Somewhere in the last few years the advice changed. Likes stopped being the number worth chasing and saves took over, usually stated as a multiplier: a save is worth five likes, or ten, or twenty, depending on which deck you are reading. It is now common enough that people restructure whole videos around it.
The multiplier does not come from anywhere. TikTok has never published a weighting for saves, or for likes, or for any other engagement action, and nobody outside the company can measure one from the outside. What can be checked is what a save actually is, what it does that a like does not, and how often it happens. All three are more interesting than the multiplier, and only one of them supports the advice.
What TikTok says feeds the feed
TikTok’s own description of the For You feed lists what it calls user interactions as “the videos you like or share, accounts you follow, comments you post, and content you create”. That list, in how it recommends videos, does not mention saves.
Be careful with that, because it is the kind of absence people over-read. The sentence begins “including things like”, which makes it illustrative rather than complete, and the same page mentions favourites elsewhere as a gesture you can perform. The honest reading is not that saves are ignored. It is that TikTok has never confirmed they are counted, never mind how heavily, and anyone quoting a ratio is filling that silence with a number they made up.
What the same document does state plainly is the shape of the weighting. “A strong indicator of interest, such as whether a user finishes watching a longer video from beginning to end, would receive greater weight than a weak indicator, such as whether the video’s viewer and creator are both in the same country.” Watch signals sit at the top. It also says follower count is not a direct factor, and neither is how your previous videos performed.
So the case for saves cannot be made around watch time. It has to be made through it, and there is a version of that argument that holds up.
A save is a claim on a future session
Every other engagement action is spent at the moment it is given. A like is a reaction. A comment is a reaction typed out. A share moves the video sideways to somebody else, immediately.
A save is the only one that points forwards. The viewer is not telling you they enjoyed it. They are telling you they intend to come back, and the video goes into their Favourites to wait. If they do come back, the return visit is a replay, a replay is watch time on the same video, and watch time is the signal TikTok has actually said it weights heavily.
That is a real mechanism, and it is worth understanding precisely, because the precision is where the advice usually goes wrong. The save is not the valuable event. The return is. A save that never gets revisited is a bookmark in a folder nobody opens, which is most of them, and it produces exactly nothing beyond a number in your analytics.
This also explains why the payoff is slow and quiet. Likes arrive in the first hours, while the video is still being tested on new audiences. Saves keep accumulating for weeks, and the replays they eventually produce land long after the post has stopped being distributed. By the time a save turns into watch time, the video’s reach has usually been decided.
How often a save actually happens
There is no published benchmark for TikTok save rate. Socialinsider, whose TikTok benchmarks are the most widely quoted set going, folds saves into one composite figure: likes plus comments plus shares plus saves, divided by views. Saves never appear on their own, and we could not find a standalone figure published anywhere else either. So when a post tells you a good save rate is one percent, it is not citing a study, because there is no study to cite.
We went and measured it instead, in the narrowest way that is still honest about itself. We read the public counts on the most recent videos from a dozen large accounts across different content types, kept the 73 videos that were posted in the twelve months to 5 September 2026, and worked out each rate against that video’s own view count. Nine accounts survived that filter.
- Likes: a median of 6.9% of views.
- Saves: a median of 0.39% of views.
- Shares: a median of 0.18% of views.
In ratio terms the median video collected about one save for every nineteen likes. Saves outnumbered shares on 53 of the 73 videos, which is the one part of the received wisdom our numbers support: saves are not the rarest thing on the list, and treating them as an exotic signal is wrong.
The spread is the interesting part. Individual videos ranged from 0.03% to 4.0% saves against views, a factor of more than a hundred. And it sorted almost entirely by what the video was for, not by how good it was. The recipe account in the sample sat at a median 1.55% save rate. The wordless comedy account sat at 0.27%, with a higher like rate. Both are enormous, both are excellent at what they do, and one of them gets close to six times the saves because the thing it makes is worth keeping and the other is worth watching once.
Nine accounts is not a benchmark and we are not going to call it one. They are all large, which almost certainly flatters the numbers, and save counts keep climbing after a snapshot is taken. Treat it as an order of magnitude: saves run somewhere around a twentieth of likes on entertainment, and closer to a third on reference content.
The trade-off nobody mentions
Once you know saves track content type, “optimise for saves” stops being free advice and starts being a decision with a cost.
Save-worthy content is reference content. A recipe, a checklist, a settings walkthrough, a list of five things in an order you can follow later. What makes it worth saving is that it is dense and complete, and density has a specific problem: it competes with the opening. Every second spent establishing that this video contains a useful list is a second not spent making anyone want to watch it, and the opening is where distribution is decided. We wrote about that mechanism in how to write a TikTok hook, and it does not stop applying because the video is useful.
The videos that manage both do it by putting the payoff first and the reference material second, so the hook is doing hook work while the saveable part is still coming. That is harder than either one alone, which is why most accounts quietly pick a side.
If you are going to pick, pick on what the account is for. An account that sells something benefits more from a viewer who returns than from one who taps a heart and forgets. An account that grows on reach benefits more from the like rate, because the like rate is downstream of the same thing the recommendation system is reading.
Read your own numbers instead
Both counts are public on the video itself, and both are in your own analytics with more detail. In TikTok Studio, under Analytics and then Content, opening a post gives you its saves alongside watch time, completion rate and new followers.
Compute the rate against views yourself and compare it with your own back catalogue rather than with anyone’s benchmark. Your last twenty videos are a far better control group than a median taken across industries, because they share your audience, your niche and your format. A video that saved at three times your normal rate told you something. A video that saved at 0.4% told you only that it is a normal video.
The same warning applies here as anywhere else in analytics: read the trend across ten posts, not the spike on one. Our piece on TikTok engagement rate goes through how to do that without panicking at a single bad week, and how to get more views covers the distribution side that all of these numbers sit downstream of. If you want a quick outside read on whether the account is underperforming at all, our TikTok Health Scan works from a public username.
What likes still do that saves cannot
There is one job the like count has entirely to itself, and it has nothing to do with ranking.
It is the number the next viewer reads. All four counts sit on the right of the screen, but the like count is the one people have been trained to interpret, and it is read in the half second before anyone decides whether to keep watching. A video showing three likes tells that person nobody thought it was worth reacting to. A save count does not carry that, partly because it is an order of magnitude smaller and partly because nobody has any intuition for what a good one looks like, which our own numbers just demonstrated.
So the two are not competitors and the multiplier framing is the wrong shape. Likes are public proof aimed at the next viewer. Saves are private utility aimed at the same viewer, later. Only one of them is doing anything in the first hour, which is the hour that decides how far the video goes.
If a post is sitting at zero and you want the empty-count problem gone while you work on the real fix, our free TikTok likes puts 25 on a video you choose, once every 30 days, at no cost and with no password. Take it for exactly what it is: it removes the floor effect so the video is judged on itself. Twenty five likes changes nothing about distribution, and it will not rescue a video that people are seeing and ignoring.
Where we stand
We sell both. There are pages on this site for buying TikTok likes and for buying TikTok saves, so weigh what follows accordingly.
Bought likes do one thing, and it is the thing described above: they change how the count reads to the next person who lands on the video, during the short window while it is still being shown to new people. That is a cosmetic effect with a narrow window, and we have written it out at length in how to get more likes.
Bought saves are a harder case to argue, and we would rather argue it honestly. The entire mechanism that makes a save worth anything is the return visit, and a purchased save does not return. It moves a number on a page that most viewers do not read closely. If you want saves because the count looks thin next to your like count, that is a real cosmetic reason and the service does it. If you want saves because you have read that they carry more algorithmic weight, buy something else, because the weighting is unpublished and the part that would matter is the part you cannot buy.
The same applies in the other direction. TikTok’s Community Guidelines state that if it detects inauthentic metrics it will remove fake likes, followers or other inflated signals, and the scale is not theoretical: in 2024 it says it prevented over 54 billion fake likes and removed a further 8 billion. No seller can call any of this risk-free, us included, and we go through what actually happens in the risks of fake engagement.
Frequently asked questions
Is a save worth more than a like on TikTok?
Nobody outside TikTok knows, and TikTok has not published a weighting for either. The specific ratios you see quoted are invented. What is documented is that watch signals carry the most weight, which means a save matters to the extent that it produces a replay later, not as a value in its own right.
What counts as a good TikTok save rate?
There is no published benchmark, because the analytics firms fold saves into a combined engagement rate. In our own sample of 73 videos from large accounts, the median was 0.39% of views, with reference content running several times higher than entertainment. Your own last twenty videos are a better comparison than any of these numbers.
Do saves show publicly on a TikTok video?
Yes. The save count sits on the right-hand side of the video with the likes, comments and shares. What is private is who saved it: TikTok does not tell you which accounts did, and it does not send you a notification when someone does.
Do saves help you get on the For You page?
Not directly, as far as anything TikTok has published shows. The credible route is indirect and slow: a save leads to a return visit, a return visit is watch time, and watch time is weighted heavily. Most saves are never revisited, so most of them do nothing at all.
Why do my saves go up after the views stop?
Because saving is something people do to content they find later, through search, a profile visit or a share. Reach is decided in the first day or two. Saves keep trickling in for weeks afterwards, which is one reason they are a poor early signal of how a video is doing.
Should I ask viewers to save the video?
Only if there is a reason to. “Save this for later” attached to a video with nothing to come back to reads as a growth-hack tic and costs you a second of retention. Attached to a recipe, a checklist or a settings walkthrough, it works, because you are naming something the viewer had already half decided.
Do bought saves disappear?
They can. TikTok says it removes inauthentic metrics when it detects them, and that applies to saves the same as to likes and followers. Anything bought from a cheap panel in a sudden rush is the most likely to be swept.
The short version
The multiplier that says a save is worth five or ten likes is not from TikTok and not from any study. TikTok’s own list of user interactions does not name saves, and its published weighting says watch signals dominate everything.
The honest case for saves runs through that weighting rather than around it. A save is a promise of a return visit, a return visit is a replay, and a replay is watch time. That is real and it is slow, and it only pays off on the minority of saves anyone acts on.
In our sample of 73 videos, saves ran at a median 0.39% of views against 6.9% for likes, roughly one save per nineteen likes, and the biggest driver was not quality but whether the video was worth keeping rather than worth watching. Which is the actual decision. Reference content earns saves and pays for it at the opening. Entertainment earns likes and gets the first hour, which is the hour that decides the reach.
Pick the one your account is for, measure it against your own back catalogue instead of a benchmark nobody has published, and be suspicious of any number presented without a source. Including ours, which is why the method is written above.
