Stop chasing likes — a like on TikTok is the cheapest currency that will not buy you a place on the FYP. If you want the algorithm to adopt your content and push it to millions, design your videos to be saved, shared, and rewatched — not merely liked. This article puts the real weights of every signal in front of you with the numbers, and translates them into practical steps you can apply today.
Signals ranked by weight
The algorithm does not treat interactions equally — it assigns every action a user takes a weight that feeds into a mathematical equation determining whether a video deserves to advance to the next distribution wave or not.
| Rank | Signal | Why the algorithm values it | Estimated weight |
|---|---|---|---|
| 1 | Completion rate and replays | TikTok's primary goal: keeping users inside the app as long as possible. Above 70% average watch time + above 30% completion = digital gold | 10 points |
| 2 | External shares (WhatsApp, Instagram…) | Brings new users to the platform from outside for free — algorithm rewards this strongly | 8 points |
| 3 | Internal shares (within TikTok) | Circulates users locally within the app | 6 points |
| 4 | Saves and favourites | Signals high reference value — raises the rating of accounts that deliver useful content | 5 points |
| 5 | Engaged comments | Writing a comment means deep engagement — reading comments raises the video's watch time silently in the background | 4 points |
| 6 | Likes | The easiest and fastest interaction — people tap like without thinking or even completing the video | 2 points |
The short version: completion + replay ← external share ← internal share ← save ← comment ← like
The quantitative gap between signals
A like is the cheapest currency on TikTok because users tap it without thinking while scrolling. A share requires a conscious effort — and that effort is precisely what the algorithm rewards.
- One share = 8 to 10 likes in terms of the mathematical weight pushing the video forward
- One save = 5 likes — a save tells the algorithm the content is valuable enough for the user to return to later
- One "Not interested" tap = minus 20 likes — the heaviest negative signal by a wide margin
These numbers explain why a video with 50 shares outperforms a video with 5,000 likes in distribution reach.
Cumulative points matrix
Here is what actually happens at each viewer behaviour inside the first test sample:
| Viewer behaviour | Points granted | Technical result |
|---|---|---|
| Immediate skip in first 2 seconds | Zero points | If repeated by 70% of the sample → video dies at 200 views |
| Like press and exit mid-video | +2 points | Weak performance, video stops quickly |
| Comment + save to favourites | +12 points | Very good → Circle 2 (5K–10K views) |
| Complete + replay + share | +18 points 🚀 | Viral explosion — algorithm pumps clip to millions of screens |
Direct comparisons with real numbers
Shares vs likes
Two videos that reached the same count in their first test sample — 5,000 views — but with completely different engagement:
- Video A (showcase content): 800 likes + just 10 shares → stopped at 7,200 views and died
- Video B (practical information): only 200 likes + 250 shares → exploded past 180,000 views in 24 hours
Same starting views — but the share points from Video B led the algorithm to classify it as high viral potential content and expand distribution immediately.
Replay rate — the most striking signal
A 6-second video achieved a retention rate of 135% — meaning users rewatched it an average of one and a half times. The impact: the next test sample size multiplied by five times in a single step. The video jumped directly from 10,000 views to 100,000 in a few hours — because dwell time inside the app rose exceptionally high.
Negative signals and their destructive weights
Just as there are positive points, the algorithm applies negative points instantly to brake a video:
| Negative signal | Penalty weight | When does it become destructive? |
|---|---|---|
| Immediate skip (first two seconds) | Light on a single video | When it exceeds 75% of the first test sample |
| Tapping "Not interested" | Penalty equivalent to minus 20 likes | If 5% of the test sample taps it — the video is immediately isolated and the next video is affected |
"Not interested" is the heaviest negative signal — one tap cancels out twenty positive likes. This explains why posting content outside your niche punishes the account so quickly: existing followers who are not interested in the new topic tap "Not interested," and the video's points collapse immediately.
How to leverage the weights in your favour
- Hook the first 2 seconds: open with a surprising hook that shocks or intrigues the viewer — retention above 75% in the first two seconds
- Design for saves: provide dense practical information impossible to absorb in one viewing + say mid-video: "save this to try tonight"
- Trigger emotional sharing: create content that touches the viewer's identity and represents a collective opinion — automatically pushes them to share with friends
- Make content worth stealing: content that gets saved for later or sent to friends feeds the algorithm the highest-value signals all at once
How the weights have changed over time
The algorithm is not static — its priorities shift as the platform's goals evolve:
- 2019–2021: likes and follows carried significant weight — videos spread simply by accumulating a fast burst of likes
- Now: watch time and completion rate account for roughly 50% of the equation's total weight — because TikTok wants to keep users inside the app as long as possible to show more ads
- Current direction: saves and shares are gaining increasing weight — as TikTok competes with Google's search engine and the spread mechanics of other platforms
To understand how the algorithm makes its distribution decisions based on these signals, read TikTok algorithm & going viral. And for the complete picture on the platform, read The complete TikTok guide.
Likes satisfy vanity — shares, saves, and replays satisfy the algorithm and open the FYP doors. Those who understand the weights and design their content around them will be guided automatically to the top by the algorithm itself.