How to Get More Views on TikTok: 6 Things We Could Actually Measure
Try it on your own video
Paste a link. Get the clips.
Almost every list of TikTok growth tips is a list of things somebody believes. We had 4,404 real posts with play counts, follower counts, durations, transcripts and on-screen text sitting in a database, so we tested the common advice instead of repeating it.
Four tactics survived. Two widely repeated ones did not. And several popular claims turned out to be things this data structurally cannot answer, which we would rather say than dress up.
How we tested. 4,404 posts across nine verticals, minus 132 paid ads, leaves 4,265. Medians, not averages. Every comparison is stratified by follower band, because most apparent content effects are really account-size effects — that single control killed two of the tactics below. We only report an effect that holds in the same direction across bands, and we suppress any group under 40 posts.
The four that survived
1. Put spoken audio in the video
The most consistent result in the whole corpus. Posts with speech beat silent posts on plays in all four follower bands (3.66x, 1.25x, 1.18x, 1.07x), and on shares and saves in all four as well. It is the only variable we found that moves every metric in the same direction at every account size.
The advantage shrinks as accounts grow, which is what a genuine effect usually looks like once other advantages accumulate. If you change one thing about a silent, text-only format, change this.
2. Match your length to your follower count
"Keep it short" is right for most creators and wrong for the largest ones. Median plays by duration:
| Duration | Under 1k | 1k - 10k | 10k - 100k | 100k+ |
|---|---|---|---|---|
| 0 - 15s | 6,652 | 50,350 | 84,550 | 169,900 |
| 61 - 120s | 920 | 10,300 | 42,450 | 379,300 |
Short wins by 7.2x under 1,000 followers and loses by 2.2x above 100,000. The full breakdown by follower count has all five duration bands.
3. Never land at 31 to 60 seconds
One duration band lost everywhere: on plays, on shares and on saves, at every account size — twelve band-level comparisons, none in its favour. Our read is that it is long enough to need real structure and short enough that most people do not give it one. Go under 30 seconds or commit past 60.
4. Open with a scenario, not a question
"Ask a question" is the first item on nearly every hook list. Across 2,421 measured openings it was the worst family we tested above 1,000 followers (0.82x, 0.83x, 0.60x). Scenario openers — POV, "when", "me when" — beat questions in all four bands by 3.0x to 5.9x. The hooks analysis has the full ranking, including the caveat that the scenario advantage is strongest in business content and reverses in some consumer verticals.
The two that did not survive
Adding text on screen
Posts with on-screen text median 34,800 plays against 24,700 without — a clean-looking 1.41x. Then we noticed the accounts using text had 11,800 median followers against 8,955. Split by band, the effect is 5.89x under 1,000 followers and 0.92x, 0.94x and 1.01x above it.
The whole result lives in the smallest accounts, and the likeliest reading there is selection rather than causation: among very small accounts, the ones bothering to burn a hook into the frame are the ones taking it seriously. Write on-screen text because most people watch without sound, not because it buys reach.
Using a trending sound
Comparing original audio against licensed or trending tracks, the effect held in 2 of 4 bands on plays and 2 of 4 on saves. There is a real-looking 1.41x advantage in the 1,000-10,000 band that replicates nowhere else, which is the signature of noise. One caveat on our own null: TikTok's flag marks creator-owned audio, which is not exactly the same question as whether a sound is trending.
What this data cannot tell you
We will not pad this out with tips we have not tested. Three of the most common questions are ones our corpus structurally cannot answer:
- Posting frequency. We hold a snapshot of posts, not accounts tracked over time. Nothing here speaks to cadence.
- Hashtags. The corpus is hashtag-sampled, so every post in it used them. There is no comparison group.
- Posting time. We have capture timestamps, not post-time-of-day matched to local audience, so any claim would be an artefact.
And none of it is watch time, which TikTok does not expose and which is the variable most likely to explain everything above. We are measuring the shadow, not the object.
The short version
- Speak in the video. It is the only thing that helped everywhere.
- Under 10,000 followers, cut to 15 seconds. Over 100,000, stop cutting your best material down.
- Do not land at 31-60 seconds.
- Open with a situation, not a question.
- Add on-screen text for the sound-off majority, not for reach.
- Do not rebuild your audio strategy around trends. We could not find the effect.
If you are pulling clips out of longer recordings, our AI clip generator applies the length bands and hook patterns above when it picks boundaries, and the free viral score checker will score a clip you already have.
Frequently asked questions
How do I get more views on TikTok?
Four things survived testing on 4,265 posts: include spoken audio, avoid the 31-60 second band, match length to follower count, and open with a scenario rather than a question.
Does posting more often get more views?
We cannot answer that here. Our corpus is a snapshot of posts rather than accounts tracked over time, so posting frequency is outside what it can measure.
Do hashtags increase views?
Our sample was collected by hashtag, so every post in it used them and there is no comparison group. We make no claim either way.
Does on-screen text get more views?
Not by itself above 1,000 followers. The apparent advantage disappears once account size is controlled for.