the official Fanvue AI creator program*
πŸ”’ Heads up β€” finish the previous lesson first. The course works in order.

Phase 3 · The Content Engine · ⏱ 45 min

Research: Find What Works

β–Ά Lesson video coming soonEverything below stands on its own.

When you're done

By the end of this lesson you will have:

Where this goes: every production block from lesson 3-3 (The Content Factory) onward shoots at least one set from these saves. Motion formats become reels in lesson 3-4 (Reels That Feel Real), and winning images get multiplied in lesson 3-5 (Carousels & More-Like-This).

Why this matters

Here is the doctrine from lesson 3-1 (The Volume Game), now with a system behind it: research beats imagination. You will never guess your way to viral. Every niche's feed is a free, real-time report on what audiences want this week, with the numbers showing. Winners collect what is overperforming, work out why it worked, and ship their own version while the format is still hot.

The AI advantage stacks on top of that. A human creator who spots a winning format needs a new shoot to copy it. You need a generation. You can out-ship the person who invented the format. That is why research comes before production in this phase: the factory you build next lesson is only as good as what you feed it.

Do not trust raw numbers. This niche is full of bought views. A faked 40,000-view reel on a 2,000-follower account looks exactly like a real winner, and a student who remixes it spends a week copying a pattern that never actually worked. That is why nothing gets saved without the 30-second bot check you learn below. It runs on every find, forever.

One more thing this lesson protects: her accounts. You will remake other creators' patterns every day, and there is a hard line between remaking a pattern and stealing a post. The rules section below draws that line, and it is not negotiable.

What you need to know

Where this lesson happens: entirely on the social platforms (Instagram, TikTok, X), on a research login. Bopgpt is not involved today.

The 30-second bot check, defined before you use it. Every outlier must pass before it gets saved:

If two of the three checks feel off, skip it. There are more outliers tomorrow. There is no recovering a week spent remixing a fake one.

The three rules of remixing. These keep her accounts alive. Break them and the account is gone:

The steps

  1. Set up your research login and its save spots. On Instagram, TikTok, and X: sign in to your own personal account on whichever of the three you have (logged-out browsing works where the platform allows it, but platforms show logged-out visitors very little).

  2. On Instagram: confirm you can save a post (the bookmark icon under a post) and create a saved collection named after her niche.

  3. On TikTok: confirm you can add a video to Favorites.
  4. On X: confirm you can bookmark a post.

You only need the platforms that load for you today. If Instagram throws a login wall and you have no personal account, work the others and see If you're stuck.

  1. Pick your study accounts: our study list first, then three accounts in her niche. On Instagram (and whichever other platforms you are using):

Study these first, the accounts we point every student at: coming soon

Then find three accounts in her niche yourself:

  1. Search two or three of her niche's terms in the platform's search bar.
  2. Open the top creator results.
  3. On each promising profile, open the suggested or similar accounts row. The platform does the niche-mapping for you.
  4. Open each candidate's reels or videos tab and look at the view counts.
  5. Keep accounts that are small but working: roughly 1,000 to 20,000 followers. Big enough to prove demand, small enough that their playbook applies to a new account like hers. (lesson 4-3 (Instagram: The Content System & Cross-Posting) raises this bar once she has a real follower count of her own.)
  6. For each of the three keepers, scroll the last 15 posts and learn its normal. You need that number in the next step.

Three accounts is the starter list. The daily 15 in step 6 grows it to 8 to 10 accounts within a month. Stay at three and those three accounts' taste becomes the limit of what she ever posts.

  1. Spot the outliers on your three accounts. On Instagram (or the platform you are working): scroll each account's grid and look for the posts sitting far above that account's normal. Most posts cluster near the average; one or two tower over it. Anything at 5 times the normal or more is a candidate. A candidate is not an entry until it passes the next step.

  2. Run the 30-second bot check on every candidate. On the platform: open the candidate and run the three checks you learned above: the view-to-comment ratio, the comment quality, the wake. Two of three off means skip it.

The comment-quality read, delegated if you want a second opinion. Paste this into Bopgpt's AI chat tab:

Here are 15 comments from one post, pasted exactly: [paste]. The post has [views] views and [comments] comments. Sort the comments into SPECIFIC (questions, quotes of the caption, disagreements, real reactions) and GENERIC (emoji strings, "so pretty", "follow me"). Give me the split and one line: does this look like real engagement or bought reach?

  1. Write the three decode notes for each outlier, then save it. On Instagram, TikTok, or X: for each outlier that passed, make the three decode notes (hook, structure, because-theory), then save it into the niche collection. Type the notes into the save's note field where the platform offers one; otherwise say them out loud before you hit save. A save without the three notes cannot be turned into a recipe later, so it does not count.

Example decode, so you know what done looks like:

  1. Pick tomorrow's patterns and install the daily 15. On Instagram, TikTok, or X: from today's decoded saves, pick the 2 to 4 patterns your first production block will shoot, and say them out loud. From tomorrow, this is the first 15 minutes of the block time you set in lesson 3-1 (The Volume Game):

  2. 10 minutes: scan your study accounts and each platform's discovery surface. Save, vet, and decode 1 to 3 new outliers. Delete stale saves; formats age out in weeks.

  3. 5 minutes: pick today's 2 to 4 patterns. You must be able to say them before Bopgpt opens.

At least one of each day's three production sets comes from these saves. lesson 3-3 (The Content Factory) teaches the block that shoots them tomorrow, and its starter menu covers any set your research did not claim. Later, once her remixes are posting to real traffic, each one gets a 48-hour numbers check and the pattern gets kept or killed; lesson 4-3 (Instagram: The Content System & Cross-Posting) runs that loop on Instagram.

  1. Post your best find in #content. On Discord: post your best decoded outlier in #content as a pattern note in your own words (hook, structure, theory), never as a re-upload of the post itself.

Do it now

You're done when

Your research login holds three vetted, decoded outliers in its saves, and you can say tomorrow's 2 to 4 patterns without looking. You will never again sit down to a blank "what should I post?" The market answers that question every morning, and you are the one who reads the answer. Mark the lesson complete.

If you're stuck

I can't tell an outlier from a normal post

Divide by that account's typical numbers, never by absolute numbers. Scroll the account's last 15 posts; most sit near one level, and anything at 5 times that level stands out on sight.

It passed two checks and failed one

Let the wake break the tie. Real accounts chase their own spike: same pattern, next posts, visible in the grid within days. A spike its owner never followed up on is a spike its owner did not believe. Do not believe it either.

Instagram keeps throwing a login wall at me

Common for anyone without a personal Instagram account, and X behaves the same way logged out. If you have a personal account, use it. If you do not, do not create one today: work the platforms that load for you and let the Instagram saves stay thin for now. If Instagram still walls you when lesson 4-3 (Instagram: The Content System & Cross-Posting) builds on this system, that is the day to create a personal research account: yours, never hers. Thin Instagram saves for two weeks are fine. A skipped daily 15 is the real problem.

My niche has no accounts in the 1,000 to 20,000 range

Widen the search terms one level (from "gym girl" to "fitness creator," from her micro-niche to its parent niche). The suggested-accounts row on any mid-size profile will surface smaller neighbors within two or three hops.

How many patterns should I ride at once?

2 to 4 live patterns, refreshed as they cool. Once her posts have real numbers, one of them is always a banker: a pattern that already worked for her, kept running while new ones get tested. Fewer than 2 and you are under-testing. More than 4 and you cannot read what is working.

Next up

Research is live and tomorrow's patterns are picked. Next lesson builds the factory: the daily production block in Bopgpt that turns those patterns into 8 or more post-ready pieces in one sitting.