Average TikTok views vs the median: how to read a creator's numbers

A creator’s media kit usually hands you three numbers: average views, a follower count, and a few recent posts. All three are useful. Each one is also off in a predictable direction. The average sits well above what a typical post does. The follower count suggests reach that large accounts do not get. And the posts in a campaign report are often read while they are still collecting a large share of their views.

We measured how far off each one runs, using profile snapshots of creators we evaluated for client briefs on TikTok and YouTube, with a relative check on Instagram. Most of those briefs were for AI apps, so treat the figures as a benchmark from that pool. They are not a census of either platform. The methodology note at the end says exactly what is in it. None of these figures are UniSong campaign results. They describe creators’ own posts.

The short version:

• For the typical TikTok creator, the average views of their last 10 videos (each four weeks to six months old) run about 2x their median video.
• Forecast a single post from the median. For a four-post package the average is the better guide, but on TikTok it still runs high.
• Accounts with 10x the followers typically see about 3–5x the views on a typical post (median of their last 10 videos, each four weeks to six months old, on TikTok and YouTube Shorts), so views per follower fall as accounts grow.
• On TikTok and Shorts, likes per view hold up on bigger accounts. Comments per view thin out.
• Don’t read TikTok views at day 7. Read them no earlier than day 30, and at day 45 to be safe.

Average TikTok views are not the typical post

Media kits and creator tools usually quote average views. The trouble is that a creator’s views are not spread evenly across their posts. A few videos break out and most do not, and the average absorbs the breakouts.

For the typical TikTok creator, the average views of their last 10 videos (each four weeks to six months old) run about 2x their median video (2.0–2.3x). On YouTube Shorts the gap is about 1.5x (1.4–1.7x), and on long-form YouTube about 1.3x. The spread between creators is wide: for the middle half of TikTok creators, the average of their last 10 videos (each four weeks to six months old) runs about 1.4x to 4x the median.

Most of that gap comes from a single post. Remove each creator’s single best video from those last 10 and the TikTok gap falls to about 1.5x (YouTube 1.1–1.2x). One video does most of the inflating.

You can see it directly in how views are shared out. For the typical TikTok creator, one video carries 35–40% of the views across their last 10 videos (each four weeks to six months old), where an even spread would give 10%. On YouTube Shorts the best video carries about 30%, and on long-form about 25%.

So most videos fall short of their creator’s own average. About 30% of a TikTok creator’s last 10 videos (each four weeks to six months old) beat it. On YouTube it is about 1 in 3. If a brand books a creator expecting the average, most posts will come in under it.

The window matters too. Measure over more videos and the gap grows: on TikTok, the average runs about 2.4–2.5x the median over the last 15 videos and 3x over the last 20 (each four weeks to six months old). A longer window has more chances to include a breakout. An average quoted without its window can’t be compared with another creator’s average.

How many views is good (or viral) on TikTok?

A post above the creator’s own median of their last 10 videos is a good post for them. For a size check, multiply their followers by the tier range in the views-per-follower table below. A post well above both is a breakout for that creator, whatever the platform-wide numbers say, and a viral post is the kind of breakout that already inflates a creator’s average. For absolute view levels across TikTok videos, see the view distribution in our TikTok engagement rate calculator.

Average views vs median views: forecasting one post or a package

If the average overstates the typical post, should you forecast from the median instead? For one post, yes. For a package, not quite.

We tested this by forecasting each creator’s 4 newest videos (each four weeks to six months old) from the 8 before them, using either the median or the average of those 8.

For a single post, start from the median. On TikTok, a forecast from the median is typically off by about ×2.1–2.3 (a post at half or double the forecast both count as a ×2 miss), and a forecast from the average by about ×3. Half of next posts miss by more, so treat any single-post forecast as a wide range. The median gave the smaller miss for about 2 in 3 TikTok creators. On Shorts the edge is small (×1.9 vs ×2.3; the median closer for more than half of creators). On long-form YouTube neither summary wins.

The average also sets a bar that most next posts don’t clear. Only about a quarter to a third of a TikTok creator’s next posts reached their previous average, against close to half for the median.

For a package of four posts, the average is the better guide, but on TikTok it still runs high. The typical creator’s next four posts averaged about three-quarters (0.7–0.8x) of their previous average and about 1.5x their previous median. On YouTube the average held up better (about 0.9x).

The two results fit together. A four-post package has a fair chance of catching a breakout, which lifts its total above four median posts. It is less likely to catch one as big as the breakout already sitting in the creator’s average, so on TikTok it lands below the average.

In practice:

• One post: forecast from the median, and expect real misses in both directions.
• A TikTok package: start from the average, then plan for delivery at about three-quarters of it.
• A YouTube package: the average is a reasonable starting point.

Our influencer rate card benchmarks estimate implied CPM from recent average views. That is a reasonable starting point for a YouTube package. For a TikTok package, discount the average by about a quarter first, and for a single post run the same sum on the median, so the brand sees the range it is buying.

Views per follower: bigger accounts get fewer views per follower

Follower count is the first filter most brands use, and the tier labels on a rate card are built on it. But views don’t scale with followers one-for-one.

Accounts with 10x the followers typically see about 3–5x the views on a typical post (the median of their last 10 videos, each four weeks to six months old), on TikTok and YouTube Shorts. Long-form YouTube points the same way. So views per follower fall as accounts grow. Views include people who don’t follow the account, so treat views per follower as a ratio, not the share of followers who saw the post.

Here is the TikTok picture by rate-card tier. The first column is the typical creator’s typical post, shown as a range because it moved when we cut the data different ways. The second shows the middle half of creators in the tier.

TikTok followers	Views per follower, typical post	Middle half of creators
10k–30k	about 0.33–0.44	0.18–0.94
30k–100k	about 0.22–0.33	0.12–0.56
100k–300k	about 0.15–0.29	0.07–0.46
300k–1M	about 0.11–0.16	0.05–0.34
1M–3M	about 0.04–0.07	0.025–0.15
3M+	about 0.03–0.07	0.018–0.10

Source: UniSong Creator Studio analysis of creators evaluated for client briefs (mostly AI apps); median views of each creator’s last 10 videos (each four weeks to six months old) divided by followers at the time of the snapshot; anonymized aggregate, rounded. Tiers below 10k followers are not reported.

YouTube Shorts follows the same slope in coarser tiers: views per follower on a typical post (median of the last 10 videos, each four weeks to six months old) are about 0.2–0.3 at 10k–100k followers, about 0.1 at 100k–1M, and about 0.04–0.05 at 1M+.

Instagram shows the same pattern. Using the average of recent posts, 1M+ accounts drew roughly a quarter to two-fifths of the views per follower of 10k–100k accounts. Our Instagram data supports only that relative comparison, so it is not on the chart.

Two uses for the table. First, it answers “is this many views good for this creator”: multiply the follower count by the tier’s range and compare. Second, it shows that smaller accounts get more views per follower. That does not make their views cheaper: our rate card found implied CPM lower for large accounts than for small ones. Our guides to micro-influencer marketing and nano-influencer marketing cover when smaller creators are still the right buy. The middle-half column is a reminder that tier alone decides little. Inside any tier, creators sit several times apart.

Engagement rate by follower count: judge big accounts per view

Large accounts usually look weak on engagement rate. In our data that gap is mostly reach.

Likes per view hold up. Likes per view do not fall with account size on TikTok (1M+ vs 10k–100k: about 0.9–1.3x) or on Shorts, where there is no consistent decline. These are each creator’s median over their last 10 videos (each four weeks to six months old).

Comments thin out faster. A 1M+ TikTok account drew about 0.6x the comments per view of a 10k–100k account. On Shorts it was a third to a half. Long-form comments thin out too. If the brief needs conversation under the post, such as product questions or replies to a call to action, smaller accounts are likely to deliver more of it per view.

The follower-based gap is mostly views. Put together, a 1M+ account earns roughly a fifth of the likes per follower of a 10k–100k account on TikTok, and about a quarter to a third on Shorts. The whole gap, or nearly all of it, is fewer views per follower, not weaker engagement per view.

So do micro-influencers have higher engagement? Per follower, yes, by a wide margin. Per view, not for likes. For comments, yes.

So when someone asks what a good engagement rate is for a creator of a given size, the follower-based answer will tend to favor small accounts. Compare engagement per view, and compare it only within the same follower tier. Our TikTok engagement rate calculator runs both the view-based and follower-based formulas side by side, with benchmark ranges.

How long do TikToks keep getting views? When to read results

A campaign report is a snapshot of a moving number. Read it too early and every creator looks weaker than they will be.

Within the first 24 hours, a TikTok has roughly a tenth of the views of the same creator’s typical 2–6-month-old video. At about a week, it is at about half (45–55%). At one to two weeks, a little over half (roughly 50–60%). At two to four weeks, roughly two-thirds to three-quarters (about 65–80%).

Long-form YouTube is further along at the same age: about 70–75% of the creator’s typical 2–6-month-old video at one to two weeks. That contrast with TikTok is a narrow one, so don’t lean on it hard.

These are cross-sectional figures. Each video was measured once, so we are comparing younger and older videos from the same creators. No single video was tracked as it grew.

The reporting rule we take from this:

• Don’t read TikTok views at day 7. Read them no earlier than day 30, and at day 45 to be safe.
• Read long-form YouTube no earlier than about day 30; later is safer.
• Views keep growing after that on both, so compare creators at the same age.

If your reporting template closes at two weeks, label the TikTok views as partial and schedule a second read. Our influencer marketing reporting KPIs guide covers which metrics belong in the early report and which in the final one.

A checklist for reading a creator’s numbers

1. Ask for the last 10 posts that are at least four weeks old, rather than an all-time average.
2. Take the median, and look at the spread. One breakout can carry a large share of the total.
3. Check views per follower against the tier range, and expect it to fall as follower count rises.
4. Separate reach from engagement per view, and compare engagement only within the same follower tier.
5. Forecast a single post from the median. Forecast a TikTok package from the average, discounted to about three-quarters.
6. Report no earlier than day 30, with every creator measured at the same post age.

If you want a team to run this vetting and the campaign around it, our influencer marketing service starts from the same checks.

Related reading

• Influencer rate card benchmarks: tiers, relative quotes and implied CPM.
• TikTok engagement rate calculator: view-based and follower-based formulas.
• Micro-influencer marketing and nano-influencer marketing: buying reach from smaller creators.
• Influencer marketing reporting KPIs: what to report, and when.

Methodology

The pool. Every figure comes from profile snapshots of creators UniSong evaluated for client briefs, most of them for AI apps, with app and tech categories over-represented. This is not a random sample of TikTok, YouTube or Instagram, and pools from other categories may look different. The figures describe creators’ own posts. None of them describe UniSong campaign results.

Platforms. TikTok and YouTube figures come from video-level snapshots. YouTube is split into Shorts and long-form throughout, because the two behave differently. Our Instagram data has no per-video rows, only the average of recent posts and follower count, so Instagram appears only in one relative comparison.

The window and the last-10 rule. Sections on skew, size and engagement use each creator’s last 10 videos that were four weeks to six months old when the snapshot was taken. Younger videos are still collecting views. Older ones fall outside a recent picture of the account. Creators without 10 videos in that window are left out, which under-represents creators who post several times a day, because their recent videos rarely reach back four weeks. Skew figures depend on the window, as the 15- and 20-video results show, so every skew figure here carries the last-10 qualifier. The forecast test uses the last 12 (see below).

Pinned posts. TikTok pinned videos sit at the top of a profile regardless of date, and often have far more views than their neighbors. We removed them with a date-gap rule that flags a video much older than the run of posts after it. The four-week-to-six-month window already excludes most of them.

Forecast test. For each creator we took the last 12 videos in the window, used the older 8 to make a forecast from the median and from the average, and compared the forecast with the newer 4. A single-post miss is measured as a factor in either direction (a post at half or double the forecast both count as a miss of ×2), and we report the median miss per creator, then the median across creators. The package result divides the average of the newer 4 by each forecast, then takes the median across creators.

Timing. For each creator, the reference is the median views of their videos aged two to six months. Each younger video is compared with that reference, grouped by age, summarized per creator and then across creators. No video was measured twice, so this is a cross-section of different videos at different ages. Creators need at least three videos aged two to six months to have a reference, so creators who post more than about daily drop out (their snapshots rarely reach back two months), and the youngest age groups over-represent creators who post more often. We don’t publish a Shorts timing level because it moved too much across client groups.

Summaries and intervals. Every figure is computed per creator first, then summarized across creators as a median, the middle half, or an equal-weight mean, so no single viral creator can move a headline. The size gradient (views for 10x the followers) is a slope fitted across creators. Engagement comparisons by tier use geometric means of creator medians. Uncertainty comes from bootstrap 95% intervals that resample creators rather than videos; package intervals are exact intervals for a median. YouTube upload dates only record the day, so “last 10” can depend on how same-day uploads are ordered. We checked many orderings and published only figures that held across them.

Client-group check. Every headline was re-estimated with the largest client group left out, and again with client groups weighted equally. A figure is published only if the estimate without that group falls inside its published range. Where checks moved a result, we publish the range rather than a point.

What we don’t publish. Creator, video and campaign counts, client and brand names, internal system names, and any price. Figures are rounded.