AI CREATOR LAB
the official Fanvue AI creator program*
📓 Workbook
You'll have the IG research system that removes guessing forever — the outlier board — and today's IG slate built from it, using inventory you already produce. Nothing new to create; everything smarter to aim.
You've been running the research system since 3-5 — the swipe file, the daily 15, the remix tools. This lesson is its Instagram tuning: the outlier board is your swipe file's IG section with the discipline turned up, because Instagram punishes randomness and rewards pattern-matching harder than anywhere. The creators who grow here aren't more creative — they're better students of what's already working, and by now that's just your morning routine pointed at one platform.
Promote the IG section of your swipe file to a proper outlier board (15 minutes). You seeded it in 3-5; now make it IG-sharp. Confirm you're tracking three creator accounts reaching the audience you want — same niche, ideally 5–20× her size (big enough to learn from, small enough to be relevant). For each, hold at least one current outlier: a reel or post with clearly outsized numbers versus that account's normal. (An account averaging 2k views with one reel at 40k — that's an outlier. A million-follower account getting a million views is just Tuesday.)
Decode each outlier — three notes in the workbook: the hook (what the first second shows/says) · the structure (what happens, in what order) · why it worked (your one-sentence theory). Three outliers, nine notes. That's the board.
Adapt one pattern — never copy the post. Take the strongest pattern and make her version: same hook mechanism, same structure, her face, her scenes, her voice. The pattern is transferable; the content is hers. (This is also the line between research and theft — patterns are free, posts are someone's.)
every adapted pattern tests ONE variable from your best guess (hook, scene, caption style) — post it, read the result. The board refreshes on the daily 15; outliers go stale in weeks, and at daily volume you'll ride 2–4 live patterns at once (3-5's rule).
| Post | Warm-up (wk 1–2) | Full volume (wk 3+) | Source |
|---|---|---|---|
| Reels | 1/day | 3–5/day (pattern remixes + bankers) | Next lesson makes these; Reel Studio + your keepers |
| Grid post or carousel | 2–3/week | ~1/day | Straight from the daily batch |
| Stories | most days | daily | 20-second job inside the daily ten (next lesson) |
| Trial reels | — | on top of the 3–5 | Next lesson — the testing lab |
~20 minutes; the board then lives inside the daily 15.
The outlier board exists with three decoded patterns, and today's IG slate — including an adapted-pattern reel — is on paper. You now aim before you shoot, permanently.
You're probably looking at accounts too big (everything's huge) or too small (nothing has signal). Recalibrate to 5–20× her size and sort their reels by views — the spikes announce themselves against the account's own average.
Test: could someone see hers and theirs side by side and call it the same post? Then it's too close — change the scene, the words, the specifics, keep only the mechanism. Hook mechanisms ("get ready with me," "things nobody tells you about ___") belong to everyone; executions belong to their makers.
One flop is one data point — the system's whole design is that a flop costs you one slot and teaches you one variable. Check the theory note: was it a reach pattern when you needed a personality pattern? Adjust one variable, run it again in the next batch. The board wins on iteration, not on first tries.
Same answer as X: a couple of relevant ones at most, zero walls. Distribution comes from the pattern and watch-time, not tag spam. (Yes, everyone's cousin disagrees. The tracker will settle it with your own numbers.)
The system knows what to make — next lesson covers making it: reels and Stories, the two IG formats that do all the heavy lifting, straight from the studio.