A power law shows up whenever a few things absorb almost everything and most things get almost nothing, and that shape keeps repeating no matter how far you zoom in. City populations look this way, a handful of megacities hold a huge share of humanity while a long tail of small towns barely register. Earthquakes work the same way, most tremors are nothing and the rare big one releases more energy than the rest of the year combined. The familiar name for this shape is the 80/20 rule. An Italian economist named Vilfredo Pareto noticed in 1896 that about 80% of the land in Italy belonged to roughly 20% of the people, and versions of that same lopsided split, sometimes 70/30, sometimes 90/10, kept turning up everywhere he looked afterward.
There's a second shape that gets mistaken for the first constantly, called log-normal, and it comes from a genuinely different process. Not a few things reinforcing their own advantage, just ordinary growth that happens by percentage instead of by a fixed amount. If a pile of money grows by some random percent a year, up some years, down others, the spread of outcomes across a lot of people doing that ends up badly skewed too, a handful well out ahead, most people bunched lower down, without any single account actively feeding off its own size the way a true power law implies. Over the range most people actually look at, it produces almost the same lopsided picture as a real power law, which is exactly the trap. A true power law never really tapers off, the richest sliver keeps pulling further away as the group gets bigger. A log-normal shape is still unequal, sometimes wildly so, but it settles down out at the tail. Calling something an "80/20 situation" usually means assuming the first kind without checking, when a lot of real, heavily skewed data turns out to be the tamer impostor instead.
That distinction gave me two separate things to check on a platform people already treat as obviously power-law-shaped: whether followers themselves pile up on TikTok in that self-reinforcing way, and whether an account's size predicts how its audience behaves in that same fixed-ratio shape. The second question turned out to have a much cleaner answer, so I'll start there.
If you're deciding whether to put an influencer budget behind one large account or several small ones, the pitch decks already claim an answer: small accounts engage their following better than large ones. I checked, and it holds up, more precisely than the slogan suggests, on a sample of a thousand TikTok profiles. A thousand is enough to see the pattern clearly, not enough to trust blindly at the extremes, and this sample runs out well short of a million followers, so I'd rather flag that now than let it surface later.
Here's the actual shape of it: every time follower count multiplies by ten, engagement rate drops to about a third of what it was, and that ratio holds across the whole range of the sample, not just at one lucky slice of it. It's the exact kind of fixed-ratio decline that makes something a power law rather than just a fuzzy downward trend, and it's a real enough gap to change how a budget gets split, not just decorate a slide.
The straight line is a simplification, so I checked it against a flexible trend line that doesn't assume any particular shape. The two agree almost exactly through the middle of the range, and only pull apart at the two ends, where the smallest and largest accounts each keep a little more engagement than a single exponent predicts. Both of those ends are also the thinnest parts of the sample, the same accounts already too few or too noisy to trust on their own, so I'm reporting the straight-line number as the honest summary of the bulk of the data, not a claim that one exponent describes every account down to the last few dozen at either edge.
Some of this is probably just what a small audience is: more niche, more likely to actually know who the creator is, small enough that a comment might get an actual reply instead of disappearing into a feed. Part of it is arithmetic. Engagement rate is a share, and the denominator gets harder to clear as it grows, since a following that size has had time to accumulate people who followed on a whim, drifted off, or never fit the content in the first place, all of whom still count against the rate even though they were never going to like anything.
None of that means a small account generates more total engagement than a large one, and it's worth being precise about that before anyone reads this as "go small." A creator with a million followers and a 1% engagement rate is still pulling in ten thousand likes and comments, against roughly two hundred and fifty for a creator with five thousand followers at a 5% rate. The more accurate version of the claim isn't that small accounts perform better, it's that they engage their audience more deeply while large accounts deliver more total reach, and which of those a budget should chase depends on whether the goal is depth or scale.
Two things about the headline number are worth naming before I lean on it further. Engagement rate is a share, and shares built on tiny numbers misbehave: an account with three followers and one enthusiastic friend can post a 400% engagement rate, since a single like can outnumber the whole audience. Nearly a tenth of the sample sits above 100% engagement, almost all of it concentrated in the smallest accounts. I redid the comparison excluding anyone under 100 followers, just to make sure that noise wasn't the whole story. The tenfold-to-a-third pattern got a little weaker, but it held. The second issue is the sample itself: the scraping company doesn't say how these thousand profiles were chosen, and if they leaned toward accounts that were already trending, the overall level could sit higher or lower than the truth even if the underlying pattern survives.
A single sample of a thousand can't rule out a biased scrape on its own, so I went and found two more: a public mirror of the thousand largest Instagram accounts on earth, collected independently in 2022 by someone with no connection to the TikTok data, and a separate free sample of a thousand Twitter/X posts with followers ranging from 34 to over seven million. Both told the same story, a steeper decline on Twitter than on TikTok, and steeper still among the Instagram accounts, every one of them already in the millions of followers, meaning the effect doesn't level out as an account gets bigger. It keeps getting worse. Three platforms, two unrelated sources, two different years, and none of them disagreed about the direction.
Which brings the question back to the one I opened with and didn't answer yet: is the concentration of followers on TikTok the self-reinforcing, true 80/20 kind, or the tamer log-normal impostor. Follower counts in this sample are almost absurdly lopsided, by one standard measure of inequality, the same kind used for income, the top 1% of profiles hold more followers between them than the bottom 90% combined. That's consistent with either explanation, so eyeballing it doesn't settle anything. I ran a formal test built specifically to tell the two shapes apart instead of judging it by eye, and it came back close to a coin flip, leaning toward the tamer, log-normal explanation, but not by enough to rule out the self-reinforcing one. I can't confidently call this a real power law, and I can't confidently rule it out either.
So the two questions I opened with land in different places. Exactly which shape describes how followings pile up this unevenly is genuinely unclear on this data, and I'd want a bigger sample before betting on Pareto's version over the tamer one. But the relationship between audience size and how that audience behaves, the thing that actually matters for a budget decision, showed up the same way across three platforms and two sources I didn't collect myself.
Data: Bright Data TikTok profile sample (n=1,000, followers 1-812,300, log-log OLS with HC1 standard errors, sampling method undisclosed); Bright Data Twitter/X post sample (n=1,000, followers 34-7.1M); niteshkuwarbi/instagram-data-analysis top-1,000 global Instagram mirror (n=996, followers 2.8M-469.6M, scraped 2022, independent of Bright Data).