Content research: how to know what to post before you film

Monday you know you should post. Tuesday you scroll for forty minutes to “get inspired” and close the app with nothing. Thursday you film whatever occurs to you at 6pm. That gap isn’t a creativity problem — it’s a research problem, and it has a method.

Last updated 12 August 2026

Almost everyone who posts consistently has the same week. Monday you know you should film something. Tuesday you open Instagram to “get inspired”, scroll for forty minutes, and close it with nothing. Thursday you film whatever occurs to you at 6pm. It does 300 views, and the reason it did 300 views was decided long before you pressed record.

That gap — between deciding to post and knowing what to post — is what content research fills. It is not a creativity problem and scrolling is not the solution to it. It is a research problem with a repeatable method, and the method takes about forty-five minutes a week.

This page is the whole method end to end. Each stage has a deeper page beneath it if you want to go further.

What content research actually is

Content research is the work of finding out what your audience already responds to, in enough structural detail that you can build something new on top of it. Three things it is not:

• It is not scrolling. Scrolling shows you what an algorithm thinks you want to watch, which is close to the opposite of what you need. You are researching a market you sell to, not consuming a feed built for you.
• It is not copying. The output of good research is a structure, not a subject. Two videos can share a shape and have nothing else in common.
• It is not looking at the biggest accounts. A million-follower account gets a million views for reasons that will never transfer to you.

The single most useful idea in this whole page

Judge a post against its own account’s baseline, never against other accounts.

A post with 40,000 views on an account that normally gets 60,000 is a failure, and copying it will teach you nothing. A post with 9,000 views on an account that normally gets 1,200 is an outlier — something in that video overcame the account’s usual ceiling, and whatever that something is, it is the only thing worth studying.

Almost nobody does this. They look at the biggest numbers on the biggest accounts and copy things that worked because of audience size. Sorting by outlier rather than by absolute views is the entire difference between research that changes your output and research that wastes an afternoon.

The four inputs

Everything worth knowing comes from one of four places, and they are not equally weighted.

1. Your own back catalogue

The most under-used and the most reliable, because it is the only source measured on your exact audience. Take your last thirty posts, find the median view count, and mark everything that beat it by two times or more. Those are your outliers. Three or four of them will have something in common, and that thing is your format.

2. Accounts one tier above you

Not the giants. The accounts three to ten times your size in your niche — close enough that their audience is your audience, far enough ahead that they have already solved things you have not. Ten to fifteen of them is a working reference set. This is the bulk of the work, and competitor content analysis covers exactly what to record.

3. Your audience’s own words

DMs, comments, the questions people ask in the first five minutes of a sales call, the objections in your support inbox. This is where the language comes from, and using the words your audience actually uses is most of what makes a hook land. It is also the cheapest source and the one people ignore because it does not feel like research.

4. Adjacent niches

Formats travel between niches long before they saturate. A structure that is exhausted in personal finance is often brand new in home renovation. Watching one or two unrelated categories is how you arrive early rather than fourth.

The weekly loop, step by step

Step 1 — Build the reference set (once, then maintain)

Ten to fifteen accounts, one tier above you, same audience. Save them somewhere you will actually revisit. Add one and drop one every few weeks so the set does not go stale.

Step 2 — Find the outliers, not the top posts

For each account, eyeball the median view count of the recent grid, then pick out anything running two to three times above it. On a normal week across fifteen accounts you will find between four and ten. That is your research material for the week — everything else is noise.

Step 3 — Read structure, not subject

For each outlier, answer five questions and nothing else:

• What is the first spoken line, word for word?
• What is on screen in frame one?
• How many seconds until the first cut?
• Where does the payoff land — middle or end?
• What would make someone send this to a specific person?

Five answers, five outliers, and the pattern shows up fast. If you want the underlying taxonomy for the first two questions, the six hook structures covers it.

Step 4 — Convert a finding into your own idea

Take the structure and refill it with your expertise. The test of a good conversion is that nobody could tell which video you learned it from.

A finding is not an idea until it has a hook line written down. “Do a myth-busting one” is not an idea; “Soreness is not a measure of a good workout” is. The idea generation guide covers this conversion in detail, including the nine sources that produce ideas without any competitor research at all.

Step 5 — Keep a bank, and film from the bank

Never film from a blank page. Research produces a queue of eight to twelve written hook lines; filming days draw from the queue. Separating the two is what stops the Thursday-at-6pm video from happening, and it is the single highest-return habit change available here.

How much time this actually takes

Forty-five minutes a week, once the reference set exists. Roughly twenty minutes finding outliers, twenty answering the five questions, five writing hook lines into the bank. The first session takes longer because you are building the set.

If it is taking three hours, you are watching instead of researching. The five questions are a discipline precisely because they stop you enjoying the content.

Doing it manually versus using a tool

Everything above works with a notes app and a spreadsheet, and doing it by hand for a month is genuinely the best way to learn to see structure. Be honest about the failure mode though: the manual version dies in week three, because eyeballing a median across fifteen grids is tedious and the tedium is not where the value is.

What a tool should do is the mechanical part — pulling the accounts, computing each one’s baseline, flagging what beat it, and breaking down the hook and pacing — and leave you the judgement. That is what Runwave is: it watches the accounts you point it at, surfaces the reels beating their own baseline, explains the structure that carried them, and writes the script in your voice from a knowledge base of your own material.

The rest of the cluster

• How to come up with content ideas — nine sources, and why running out is a capture problem rather than a creativity one.
• Competitor content analysis — the baseline method, what to record, and what never to copy.
• Finding trending content before it peaks — the three early signals, and the window where joining a trend is still worth it.
• Content pillars — picking the four things you post about, so research has something to aim at.

Questions people actually ask

What is content research?

Content research is the work of finding out what your audience already responds to, in enough structural detail to build something new on top of it. In practice it means keeping a reference set of accounts one tier above you, finding the posts that beat their own account's baseline, and recording the structure — the opening line, the first cut, where the payoff lands — rather than the subject.

How do I do content research for social media?

Build a set of ten to fifteen accounts one tier above you, work out each one's typical view count, and pick out posts running two to three times above that. For each of those, answer five questions: the first spoken line, what is on screen in frame one, seconds to the first cut, where the payoff lands, and why someone would send it on. Then refill that structure with your own expertise.

How long should content research take each week?

About forty-five minutes once your reference set exists — twenty minutes finding outliers, twenty answering the structural questions, five writing hook lines into an idea bank. If it is taking hours, you are watching content rather than researching it.

Should I copy what my competitors post?

Copy the structure, never the subject. A hook shape, a length, a place where the payoff lands — those transfer and are not anyone's property. The topic, the script and the personality do not transfer and copying them produces a worse version of someone else's video.

Why do big accounts' posts not work for me?

Because a large account's numbers are driven partly by audience size, so its top posts do not isolate what made the video good. Judging a post against its own account's baseline is what separates a genuine outlier from a post that simply had a big audience.

What is the difference between content research and scrolling for inspiration?

Scrolling shows you what an algorithm has decided you personally want to watch, which is close to the opposite of what you need. Research is deliberate: a fixed reference set, a baseline comparison, and a fixed list of structural questions answered in writing.