SYSTEM · ONLINE// ALGORITHM_SIGNALSBUILD · 2.5.β--:--:-- UTC

// ALGORITHM_SIGNALS

What the FYP actually rewards

The TikTok algorithm doesn't "see" your ad — it watches how viewers react to it for the first hour after you ship. Five upstream signals tilt that early reaction in your favor before a single user thumbs through. The Deep Bench report scores all five.

1. FYP distribution likelihood

A composite score: how much of your creative reads as FYP-native based on the platform's known distribution biases. Strong signals: vertical 9:16 framing, native captions, sticker-style on-screen text, conversational pacing, raw lighting. Weak signals: letterboxed footage, brand chrome over the corners, slow camera pans, jingle-style audio.

The algo doesn't literally check these — but viewers respond differently to FYP-native ads (longer watch times, more saves), and the algo reads thatresponse. So the signal is predictive even though it's not direct.

2. Sound-trend alignment

Whether the audio track is currently trending, recently trended, or custom. Trending sounds get a brief but meaningful distribution boost — TikTok's algo bucket-tests viewer reactions to trending audio more aggressively than to original audio.

The catch: trending sound has a short half-life. A sound that was peaking last week might be flat this week and actively counter-productive (looks late) next week. The Deep report scores this contextually, but for fast-moving creators the signal is best checked manually before launch.

For brand-safety paid placements (Spark Ads boosted across non-TikTok accounts), trending sounds also carry licensing risk. The rubric flags audio-copyright concerns separately under risk flags.

3. Hashtag effectiveness

Less impactful than people think — but not nothing. Hashtags help the algo classify the content into a discovery bucket; they don't drive distribution by themselves. The score weighs vertical-specific hashtags (e.g. #tiktokshop, #tiktokmademebuyit for shopping content) higher than generic ones (#fyp, #viral).

Three to seven hashtags is the sweet spot. More than ten reads as spam; fewer than three loses classification signal. Mixing high-volume (millions of posts) and niche (thousands of posts) tags performs best — high-volume gets you in the discovery pool, niche keeps you in the relevant cohort.

4. First-frame thumb-stopper

A subset of the hook score, scored independently because it measures one specific thing: does the literal first frame — the one TikTok auto-plays as the user thumbs past in the FYP — give the eye a reason to stop.

Strong: a face mid-expression, an unexpected object, a contrast-y color block, sticky text already on screen. Weak: a logo card, a slow zoom into the subject, a black/empty frame waiting for the subject to enter, a long stationary establishing shot.

The first frame matters more than any other frame. If the eye doesn't catch on it, the rest of the ad doesn't exist.

5. Spark Ad suitability

Whether the creative is suitable for being amplified as a Spark Ad (TikTok's native paid placement that boosts an organic post rather than running a separate creative). The signal weighs:

  • Native organic feel — if it looks like an ad, Spark Ads accelerate the scroll, not the conversion.
  • Audio licensing— Spark Ads can't use most trending sounds (rights). Original audio or properly-licensed tracks score higher.
  • Scrollable thumbnail— first frame works as a still in the "sponsored" feed slot.
  • Risk-flag clean — Spark Ads go through extra TikTok policy review; risk flags here are effectively disqualifying.

A high Spark Ad suitability score on an organic-feeling clip is the most cost-efficient ad you can run on TikTok — you boost a post that already has organic engagement signal, instead of paying to generate signal from cold creative.

// PUT_IT_TO_WORK

Run one of your own ads through the rubric and see the scores against everything you just read.

ANALYZE_AD →