I've been quiet on X for a while. Not because I stopped building, but because I was building HairOver.
Before posting anything, I wrote the whole plan down in a spreadsheet: 30 days of posts, a content mix, a funnel, metrics, creator outreach, and DM templates. This post is that plan.
I'm publishing it before I have the results. Once the results are in, it's too easy to rewrite the plan so it looks like it predicted them.
The core decision: two accounts with different jobs
| Founder account (me) | HairOver account | |
|---|---|---|
| Job | Build trust, tell the story | Show the product |
| Cadence | 4–5 original posts a week, meaningful replies every day | 5–7 posts a week, reply to users and creators every day |
| Rule | Founder-led. Not a sales account. | Visual, simple, playful |
| Framework | Story → Emotion → Lesson → Product | Problem → Visual → Curiosity → Product |
The two accounts work together 1–2 times a week, with a rule I like: the founder account explains, the HairOver account demonstrates. Quote and reframe each other's posts. Don't post the same thing twice.
Why split it: a founder account that keeps saying "download my app" stops being worth following. A product account that writes long reflections gets scrolled past. Giving them different jobs lets each one stay what it is.
The content mix
Founder account
| Pillar | Share | Why |
|---|---|---|
| Founder story | 30% | Make the journey human |
| Build in public | 25% | Progress, decisions, failures |
| AI / technical insight | 15% | Use the developer audience I already have |
| Product / UGC | 15% | Show proof without hard selling |
| Startup lessons | 15% | Useful even if you don't care about HairOver |
HairOver account
| Pillar | Share | Why |
|---|---|---|
| Transformations | 35% | The main visual growth engine |
| Hair education | 20% | Searchable, shareable |
| UGC / community | 15% | Social proof and participation |
| Product demos | 10% | Show the product quickly |
| Polls / questions | 10% | Conversation |
| Memes / trends | 5% | Reach |
| Product updates | 5% | Keep users informed |
Transformations get the biggest share because HairOver's value is easiest to show visually. Same face, five different haircuts. Which one wins? explains the app faster than any description could.
The funnel
Each stage has a single metric, so I can tell where people drop off:
| Stage | Goal | What moves it | Metric |
|---|---|---|---|
| 1 | Attention | Transformations, hooks, founder story | Impressions |
| 2 | Curiosity | Profile visits, replies, video completion | Profile visits |
| 3 | Trial | Site or Play Store | Clicks / installs |
| 4 | Activation | Upload a photo + first generation | First generation rate |
| 5 | Retention | Repeat generations | 7-day / 30-day retention |
| 6 | Advocacy | Share result, UGC, referral | UGC shares |
The stage I care about most is 4. Impressions are easy to get excited about. A stranger uploading their own photo and generating a look is the first sign the product is doing something for them.
The rules
- Use native short video. Don't open a video with the logo.
- Ask one simple question per post.
- UTM links on everything, so I can measure X → site → install.
- Double down on formats that produce installs, not just likes.
- No engagement bait. Replies should add an actual thought.
- No hashtag stuffing.
#HairoverChallengeis only for the challenge campaign.
The 30-day calendar, in short
The calendar alternates between the two accounts every day. A few of the anchor posts:
- Day 1 (founder): the comeback post. "I've been quiet on X for a while… It's called Hairover."
- Day 3 (founder): the problem. "You find a hairstyle you love. Save it. Show it to your barber… 'Why does it look so different on me?'"
- Day 4 (HairOver): "Same face. Five completely different hairstyles. Which one wins?"
- Day 9 and Day 23 (founder): Build Log #01 and #02
- Day 15 and Day 29 (founder): two-week and one-month recaps
- Day 27 (founder): transparent numbers, even when they aren't good
- Day 30 (HairOver): the
#HairoverChallenge. One photo, five hairstyles, pick the one you'd actually get.
Creators and outreach
Four types of partners, each with a different offer:
| Type | Size | Approach |
|---|---|---|
| Micro creators (hair, beauty, grooming, fashion, AI) | 5k–100k | Free generations; ask for a transformation |
| Barbers and stylists | 1k–50k | Test HairOver with client-style examples |
| AI creators | 5k–100k | A product experiment or demo |
| Existing users | Any | Feature their result |
The outreach message ends with "No pressure to post — I'd genuinely love your feedback first." I meant it. At this stage, a creator's honest reaction is worth more than one sponsored post.
A problem with my own plan
When I reread the calendar, I noticed that some posts are written for events that hadn't happened yet:
- Day 19: "Someone used Hairover today and sent us their result."
- Day 23: Build Log #02 still has placeholders:
[feature],[improvement],[fix].
Writing posts ahead of time is fine. Publishing them before they're true isn't. So my rule for the calendar: a post that describes an event only goes out if the event actually happened. If it didn't, the slot gets a different post. The whole point of building in public is that the posts are true.
What I'm measuring
Reviewed weekly: impressions, profile visits, link clicks and CTR, Play Store visits, first generation rate, repeat generation and 7-day retention, UGC posts and mentions.
Reviewed monthly: paid conversion and revenue that can be traced back to X.
What I don't know yet
All of it. This is a plan. I don't know whether founder posts will drive installs, whether transformations will get reach, or whether any creator will reply. The results post will include the real numbers, including the ones that make the plan look wrong.
Key takeaways
- Two accounts, two jobs: the founder explains, the product demonstrates.
- Give every funnel stage one metric; activation (first generation) matters most.
- Judge formats by installs, not likes. Use UTM links to know which is which.
- Pre-written posts are fine. Pre-written events aren't. Only publish what actually happened.
- This is the plan, published before the results.