89% of TikTok Videos Have No Call to Action (And They Perform Better)

The finding nobody expects

Disclosure. FYPNow sells the analytics product that produced this data. We have an

obvious interest in you finding it credible, which is why the method, the sample bias and

the places the numbers disagree with our own marketing are all stated below rather than

buried. The figures are published under CC BY 4.0 — check them.

We analysed 2,584 TikTok videos with a video model that records, among other things, whether the video asks the viewer to do anything.

2,298 of them — 88.9% — ask for nothing at all.

And the videos that do ask engage worse. (This table needs a view count to compute rates, so it covers the 2,551 videos that have one — hence the slightly lower count in the first row.)

A video with a link CTA engages about 43% worse than a video with no CTA at all.

Read that carefully, though

This is the part where most data posts overclaim, so let us not.

This is correlation, and there is an obvious confound. CTAs cluster in promotional content, and promotional content underperforms for reasons that have nothing to do with the CTA. Look:

Promotional videos have a CTA 60% of the time and are the worst performers. Comedy has one 2.3% of the time. So "CTAs correlate with worse engagement" is partly just "ads correlate with worse engagement", which surprises nobody.

What survives the confound is narrower and still useful: there is no visible engagement bonus for adding a CTA. The widely repeated advice that you should always tell viewers what to do next is not supported by anything in this sample. If a CTA serves your business — driving a link, building a list — use it knowingly, as a trade rather than as a free lift.

The one CTA that did not hurt saves was follow (0.63%, the best of the four), while having the worst engagement rate. Asking for a follow appears to convert the people who were already going to stay, and to cost you everyone else.

What this looked like before we checked it

The first version of this analysis reported that 2,292 of 2,584 videos had no CTA. That is two

different denominators stitched together: 2,292 is the count among videos that have a view

count (needed to compute any rate), while 2,584 is the whole corpus. The real full-corpus

figure is 2,298.

A 6-video error changes nothing about the conclusion, and we would have got away with it. We

are flagging it because it is the characteristic way a data post goes wrong — not a fabricated

number, but a filtered subset quietly reported against an unfiltered total. If you are reading

someone else's study and the counts in the prose do not reconcile with the counts in the

tables, that is usually why, and it is worth asking what else was filtered.

Every table below states its own filter.

The strongest signal in the dataset: text overlays

Not the CTA. This:

A 2.7× difference in save rate, the largest gap we found on any binary attribute.

Two honest caveats. The no-overlay sample is small at 145 videos. And overlays are near-universal at 94%, so videos without one are unusual in ways beyond the overlay — they may simply be less produced overall.

Still: 94% adoption plus the largest measured gap makes this the closest thing to a floor requirement in the data. If you are shipping videos with no on-screen text, that is the first thing to change.

Captions: cheap, and it shows

About 27% higher save rate, on a near-even split — so this is the best-balanced comparison in the study, with no small-sample problem on either side.

Captions are also the cheapest intervention available. Every editor generates them automatically now, and the sample is still 54% without.

Showing your face: less mandatory than advertised

90% of the sample shows a face, and the 10% that do not do slightly better on both metrics.

The gap is small and the faceless sample is a tenth of the size, so this is not "faceless is better". But it is reasonable evidence that faceless content is not the handicap it is usually described as — which matters if you have been avoiding TikTok because you do not want to be on camera.

What educational content is doing right

Worth pulling out, because it is the clearest positive pattern here.

Educational content had the best median save rate of any content type at 0.90% — nearly double lifestyle's 0.49% and almost three times comedy's 0.33%.

Saves are the metric worth caring about most, because a saved video gets rewatched, and rewatches are the behaviour the algorithm reads most favourably. Educational content earns them structurally: the reason to save is that you will want it again.

Transformation content had the best engagement rate (9.22%) while saving less — people react, then move on.

The short version

• 89% of videos have no CTA, and adding one shows no engagement benefit in this sample. Use one when it serves a business goal, not because you were told to.
• Link CTAs are the most expensive — 4.34% engagement against 7.62% with none.
• Use a text overlay. 94% do; the ones that do not have a 2.7× worse save rate.
• Add captions. 27% better save rate, and it costs you one click.
• You do not have to be on camera. Faceless content did marginally better on both metrics.
• If saves matter to you, teach something. Educational content saved at nearly double the rate of lifestyle.

Method

2,584 TikTok videos, each analysed by the same pipeline: a video model records CTA type, overlay and caption presence, whether a face appears, and classifies content type. Engagement, save and share rates come from each post's own metrics. Medians throughout, because view counts are heavily skewed. Content types with fewer than 80 videos and CTA types with fewer than 30 are excluded.

The sample is videos FYPNow users chose to track, which skews toward content someone already thought worth watching. Nothing here is causal — every comparison has confounds, and the CTA one has a large and obvious one that we have shown rather than hidden.

Related

• We analysed 2,584 TikTok videos: the most common hook isn't the best one — the hook half of this study
• Looking at the camera underperforms: presenter data from 2,499 videos — eye contact, expression and delivery
• The anatomy of a TikTok video — length, pacing and audio
• What is a good TikTok save rate? — benchmarks for the metric doing the most work here
• Find viral TikTok videos in 4 steps — run this on your own niche