Tony Wang8 min readThe Median TikTok Creator Has 12,000 Followers. That's Our Code, Not TikTok.
Our TikTok index holds 80x more creators just above 1,000 followers than just below. That cliff is a constant in our own ingest code, not a fact about TikTok.
Our TikTok Creators Index covers 6,133,853 accounts. Its median creator has 12,000 followers. Two months ago the same index covered 3,485,257 accounts and its median creator had 29,000 followers.
TikTok did not change. We crawled more.
That is the whole problem with a "typical creator" number sourced from a discovery-built dataset, and it is worth showing precisely rather than hand-waving at it — because in our case the distortion is not subtle, not statistical, and not really about TikTok at all. It is a constant in a Go file.
There are 80 times more creators just above 1,000 followers than just below
Slice the index finely around 1,000 followers and the distribution stops behaving like a distribution.
| Follower band | Creators in band | Creators per follower of width |
|---|---|---|
| 900–949 | 1,148 | 23 |
| 950–999 | 1,305 | 26 |
| exactly 1,000 | 2,660 | — |
| 1,001–1,049 | 102,907 | 2,100 |
| 1,050–1,099 | 92,566 | 1,851 |
| 1,100–1,199 | 158,977 | 1,590 |
Twenty-six creators per follower-unit on one side of the line, 2,100 on the other. Above 1,000 the curve behaves itself — a smooth power-law decay, each band a little thinner than the last, all the way up. Below 1,000 it simply stops.
This is not a rounding artifact. Rounded follower counts would pile up on round numbers, and they do not: the bands decay monotonically through 1,100, 1,200, 1,300 with no clustering. It is not attrition either — deleted and private accounts are excluded from these counts, and the under-10,000 band has the lowest disappearance rate in the whole index.
One honest caveat about the small population that does sit below the line: 35,422 of those 170,908 accounts have a follower count of exactly zero, which in practice means a crawl that failed rather than a creator with no audience. The genuine 1-to-999 population is about 135,000 accounts, or 2.2% of the index.
The cliff is six lines of Go
Here is the entire explanation, from our own ingest worker:
// defaultMinSeedFollowers is the quality floor — keyword/video search returns plenty
// of tiny/junk accounts alongside real creators. 1k keeps nano-influencers (real,
// marketable) while dropping the obvious noise; query-time min_followers filters
// higher.
const defaultMinSeedFollowers = 1000
Every creator a discovery vein finds passes through one filter before being indexed, and that filter discards anything under 1,000 followers. The comment is not wrong — keyword search really does return a lot of junk, and 1,000 really is a reasonable place to draw the line for a dataset meant to support brand and agency work. It is a defensible product decision.
It is just not a measurement. The sharpest-looking feature in our entire follower distribution is a number somebody typed.
The residue below the line has a mundane explanation too: it is mostly a bulk import that predates the discovery veins and therefore never passed through this filter, plus accounts that were above 1,000 when we indexed them and have shrunk since.
The floor sets the floor. Crawl depth does the rest.
Neither half of the "median creator" number is a property of TikTok. The floor decides where the distribution starts; how hard we crawl decides how much mass piles up on top of it. Watch both move in eight weeks:
| Follower band | 2026-06-29 (3.49M) | 2026-08-29 (6.13M) |
|---|---|---|
| 10M+ | 0.06% | 0.03% |
| 1M–10M | 2.03% | 1.32% |
| 100K–1M | 19.22% | 12.58% |
| 10K–100K | 61.55% | 40.34% |
| 1K–10K | 12.09% | 42.94% |
| Under 1K | 5.06% | 2.79% |
The 1K–10K band grew from 421,260 creators to 2,634,089, which is 84% of everything the index added. The modal band moved down a rung, the median fell by more than half, and the share under 1,000 followers also fell — from 5.06% to 2.79% — because the new arrivals all landed above the floor while the pre-floor residue stayed flat.
So in eight weeks the index got closer to the platform's real shape in the middle and further from it at the bottom, and the headline "typical creator" number dropped 60%. Nothing about TikTok moved.
What a representative measurement actually says
The right external comparison is Pew Research Center's account-level work, and the first thing worth knowing is what it does not offer. Pew's February 2025 analysis of creator posting behavior defines creator accounts as "those with at least 5,000 followers." There is no sub-1,000 figure to compare against, because that population is excluded from the analysis by construction — the same move we make at 1,000, made at 5,000, and stated openly.
What Pew does provide is an account-level distribution. Their October 2024 study surveyed a nationally representative group of 664 US adult TikTok users and catalogued roughly 250,000 accounts those users follow. About four-in-ten of those followed accounts have fewer than 5,000 followers, and 6% have a million or more.
share with under 5,000 followers
share with under 5,000 followers
Thirty-eight percent against 35.8% looks like validation. It is not, and this is the part worth sitting with.
Pew describes their sub-5,000 group as small personal accounts: more than half have posted fewer than twenty videos in the account's entire life, and a quarter have never posted at all. Ours is the opposite kind of population — a wall of accounts stacked between 1,000 and 5,000 followers, every one of which cleared a filter designed to keep only accounts worth marketing to, with near-nothing underneath. Two datasets, one nearly identical headline share, and almost no overlap in what the share is made of.
The divergence runs the other way at the top. Six percent of the accounts Americans follow have a million-plus followers; only 1.35% of our index does. A following list is popularity-weighted by definition — people follow famous accounts — while an index is not. Neither number is the platform.
What to ask a creator index
Every dataset built by discovery has a line like ours somewhere. Crawl-based, search-based, seed-list-based — it does not matter; something decides what is too small to bother with, and that decision propagates into every summary statistic the dataset ever produces. The same skew shows up in our X dataset, for the same structural reason.
So the useful question is not "what is the median creator's follower count." It is:
- What is the ingest floor, and is it published? Ours is 1,000, it now reads from an environment variable rather than a recompile, and this post is where it gets stated.
- How much of the distribution sits in the first band above that floor? For us, 42.9% — which means the modal creator in our index is, structurally, an account that just barely qualified to be in it.
- Has the dataset's size changed recently? If it has, its median has moved, and any budget or benchmark derived from the old median has silently gone stale.
We publish the index because the discoverable layer is genuinely the useful one for most work — you cannot run an outreach campaign against creators no search surface will ever return. But "the median TikTok creator has 12,000 followers" is a sentence about our crawl, not about TikTok, and it should be read that way every time.
Query the TikTok creator dataset directly
6.13M creator profiles, follower bands, verification and contact-presence rates — as structured JSON, with the ingest floor stated up front.
Related reading
Frequently asked questions
What's a typical TikTok creator's follower count?
It depends almost entirely on how the dataset was built. Our discovery-based index of 6.13M TikTok accounts puts the median at 12,000 followers — but two months earlier, at 3.49M accounts, the same index put it at 29,000. TikTok did not change; we crawled more. Any median from a discovery-built dataset is a statement about the crawl, not the platform.
Why does the dataset have so few creators under 1,000 followers?
Because our ingest code drops them. Discovery veins pass every account through a quality floor of 1,000 followers before indexing, so creator density is roughly 80x higher just above the line (2,100 accounts per follower-unit between 1,001 and 1,049) than just below it (26 between 950 and 999). The small population that does sit below the line arrived through a bulk import that predates the filter, or has shrunk since being indexed.
What does Pew Research say about TikTok creator follower counts?
Pew's February 2025 analysis defines creator accounts as those with at least 5,000 followers, so it reports no sub-1,000 figure at all. Their October 2024 account-level study of roughly 250,000 accounts followed by 664 US adults found about four-in-ten have under 5,000 followers and 6% have a million or more. Our index is at 35.8% under 5,000 and 1.35% over a million.
What should brands and agencies do with this?
Ask any creator index for its ingest floor before using its median follower count to size a campaign budget. Ours is 1,000, configurable at runtime, and 42.9% of the index sits in the single band directly above it — meaning the modal creator in the dataset is an account that just barely qualified to be in it.