TikTok has the widest gap between what creators believe about hashtags and what the platform does with them. Almost every video carries #fyp, which is precisely why it carries no information.
The #fyp problem
A tag appearing on hundreds of millions of unrelated videos tells a classification system nothing. It cannot distinguish, cannot predict and cannot target. Including it is harmless but pointless, and it displaces a tag that would have done work.
What TikTok reads instead
TikTok's understanding of a video is built primarily from the video itself: on-screen text, spoken words transcribed automatically, the audio track, the visual content and the caption. Tags are a supporting signal, not the main one.
This has a direct practical consequence. Your tags should agree with your on-screen text and your spoken opening line. When a video says "three kettlebell mistakes" out loud, on screen and in the tags, every signal points the same direction and the first audience is better matched.
A three-tag structure
One subject tag naming exactly what the video is about.
One audience tag naming who it is for — a niche or community tag people actually follow.
One format or trend tag if the video genuinely uses that format. A tutorial tag on a tutorial. A challenge tag on a real entry into that challenge.
Add a fourth or fifth only if each one is doing distinct work.
Trend tags need timing
Trend participation on TikTok has a short window — typically days, not weeks. Entering a trend after the peak means competing against the accounts that defined it, with a video the audience has already seen versions of. Check the Creative Center for where a trend sits in its curve before building content around it.
Search is growing
A large share of TikTok usage is now explicit search, particularly among younger users treating it as a first-stop reference tool. That makes descriptive, question-shaped tags more valuable than they used to be. Tags that mirror how someone would phrase a query out loud tend to outperform clever ones.