Start 7-day trial
9 min read · ·

One video went viral and the rest did not: how to find out why and repeat it

A single hit is the most confusing result a creator can get. It proves you can do it and gives no instructions. Here is how to pull the repeatable part out of the lucky part.

Short answer. To find out why one video went viral, compare it with your normal videos on five things: the topic, the opening, the format, the length and who shared it. Then check for outside causes, such as a large account sharing it. Whatever differs and is under your control is your hypothesis. Test it with three new videos that change only that one thing. Part of any hit is luck, so aim to repeat the conditions and not the exact number.

Start by accepting two things

First, some of it was luck. Timing, an early share or a viewer with a big following can all multiply a result, and you cannot control those. Second, some of it was not luck. Your own video beat your own normal by a wide margin, and something you did contributed to that.

Creators tend to go wrong in one of two ways. Some decide the hit was all skill and keep repeating surface details, like the same shirt or the same song. Others decide it was all luck and learn nothing. The job is to separate the two.

Step 1: Measure the size of the outlier

Work out your median, as described in how to find outlier posts. Use only posts from before the hit, because the hit may have changed your audience. Then divide. A video at 250,000 against a median of 1,000 is 250 times normal, which is a very different event from a 5× post. Huge multiples usually mean the video escaped your audience and reached a much wider one. That is exactly the kind where outside causes deserve a check.

Step 2: Find out how people got there

Before you study the video, study the traffic. This step stops you from drawing the wrong lesson.

If the cause was mostly outside your control, note it and do not over-learn. If the post spread through the normal recommendation surfaces, there is more to study.

Step 3: Compare it with your normal posts

Pick three of your typical videos and put them beside the hit. Use this table and fill it in honestly. Then circle what is different.

ElementWhat to write downCommon difference in hits
TopicWhat it is about in five wordsBroader, or closer to a shared problem
OpeningFirst line and first frameClear subject and tension at once
FormatTalking head, list, story, demoA format you rarely use
LengthSecondsOften tighter, sometimes much longer
PayoffWhat the viewer gets at the endA clear ending worth waiting for
EmotionWhat it makes people feelStronger than usual
ShareabilityWho you would send it toA specific kind of friend

The third column is what we commonly notice when reading outliers, and it is a prompt and not a rule. Your hit may differ in another way.

Step 4: Look at the first viewers' behaviour

If the platform shows a retention graph, compare the hit with a normal post. Three shapes are useful.

TikTok's 2020 explanation, as reported, listed finishing a video and sharing it among strong signals. Coverage of Adam Mosseri's comments lists watch time and sends for Reels. So it is reasonable to treat attention and sharing as the two things your hit did better. Neither platform publishes weights, so this is an inference, not a formula.

Step 5: Write one hypothesis

Now write one sentence: "This worked because..." Make it specific enough to be wrong. "It was relatable" cannot be tested. "It named a mistake beginners make in the first line" can be.

Good hypotheses describe a decision you can make again. Here are some examples of the shape to aim for.

Step 6: Test with three posts

Make three new videos that keep the suspected cause and change the topic. If you make the same video again, you are testing whether people want a repeat, which is a different question. If you keep the mechanism on new topics, you are testing whether the mechanism works.

Do not judge these against the hit. Judge them against your old median. If they land at 3× or 5× your old median, you have found something real, even though they are nowhere near the viral number. That is the realistic win.

If all three fall flat, go back to your table and try the second difference. Most creators need a couple of rounds.

Should you make a part two?

Often yes, and quickly. A direct follow-up is the cheapest test you can run, and people who liked the first one have told you what they want. Keep it honest: answer a question from the comments, show what happened next, or cover the point people argued about. Pin a comment on the original that points to it. Do not expect it to match the original. Treat it as a way to learn which part people wanted more of.

What to do about your new audience

A viral post brings followers who came for that one thing. If your next posts have nothing to do with it, those people will not watch, and your numbers can look worse than before. Decide on purpose whether the hit is a direction you want. If it is, make more. If it is not, accept that those followers may stay quiet, and keep measuring against your old median.

A worked example of a good hypothesis

This example is made up. A home cook usually gets about 3,000 views. One reel, "The mistake that makes your rice stick," reaches 180,000. The other posts in that month were recipes.

A weak hypothesis would be: "People like rice." It gives you nothing to test. A better one names what was different and what you expect to see:

"This post beat my median because it opened on a mistake the viewer might be making, not on a recipe. If that is right, three posts about other common mistakes should beat my median, and three new recipes should not."

That sentence is useful because it can fail. If the mistake posts do no better than the recipes, the opening was not the reason, and you have saved yourself months of chasing it. Now try the same thing with your own hit, and fill in the blanks:

"This post beat my median because ___. If that is right, posts that ___ should beat my median, and posts that ___ should not."

Reasons a video can go viral that you cannot repeat

Be honest about these. Some hits had help you will not get twice, and it is better to know that before you plan around them.

None of this means you learned nothing. It means you should give that hit less weight. A post with an obvious outside cause is weak evidence, and two smaller outliers with something in common are better evidence than one enormous fluke.

What to do in the 48 hours after a video takes off

You will not get many moments when strangers are paying this much attention. Use them, but do not panic-post.

  1. Reply to the early comments. Questions there are free ideas for your next post.
  2. Pin a comment that says what the account is about and what to watch next.
  3. Check your bio. New visitors are deciding in seconds, so see the profile checks in this guide to views without followers.
  4. Save the numbers now: views, shares, follows, and where the views came from. You will want them later.
  5. Write down what you remember about making it. Why you picked it, what you changed, how long it took. That detail fades quickly.
  6. Plan your next post around the hypothesis. Do not rush out something unrelated.

Why the next post nearly always looks worse

Expect a drop after a hit. An outlier is by definition far above what you normally get, so the next post will almost certainly be closer to your median. Statisticians call this regression to the mean. It does not mean the platform turned on you. It means the last post was unusual.

This matters because many creators read that drop as a punishment and change everything. One wrote: "Have multiple reels with 100k+ consistently, then last week I started pulling 1k reels." A fall that large and that lasting is worth investigating, so compare the posts and check Account Status. A normal step back after one hit needs no action. Compare your next five posts with your old median, not with the hit. If your median has risen at all, you gained something real.

Do it on someone else's hit too

This method works on any account. You often learn faster from other people's outliers because you have no attachment to them. Pick an account your size, find a post at 10× or more, and run the same steps. XCut AI does much of this work: paste the link, and it reads the video itself and lays out the hook, angle, structure, pacing and call to action, so you can compare several outliers side by side. The script analyzer is a free way to check a transcript if you only want the structure.

If you want a shortlist of tools for this job, see the best tools to find viral reels.

Frequently asked questions

Why did my video go viral and then stop?

In our view, recommendations keep going while new viewers keep responding, then slow when the video runs out of receptive people. That is normal and does not mean you did anything wrong.

Why do my views drop after a viral video?

Often they did not drop. Your baseline looks low beside the hit. Viral posts also bring followers who only wanted that one thing. Compare new posts with your median from before the hit.

Can I repost a viral video?

You can, but TikTok says duplicated content is not recommended on the For You feed. Make a new version with a fresh opening and new footage.

How long until I know the test worked?

Three posts give you a direction, not proof. Six posts with a consistent result is much stronger. Give each post the same settling time before reading it.

Is going viral worth chasing?

A rising median is usually worth more than a single spike, since it means you can repeat what works. Use a hit as a source of information and not as the target.

Should I delete my low posts after a viral video so the profile looks better?

We would not bother. New visitors mostly look at your pinned posts and your first rows. Pin your best work, and leave the rest alone.

Do I need to change my niche to match the video that went viral?

Only if it matches what you want to be known for. New followers came for that topic, so test it with a few posts first and watch whether they stay.

References

See which of your posts beat your average

Paste your handle at xcut.ai/instagram for a free scan with no account, or start a trial to research any niche.

Start your 7-day trial

$5 for 7 days · 1,000 credits · cancel anytime