First party data

What a TikTok account with 746 followers actually gets

Every benchmark page will tell you the average TikTok engagement rate. None of them shows you a real account's whole table. Below are 222 videos, every view, like, comment, repost and save, and the percentiles that come out of them.

Every figure on this page is recomputed at build time from the published table, a row per video export of the public TikTok account @viviandolorrr taken on 4 September 2026 and recomputed on 7 September 2026. Nothing here is typed by hand.

The short answer, if you need a number to cite

Two of the ten benchmark pages read for this page disclose a sample you can weigh. Use one of those and name the denominator, because a rate measured against views and a rate measured against followers are different quantities and mixing them is the most common error in this subject.

The two published TikTok benchmarks with a disclosed sample
Use thisFigureDivided bySample the page states
Socialinsider, Q2 20263.85%views2 million videos across 214,507 profiles, January 2024 to December 2025
eMarketer, overall4.1%views917 accounts, October to December 2023, accounts under 1,000 followers excluded
Rival IQ, median1.73%followersnot stated, but it is the only one of the ten publishing a median rather than an average

Check the formula as well as the denominator. Socialinsider, HypeAuditor and 1stCollab divide total interactions by views. Hootsuite and Rival IQ divide engagements by followers. Two of the ten, Creaticalc and Admanage, do not state a formula at all.

Across all ten pages the published range runs from 1.5 percent to 16 percent, and the spread is mostly denominator and tier rather than disagreement about reality.

On a views denominator the cluster sits between about 3 and 4 percent. On a followers denominator the same sources cluster between about 1.5 and 2 percent for a platform average, then jump to 8 to 16 percent for the small account tier, Brandwatch at 8 to 10 for under a thousand followers, Creaticalc at 8 to 14 for the same band and wask.co at 8 to 16 for one to ten thousand. That jump is the claim the rest of this page tests against a real account.

What none of the ten publishes is a distribution. That is what the remainder of this page is, and it is a single account rather than a survey, so read it as a case study of shape and read the table above for a number to cite.

The shape of a small account

583median views per video
3,370mean views per video
5.8xthe mean divided by the median
93%of videos land below the mean

The mean is 5.8 times the median, and 207 of 222 videos sit below it. On one account that is a description rather than a law, but the mechanism is not exotic. It is what happens whenever one or two videos escape and the rest do not, and a small account has few enough videos that a single escape dominates the average. The single best video took 38.3 percent of all 748,190 views on its own. The top ten took 75.4 percent. The bottom half of the catalogue, 111 videos, took 4.8 percent between them.

So when a benchmark page prints an average views figure with no median beside it, that number is describing the two videos that got lucky. Ask for the median. If the source cannot give you one, it does not have the rows.

The percentile ladder

All 222 videos sorted by views, from 21 to 286,400.

  • p10144
  • p25376
  • p50583
  • p751,194
  • p902,860
  • p953,606
  • p9947,074

The bars are on a logarithmic scale, anchored at the smallest and largest video in the table. On a linear scale the bottom three rungs would all render at the same minimum width, because the top of this distribution is 1,989 times the bottom, and a chart that flattens the low end would be arguing against the rest of this page. The numbers on the right are the real ones.

View count percentiles across 222 videos
PercentileViewsWhat it means
10th percentile144One video in ten did worse than this
25th percentile376The bottom quarter sits below here
50th percentile583The middle. Half the account is under this line
75th percentile1,194The top quarter starts here
90th percentile2,860One video in ten reached this far
95th percentile3,606One in twenty
99th percentile47,074One in a hundred, and this is where the outliers begin

How many videos clear a round number

Videos reaching each view threshold
Views at leastVideosShare of 222
10022199.5%
50012154.5%
1,0007031.5%
5,00073.2%
10,00041.8%
50,00031.4%
100,00020.9%

Seven videos out of 222 ever passed five thousand views. Four passed ten thousand. Two passed a hundred thousand. A creator who has read a benchmark page saying the average small account gets a few thousand views will conclude something is wrong with them. Nothing is wrong with them. They are reading a mean.

Find your percentile

Where your median lands against 222 real videos

Nothing is sent anywhere. The arithmetic runs in your browser against the ladder above.

If you do not know your median, sort your videos by views, take the middle one, and use that. The media kit tool works it out for you if you paste the numbers in.

Engagement rate, and which denominator was used

Likes divided by views across every video gives 2.30 percent. The median video scores 2.77 percent on the same formula. Median likes divided by 746 followers gives 2.14 percent. Three numbers, one account, one day. Any benchmark that does not name its denominator is not comparable to anything.

Likes per view percentiles
PercentileLikes per view
10th percentile0.65%
25th percentile1.63%
50th percentile2.77%
75th percentile4.22%
90th percentile6.19%

Comparing like for like

Socialinsider, HypeAuditor, 1stCollab and several others do not divide likes by views. They divide total interactions, likes plus comments plus shares and in Socialinsider's case saves, by views. Putting a likes only figure next to theirs compares two different quantities, which is the second most common error in this subject after the denominator. On the same formula they use, this account scores 2.73 percent, from 20,414 interactions over 748,190 views.

Against Socialinsider's 3.85 percent for Q2 2026, fetched from their page on 7 September 2026 and measured across 2 million videos, this account runs about a percentage point behind the platform. That is the honest comparison. The likes only figure of 2.30 percent is the one to use only when the thing being compared also counted likes alone.

Comments are the number nobody publishes. 120 of 222 videos here have zero comments and the median video has none. Total comments across fourteen months come to 971, against 1,158 saves and 1,062 reposts. Saves outrun comments on a small account, which matters if a brand is judging you on comment counts under a proposal.

What fourteen months did to the numbers

The first six months of the account returned a median of 1,597 views across 57 videos. The last six returned 397 across 127. Reach per video fell about 75 percent while output more than doubled. Posting more did not buy more.

Videos published, median views and mean views by month
MonthVideosMedian viewsMean views
2025-063406399
2025-079783829
2025-08161,75319,820
2025-09111,7367,293
2025-1081,2621,667
2025-11103,03420,160
2025-12141,2741,994
2026-01108362,334
2026-027641714
2026-037431536
2026-04158531,178
2026-0541167265
2026-0627281373
2026-0715451852
2026-0826477529
2026-093438691

The table has 16 rows because the run touches 16 calendar months. Elapsed time from the first video to the last is 432 days, which is 14.2 months, and the two counts are different measurements rather than a disagreement.

Two other things fall out of the table. Videos longer than thirty seconds have a median of 838 views across 74 of them, against 531 for the 57 videos of fifteen seconds or less, on a catalogue whose median length is 23 seconds. And the busiest month, 2026-05 with 41 videos, has the lowest median of any month in the set. Neither is proof of a rule on one account, and neither would survive a significance test at this sample size. Both are the kind of thing you can only see with the rows in front of you.

Against the benchmarks that already rank

Ten pages that rank for these queries were fetched on 7 September 2026 and their published figures read off the page. They do not agree, and the reason they do not agree is mostly that they are dividing by different things.

Published TikTok engagement benchmarks compared
SourcePublished figureTierDivided bySample stated on the page
socialinsider.io3.85 percent, Q2 20261K to 1Mtotal interactions over views2M videos across 214,507 profiles for the January 2024 to December 2025 study. The Q2 2026 figure is a later quarterly reading and the page does not restate a sample for it
rivaliq.com1.73 percent median by follower, 3.4 percent median by viewcross industryboth, separatelynot stated
emarketer.com4.1 percent overallaccounts of 1,000 followers and upviews, implied by the chart title, no formula given917 accounts, October to December 2023, sub 1,000 accounts excluded by design
blog.hootsuite.com1.5 percent platform averageplatformfollowersnot stated
brandwatch.com8 to 10 percentunder 1,000 followersfollowersnot stated, and the page says it is external research rather than its own
creaticalc.com8.0 to 14.0 percentunder 1,000 followersnot stateda 350,000 account study credited to SociaVault, with no date range and no published sample
blog.hypeauditor.comabout 10 percent is goodno tier givenviewsnot stated
1stcollab.com4.1 to 10.8 percentnano to megaviews100 million creator profiles claimed, no date range
wask.co8 to 16 percentnano, 1K to 10Kboth formulas given, not reconcilednot stated
admanage.ai43 views per 100 followersunder 5,000 followersreach, not engagement747,000 TikToks claimed, no date range

Two of the ten state a sample you could argue with. Socialinsider gives an account count, a post count and a date range. eMarketer gives all three and then excludes every account under a thousand followers, which is the band this page is about. The other eight publish a percentage with no disclosed sample at all.

Now put this account against them. On the followers denominator it scores 2.14 percent, which Brandwatch and Creaticalc would both call a failure for a sub thousand account, they say 8 to 10 and 8 to 14. On reach it gets 78 views for every hundred followers, against the 43 that Admanage publishes for accounts under five thousand.

By one benchmark it is far behind and by another it is well ahead, on the same day, from the same rows. One account cannot prove a benchmark wrong. What it can show is that two published benchmarks put the same account on opposite sides of normal, and because eight of the ten will not say what they measured, a reader has no way to work out which one to believe.

One more thing worth knowing before you quote a number from a benchmark page. Sprout Social publishes 5.7 percent and names Rival IQ as its source. Brandwatch publishes the same 5.7 percent for accounts under fifty thousand followers and names no source, saying only that it draws on external research. Rival IQ's own page, fetched the same day, shows 1.73 percent by follower and 3.4 percent by view, and 5.7 appears nowhere on it. A figure can travel between well known sites and stop matching the source it is credited to.

Take the table

The full file holds 222 rows with a publish date, views, likes, comments, reposts, saves and duration for each, plus the computed summary this page renders from. Captions and video ids are not in it. Use it, check the arithmetic on this page against it, cite it if it is useful.

Licence. Creative Commons Attribution 4.0. Credit iamvivi.com. Exported 4 September 2026 with yt-dlp from the public channel. Recomputed 7 September 2026. Rows 222. Size about 14 kB.

Two of the 224 videos the account has published were private or removed at export time and are not in the table. Instagram is not in it either, because a public Instagram grid does not expose per post counts, so those figures would have to be hand read and this file only holds what a machine measured.

Methodology

How we know these numbers. The public TikTok channel was exported with yt-dlp on 4 September 2026, giving one row per video with a publish date, view count, like count, comment count, repost count, save count and duration. 222 of the 224 videos the channel reports came back, and the export does not say why the other two were absent, so no reason is claimed for them. Every per video figure divides by 222.

The table is published at /data/creator-dataset.json and the site recomputes from it at build time, so a page and its data cannot drift. Figures were last recomputed on 7 September 2026.

Why every figure is computed twice. The build recomputes each summary value a second time from the row array rather than trusting the summary block. That check is what caught a published median of 590 when the rows said 583, and it is the reason the fetcher the pipeline uses cross checks the feed against the live media kit as well.

Where the benchmark comparisons came from. Ten pages that rank for these queries were fetched on 7 September 2026 and each figure was read off the page itself, along with whatever that page states about its own sample and its own formula. Nothing was taken from a summary of them. Where a page names another source for a number, that is recorded rather than resolved, which is how the 5.7 percent discrepancy surfaced.

What the export does not carry. A yt-dlp export holds no follower field, so the 746 TikTok followers and 404 Instagram followers were read off the public profiles by hand on the export date. They are the only hand entered numbers on this site. Instagram performance is excluded entirely because a public grid exposes no per post counts, and audience demographics are excluded because only the account owner can see them.

Limits, stated rather than buried. One account is one account. 222 videos from a single 746 follower creator is a case study of shape, not a survey, and no comparison here would survive a significance test at this sample size. What it offers that the ten benchmark pages do not is a complete table anyone can download and recompute.

Questions

What is a good engagement rate on TikTok

It depends entirely on whether the rate is measured against followers or against views, and most published benchmarks do not say which. Measured as likes divided by views, the account documented here runs at 2.30 percent across all 222 videos, with a per video median of 2.77 percent. Measured as median likes divided by 746 followers it is 2.14 percent. Same account, same day, two different answers. Ask which denominator a number used before comparing yourself to it.

How many views does a TikTok account with under 1,000 followers get

For this one, a median of 583 views per video across 222 videos and 746 followers, which is 0.78 views for every follower. The mean is 3,370, but 207 of 222 videos came in below that mean, so the mean is describing two lucky videos rather than the account.

Why publish the low numbers

Because a brand can check them, and because the benchmark pages that rank for these queries publish an average with no sample size, no median and no table behind it. A number nobody can audit is worth less than a small number anybody can. The whole table is downloadable at the end of this page.

Is one account a big enough sample

No. One account is one account, and this page says so rather than dressing 222 videos up as an industry study. What it gives you that an industry average cannot is the shape. You can see how far the top of a small account sits from its middle, which is the part an average hides.

How was the data collected

The public channel was exported with yt-dlp on 4 September 2026. Each row carries a publish date, view count, like count, comment count, repost count, save count and duration. The published copy keeps the counts and the dates and drops the captions and video ids. Figures are recomputed from that file every time the site builds.

The account behind the data

Vivian Dolor films island life across Southeast Asia and publishes the numbers rather than a screenshot of them. If a small, checkable account is what your brief needs, the rates and the terms are already written down.