You post. It gets 20 likes, like the last ten. Then one gets 200 and you have no idea why. We checked 7,887 business accounts to see how often that happens.
Short version: more often than you'd think, usually more than once, and the numbers can't tell you which post is next.
TL;DR
For business accounts whose usual post gets 10 to 39 likes. Their last 12 posts cover about six weeks.
First, the ground rules
For every account we took the last 12 posts (pinned ones don't count) and found the middle one by likes. That's the account's usual. If your usual is 20 likes, it means half your posts got fewer than 20 and half got more.
A hit is a post that got a set number of times its usual. 3× on a 20-like account is 60 likes. 5× is 100. 10× is 200. We picked those thresholds ourselves, so don't treat any single one as sacred. The shape across all of them is the point.
We start with accounts whose usual is 10 to 39 likes because that's what “I get about 20 likes” looks like in numbers. There are 1,301 of them. Every other like range is further down the page.
How often it happens
Only 12% of these accounts never got a post to even double their usual. The rest did, and plenty went much further. Nearly 3 in 10 had a post at 10× within about six weeks.
Business accounts whose usual post gets 10 to 39 likes. n=1,301. Last 12 posts each, about six weeks.
Read it like this: for an account that usually gets 20 likes, 5× is a post at 100 or more. About half of them had one.
Two things worth saying plainly. First, the typical best post out of 12 was 4.7× the usual. Not 50×. If you were picturing viral, this is the boring middle: a post that does 100 when you normally do 20.
Second, 12 posts is a small sample, and the best of any 12 will beat the middle one even when nothing special happened. So some of this is plain variation. That's why the next chart matters more than this one.
The part we didn't expect
The story most owners tell is that they had one post that popped and the rest are normal. We looked for that pattern: one post way up, everything else flat. Strictly, with a 5× post and nothing else above 3×, it was 10% of accounts. Among those that had a 5× post at all, 79% had more than one post above 3×.
n=629 accounts, out of the 1,301.
21%
Just one
26%
Two
20%
Three
33%
Four or more
So the real grid isn't a row of 20s with one bright tile. It's mostly 20s, with a couple of hits mixed in, and the owner can't pick them in advance.
That matters for how you read your own account. A single fluke would be a reason to shrug. Hits that keep turning up are more like a sign that some posts get people to react and most don't. We can't prove that from likes. That's hard to square with “it's all random,” though it doesn't rule out a big shared cause like a seasonal moment.
Does it depend on size?
An obvious objection: small accounts have small numbers, so a few extra likes look like a huge jump. We split by usual likes to check.
| Usual likes per post | Accounts | Median followers | Best post 3×+ | 5×+ | 10×+ | Typical best vs usual |
|---|---|---|---|---|---|---|
| Under 10 | 1,011 | 6,623 | 71% | 61% | 51% | 10.7× |
| 10 to 39 | 1,301 | 4,280 | 69% | 48% | 28% | 4.7× |
| 40 to 99 | 1,400 | 9,655 | 69% | 48% | 29% | 4.7× |
| 100 to 499 | 2,065 | 42,617 | 78% | 60% | 35% | 6.6× |
| 500 and up | 2,107 | 372,946 | 80% | 57% | 30% | 5.9× |
Between 48% and 61% of accounts hit 5× in every group, and 28% to 35% hit 10× in every group from 10 likes up. The 10-to-39 row, the one we used for the headline numbers, is the shaded one.
One honest caveat: the “under 10” row looks wilder (51% at 10×) mostly because of small-number math. Going from 5 likes to 50 is a 10× jump that takes very little. We wouldn't read much into that row.
What the hits look like
For accounts with a 5× post, we checked what kind of post it was. Video made up 41% of everything they posted but 56% of their best posts. Photos went the other way: 55% of posts, 40% of the best ones.
Accounts with a 5× post, 10-to-39-likes group.
Video (Reels)
Photo
Carousel
That's a lean, not a law. Four in ten of the best posts were still photos. Don't read this as “stop posting photos.” Read it as “if you never post video, you're skipping the format that produced more than its share.” We go deeper on this in our Reels vs photos study.
So, why?
Anyone who gives you a clean formula for a high-performing post is guessing, including us. A like count doesn't say whether people liked the hook, the face, the offer, the timing, or the fact that it made them laugh. Maybe all of it. Maybe none of it, and a bigger account shared it.
Here's what we can say with reasonable confidence. Instagram doesn't pick winners out of the blue. As far as anyone outside the company can tell, it shows a post to a small group, watches what they do, and shows it to more people if they react. So a post that got 5× the usual likes did something to the people who saw it first. They liked it, sent it, saved it, stayed on it. The reach came after that.
The order matters. It starts with people, then the algorithm follows. That means a hit usually has a reason, even when you can't name it, and it's more likely to be about the idea than about the posting button.
Our data adds one more piece. If hits come one at a time, once a year, you could call each one a fluke. They don't. The typical account here had several in six weeks. Calling that luck is a comfortable story and not much more.
What to do with this
If half of businesses like yours land a 5× post every six weeks, yours isn't a one-off. Go back and look at it. Look at the second best one too. What do they have in common that the other ten don't?
One hit tells you almost nothing, because the best of 12 beats the middle one by chance alone. Two or three that share something is a pattern. Count hits across a quarter, not a week.
Your own 12 posts are a tiny sample. The businesses competing for your customers have hits too, and every one of those is a free data point about what your shared customers react to. You'll spot patterns across 50 of theirs faster than across 12 of yours.
If you want a shortcut for number 3
Give it your Instagram handle. It finds the businesses competing for your customers, looks at their posts, and gives you content ideas built on concepts that pulled in a lot of engagement for them.
It doesn't write your captions, schedule anything, or run your account. You still decide what's worth posting and you still have to make it. It also can't promise a post will hit, because as you just read, nobody can. What it does is point you at ideas with proof behind them instead of a blank page.
First month free, then $24.99/month.
Methodology
Sample
7,887 public Instagram accounts flagged as business accounts, each with at least 9 of their latest posts stored (12 for most). Pinned posts were dropped. We removed accounts whose usual post had zero likes, since a ratio to zero means nothing. The headline group is the 1,301 accounts whose usual post gets 10 to 39 likes.
Usual and hit
Usual is the median likes across the account's stored posts. A hit at N× is a post with at least N times that. Median instead of average so one big post doesn't drag the baseline up.
Time window
The 12 posts cover a median of about six weeks (44 days in the headline group, 37 across all business accounts). Accounts that post more often cover less time. We didn't normalize for that.
It's not a random sample
These are accounts that turn up in niches people research with our tools. That skews toward businesses that are at least doing okay, and probably away from ones barely posting. They are not our customers' accounts and they are not every business on Instagram. Expect an average small business to sit a bit lower.
Best-of-12 effect
The best of 12 posts will beat the median even with no real difference between posts. That inflates every number here a little, and we haven't quantified by how much. The “more than one hit” result is less affected, but it isn't immune either.
Likes still arriving
Recent posts hadn't finished collecting likes when we read them. That slightly lowers the usual and slightly raises the ratios.
What we can't see
Reach, saves, shares, watch time, paid promotion, or who shared the post. A post could have been boosted or picked up by a bigger account and we wouldn't know. This is observational data: it shows how often big gaps appear, not what caused them.
Questions people ask