Should you copy the reference video with the most views?

A brand sells a glass bottle you fill with coffee and water and leave in the fridge overnight. Somebody spent two afternoons searching the app and dropped six links in a thread: six videos, six accounts, none of them the brand's.

Monday there's a meeting about which one the next campaign starts from, and the thread is already sorted. Top of the list, 2.3 million. Bottom, forty thousand.

The sort happened before anyone said what it was for.

What is the count under somebody else's video a fact about?

How far the video got, and nothing past that.

Before you lean on it, read what the platform says the number is. Its advertising documentation defines a video view as a play that started, and keeps the count of unique people who saw an ad at least once in a separate metric. Those definitions are written for ads. How a public post's number is put together is something we couldn't confirm, and the post doesn't tell you either. So you can't read the number on your six links as people, and nothing on the post says what it does count.

Our guide to telling whether a creator video worked is about a video you commissioned, and its wording for the count — "That number counts how many people the video got in front of. It's a fact about the video's audience" — says people where the advertising documentation says plays. What holds either way is the half that matters on Monday: the count is about how far the video traveled, not about what landed. That guide settles the rest differently: "The answer is in your own notes, or it doesn't exist." The notes are a sentence written before the video posted and a list of what would show if the sentence came true.

None of that exists here. You didn't hire the bottle videos, you wrote no sentence for them, and whoever did isn't going to send it to you. The count on a stranger's post is a number with nothing on your side to check it against.

What doesn't the post say about where that count came from?

Which of the things it never splits apart sent it.

A post can travel because of the way it's built, because of the account it was posted from, or because somebody paid to push it. The platform sells that last one: a feature that turns an existing post into a paid ad. One number covers whatever combination you happen to be looking at, and there's no breakdown under it.

Nothing here says any of the six was boosted. The point is narrower and worse: you can't rule it out from the post, so every count on your shortlist has a budget you can't see among its explanations.

The account is the explanation that costs you the most, because it's the one you can't take with you. If the 2.3 million came partly from an account 900,000 people already follow, then part of what carried it is the part you can't copy, and nothing on the post tells you how much of the rest the structure did. That is what the first reading is for.

Why doesn't "how many views is viral" settle the shortlist?

Because it's a question about one post, and you have six.

There's no shortage of answers to it. A search for one in the US in October 2026 turns up pages titled what is a good view rate for tiktok, How many views on TikTok is good?, and How many views is viral on TikTok?, most of them published by tool vendors, and the figures they hand back don't line up — one treats a few hundred views on a small account as a good result while another puts virality in the hundreds of thousands. You can't clear a line that moves depending on whichever article you opened.

Our guide to comparing creators by engagement rate is answering a question about a percentage rather than a view count, and what it says about the figure holds for both: "A threshold also has to be a threshold for something. A quoted average needs the set of accounts it was computed over, because that set is what decides what it is an average of."

And even a threshold with its set printed next to it would answer the wrong question. It would tell you whether one post did well for an account of that size. Your six candidates sit on six accounts of different sizes, and what you need is an order among them.

Does the structure show up on accounts that have nothing to do with each other?

That's the first of the two readings this page proposes, and it runs across the search rather than inside one profile.

What you can take from somebody else's video is the structure, not the execution — our guide to joining a TikTok trend names the layer that holds it: "Strictly, you don't join a trend. You join the meme format underneath it."

So go back to the search that found the video and look for the structure rather than the post: the parts that, as that guide puts it, would stop people recognizing it if you changed them. Same opening move, same order, same thing held back until the end. Then count the accounts posting it that have nothing to do with each other: unrelated subjects, no shared sound or hashtag forcing them into the same result.

Some of what holds a structure together isn't visible from outside, and the same guide's answer to that is to ask somebody who posts in it. What you can count from outside is the part that repeats where you can see it.

When unrelated accounts have each chosen to build videos this way, what you're looking at is more than one account's taste. When the whole thing lives on one account, what you've found is one account.

For the bottle videos: the one at the top of the thread is a morning routine that gets interrupted by something in the fridge, and two more links in the thread are versions of that same structure. The search turns it up again on a protein-powder account, a meal-prep account and an account that only posts dogs. The one at the bottom, forty thousand, is a tour of somebody's kitchen shelf, and nothing like it appears anywhere else. The last two links are one-offs the same way. So the first reading leaves one candidate standing and three links on a watch list.

One competitor's creator training on how to use reference videos sends creators down this road and then further: compile the posts that did well by searching the niche, and because you know they're validated content formats, "your first several videos for this new client should be copying all of the most successful concepts." Read the rest of the lesson and the certainty drains out of it — the same training treats validated formats as the anchor rather than the result, and puts the real upside in how hard you test variations around them, and its arithmetic assumes a creator posting daily, with 21 to 24 posts in a month to spend. A brand picking one reference for one campaign isn't running that many tests, so there is no testing your way to the answer. What recurrence can do for one shot is narrow which structures are candidates at all.

Does what you sell have a place in the candidate?

Run this one next, because it costs nothing and it can settle a candidate by itself.

This judgment belongs to our guide to joining a TikTok trend, and the order doesn't make it. That guide runs it on the open place in the structure and names what it needs: "what the open place is for, and a plain sentence about what you sell. No numbers, no budget, and nothing you'd have had to commission." Our guide to choosing creators is where the ordering rule comes from, and it prices a check on both halves at once: "What matters is that you never spend an afternoon on a candidate a one-minute check would have cut." The afternoon here is opening profile after profile across every version of every structure you kept. A plain sentence about a cold-brew bottle is the minute.

For the bottle videos: the interrupted-morning structure pays off on whatever is in the fridge, and a bottle of cold brew somebody made last night is a thing that can be in a fridge. The candidate clears, and the campaign has a structure to start from before anyone opens a second profile.

A candidate that doesn't clear is out whatever its counts say, and that is the whole reason this check goes early. If the open place in a recurring structure had no room for a bottle somebody fills the night before, the 2.3 million would have bought it nothing, and the next candidate on the watch list would be the one to check.

Which posting of the candidate is worth opening?

The second reading is a minute of scrolling, and it runs inside one profile.

Open the profile each version of the structure sits on and read the counts down the recent posts. You're looking for whether that post sits far from the rest of them.

Our guide to comparing creators by engagement rate already runs one measure down a single profile and says what the run buys: "A post sitting well away from the rest of that run is the one to watch." The same guide says why a single measure is the readable case: "A run of one measure is the easier case and it arrives in the same place." There it points you at a video to open while you're still choosing somebody to hire. Here it points you at which posting of a structure you've already picked is the one to study.

A post sitting among its neighbors leaves the account as the explanation you can't rule out. A post sitting far above them is a posting where something went differently from that account's usual week, and what it earns is your attention when you watch the structure back, not a verdict about your campaign.

For the bottle videos: the 2.3 million post sits inside a run of posts in the same range, which is what that account gets every week. The meal-prep account's version is several times anything around it. So the version to watch frame by frame is the meal-prep one, and the 2.3 million link — the one the thread was sorted by — is the version to watch last.

What does this hand you, and what does it not?

A structure to start from, a posting to study, and a watch list. Where the campaign ends is a different question.

The interrupted-morning structure is what the campaign starts from. The posting to study is the meal-prep account's version, which the search turned up rather than the thread. The three thread links that carry that structure stay in play as versions of it, and the 2.3 million one has the least to tell you, because its account gets that range every week. The three one-offs stay on the watch list until the search shows somebody else trying them. None of it used the number the thread was sorted by.

Handing the structure to a creator is a separate job again, and our guide to sending a creator a reference video is where that one is answered.

Be plain with your team about what each part of this is. The two readings are observations you can make on public screens without asking anyone for anything, the definitions in the first section are published by the platform, and ordering a shortlist this way is a proposal. Nobody has measured whether references picked this way produce better campaigns, and this page isn't that measurement. The first reading also depends on the search you ran, which is why the last step is a note rather than a conclusion.

Two things go in the thread next to each link, because neither one is still there when somebody opens it next week. The date you read the counts, since they keep moving. And what each reading said, in a line: which unrelated accounts you found the structure on, and where the post sat against its neighbors.

Frequently asked questions

Can't a tool just hand us the top-performing videos in our category?

It can hand you a sorted list, and the sort is somebody's decision. Ask what it's ordered by before you use it. If the answer is the counts on the posts, you have a faster version of the thread you started with, including every count that a budget or a large account explains.

One link's structure appears on no other account, but that post ran far past its own recent ones. Is it out?

It isn't what you start from, and it isn't nothing either.

The first reading is what makes a structure a candidate, and this one didn't. What you have instead is one account's result, and the two explanations for it — a structure that travels, and a structure that works because of the person in it — don't look different on the page. The thing that would separate them is another account trying it, which is why the link goes on the watch list and the search is worth running again. What doesn't change is the free check: the open place either has room for what you sell or it doesn't, and that was never a question about how far anything got.

See what this looks like for your brand

Popow suggests formats and references based on current trends and data.

Or email us: business@popow.ai