, size, colour, and position in your CapCut template — never deviate
  • Intro/outro frames: 0.5-second branded colour card at start and end — same every video
  • Thumbnail style: Every Reel cover = avatar face + 3-word hook text — same font/style
Voice Consistency Protocol
  • Lock voice settings: Document exact ElevenLabs parameters — voice_id, model_id, stability (0.5), similarity_boost (0.8) — and never change them
  • Save as API preset: If using Make.com, hardcode these values in the scenario — no manual selection
  • Per-language voice IDs: Create and document a separate clone for each language you publish in — EN clone, AR clone, DE clone, etc.
  • Pronunciation library: Keep a running list of words ElevenLabs mispronounces — add phonetic spellings to your script template (e.g. "veh-TEE-vair" for vetiver)
  • Audio post-processing: Apply identical EQ preset in Audacity to all outputs — small boost at 2–4kHz adds presence and clarity
Master Asset Library (Keep These Backed Up)
AssetFileWhere StoredNever Change
Master avatar photoavatar_master_v1.jpg (4K, 9:16 crop)Google Drive + local backup✅ Lock permanently
ElevenLabs voice cloneVoice ID string (copy from dashboard)Documented in Notion✅ Lock permanently
CapCut templatetemplate_reels_v1.capcutCapCut cloud + localVersioned — update = new version
Brand colours#0d0d0d bg, #c9a96e accent, #e8e0d5 textDocumented in Notion✅ Match ScentMatch identity
Background musicbackground_track_v1.mp3 (royalty-free)Google DriveUse same track for 2–4 weeks minimum
Intro/outro cardintro_500ms.mp4, outro_500ms.mp4Google DriveVersioned — seasonal updates OK
📱
Instagram Optimisation
Aspect ratio, hooks, captions, posting strategy — everything specific to Reels
Module 8
9:16
Aspect ratio — always. 1080×1920px. Never post 16:9 avatar video.
1.5s
Hook window — viewer decides in 1.5 seconds. Hook text must be on screen from frame 0.
45–60s
Ideal Reel length — long enough for value, short enough to replay.
3 max
Hashtags per post — Instagram 2026: fewer, more targeted hashtags outperform hashtag dumps.
Reel Structure for Avatar Content
  1. 0–1.5s — Hook frame: Bold text overlay on first frame + avatar already speaking. Never show a static title card — algorithm counts this as a pause.
  2. 1.5–5s — Setup: Avatar establishes the core tension or question. No introduction — viewers don't care who you are until you've earned their attention.
  3. 5–40s — Core value: 2–3 clear, specific points. Each point = 1 sentence max. Avatar gestures naturally (HeyGen generates subtle head movement automatically).
  4. 40–55s — Payoff: The most actionable or surprising insight. Save the best for near the end — this maximises completion rate.
  5. 55–60s — CTA: Single, specific call to action. "Save this post" is the highest-value CTA (saves trigger Feed distribution). Second best: "Link in bio for your full profile."
Caption Strategy
  • First line = hook — same as the spoken hook or a variation. Searchable keyword in first 5 words.
  • Line 2: Expand on the value — 1 sentence
  • Line 3: "Save this for later 🔖" or "Drop your fragrance family below ↓"
  • Hashtags: 3 targeted tags only — one broad (#perfume, #fragrance), one niche (#nicheperfume, #olfaction), one brand (#scentmatch). Place after the main text.
  • No links in caption: Instagram suppresses link-in-caption posts. Always "link in bio."
Posting Time Windows
  • Primary: Tuesday–Thursday, 7–9pm local audience time
  • Secondary: Saturday 10am–12pm
  • Avoid: Monday morning, Friday night, Sunday evening — lowest engagement windows
Hook Text Overlays — Copy/Paste Templates
Format 1 — Contrarian: "You've been buying the wrong fragrance. Here's why." (white bold text, bottom-left, visible by frame 1)
Format 2 — Number: "3 fragrances that smell like money (and are under €50)"
Format 3 — Question: "What does your signature scent say about your personality?"
Format 4 — Secret: "The note perfumers don't put on the bottle (but changes everything)"
Format 5 — Result: "I tested 400 fragrances so you don't have to. Here's the ranking."
✨
Advanced Human-Like Tips
Make your avatar progressively more natural, engaging, and indistinguishable from real video
Module 9
Voice Naturalness Techniques
  • Add filler pauses: Insert <break time="300ms"/> mid-sentence where you'd naturally breathe — not just between sentences
  • Vary sentence length: Mix 3-word punchy lines with longer flowing sentences — monotonous sentence rhythm is the #1 tell of AI voice
  • Script contractions: Write "you'll" not "you will", "it's" not "it is" — cloned voices handle contractions more naturally
  • Include micro-hesitations: Add "— " (em-dash) where a natural speaker would briefly pause mid-thought
  • Laugh cues: ElevenLabs v3 responds to [laughs lightly] and [sighs] inline in scripts
  • Record real laugh separately: Record a genuine laugh and splice into the audio file in Audacity — no AI laugh sounds truly human yet
Visual Naturalness Techniques
  • Add blink layer: In CapCut, overlay a subtle eye-blink video clip at irregular intervals — some tools don't animate blinks naturally
  • Micro-movement layer: Add a very subtle "breathing" scale animation (99%–100.5% scale, 4-second cycle) in CapCut — simulates natural chest movement
  • Depth of field: Slight gaussian blur on background behind avatar (CapCut background blur tool) — makes avatar feel physically present in a space
  • Lighting variation: Add a very subtle warm light flicker preset in CapCut at 3–5% opacity — simulates natural environmental light shift
  • Real hands occasionally: For 1 in 5 videos, show real hands holding a bottle in frame — breaks pattern, confirms reality
  • Wardrobe variation: If using HeyGen custom avatar, create 3 avatar versions with different clothing — rotate to avoid "same video recycled" impression
Long-Term Quality Improvement System
  • Monthly avatar refresh: Every 30 days, record a new consent/base video for HeyGen — the model improves with more data over time
  • A/B voice settings: Every 2 weeks, test one variable: stability +/- 0.1. Track completion rate. Keep whichever wins. Document in Notion.
  • Comment mining for naturalness gaps: If viewers comment "this feels scripted" or "AI voice?", it's a signal — increase pause frequency and sentence variety in next batch
  • Blend real + avatar: The highest-performing channel strategy is 70% avatar videos + 30% real face videos. The real videos build trust; the avatar videos enable scale. Never go 100% avatar — authenticity remains the primary currency on Instagram.
  • Language-specific optimisation: Each language has different natural speech rhythm. Record a native speaker reading your Arabic/German/French scripts and compare to ElevenLabs output — use it to write SSML adjustments into your pipeline.
  • Synthesia Studio (advanced): Once budget allows, Synthesia's Studio tier allows facial expression programming — map specific expressions (raised brow, slight smile, concerned) to specific script lines. This is the highest realism tier currently available commercially.
70/30
Avatar/Real split — optimal for trust + scale balance
Monthly
Avatar model refresh cadence — improves realism over time
3 vars
Wardrobe variations — rotate to avoid repetitive appearance
90-Day Avatar Mastery Roadmap
PhaseFocusDeliverableTime Investment
Week 1–2Photo shoot, avatar training, voice cloneFunctional avatar + voice on HeyGen/ElevenLabs8–12 hours total
Week 3–4First 10 manual videos — learn the pipeline10 published Reels, workflow documented3–4 hours/week
Week 5–8Set up Make.com automationSemi-automated pipeline: script → audio → video queue4 hours setup, 1h/week
Week 9–10First multilingual batch5 videos in 2 languages from single script batch2 hours setup
Week 11–12Full batch automation + consistency audit30+ videos/month with <2 hours manual time1–2 hours/week maintenance
Month 4+Optimise and scaleA/B testing results, refined voice settings, expanded language rosterOngoing — monthly review
Disclosure: Instagram's community guidelines require disclosure when content features AI-generated faces or voices in ways that could mislead viewers. Best practice: add "AI-assisted content" or "#AIavatar" in your bio or story highlights — this builds trust rather than eroding it, and positions ScentMatch as a technology-forward brand.
ScentMatch · LinkedIn Slide Studio

To Publish — Carousel Builder

Select a post from the schedule, edit the slide design live, preview all 4 canvases, and download as PDF ready to upload to LinkedIn.

Cover
Body
Recs
Your Turn
Theme:
🎨 Style Editor CLICK TO EXPAND ›
🔧 Per-Slide Overrides CLICK TO EXPAND ›
✍️ Post Text Editor
Slide 1 · Hook
Slide 2 · Body
Slide 3 · Recommendations
Slide 4 · Your Turn
🎬
Loading script…
Could not load script file.
🖼
👁 Slide Preview