Instagram Sends vs Likes: Why One DM Share Beats Fifty Hearts
- Instagram's stated signal order puts watch time first, then sends or DM shares, then likes. Sends matter more than likes for reaching non-followers.
- The logic is cost: a like is anonymous and free, while sending a post to a specific person spends social capital and predicts real value.
- Content designed for sends looks different from content designed for likes — it is useful, surprising, or relatable enough to think of a specific person.
- Sends are partly visible in Instagram Insights, so this is testable on your own account rather than something you have to take on faith.
The like button is the most prominent number on Instagram and one of the least important ones for whether anybody new sees your post.
Instagram's own restatement of its ranking signals puts watch time first, sends second, likes third. Sends — sharing a post to someone via DM — do the heavy lifting for reaching people who do not already follow you.
Why sends carry more weight
Think about what each action costs the person doing it.
| Action | Cost to the user | What it predicts |
|---|---|---|
| Like | Near zero. Anonymous, instant, no consequences | Mild approval, or a reflex |
| Save | Low. Private, but implies future use | Personal utility |
| Send | High. You attach your name and spend social capital | You believe a specific person will value it |
Sending a post is a small act of vouching. You are telling someone you know: this is worth your time, and I am the one telling you. Nobody does that casually, and nobody does it for content they merely thought was fine.
A like says "I approve." A send says "I know someone whose day this improves." Only one of those predicts that a stranger will care.
From a ranking perspective, that makes sends an exceptionally clean signal. They are hard to fake at scale, they correlate with genuine value, and they carry a built-in distribution effect — every send delivers your content to a person who did not have to be recommended it.
The same logic is showing up everywhere in 2026. Xiaohongshu's scoring formula reportedly weights shares and follows several times higher than likes. Douyin's reordered weightings moved saves and repeat visits above likes. Platforms are systematically down-weighting the cheapest action and up-weighting expensive ones.
Content built for sends looks different
If you set out to maximize likes, you make content that is agreeable. If you set out to maximize sends, you make content that makes someone think of a specific person. Those are not the same brief.
The categories that reliably generate sends:
- Useful enough to forward. A checklist, a warning, a genuinely good recommendation. The send says "you need this."
- Relatable enough to tag someone in. The "this is literally you" reaction. The send says "this is us."
- Surprising enough to be worth reporting. A counterintuitive fact or a strong take. The send says "did you know this."
- Timely enough to be relevant to someone right now. The send says "you were just asking about this."
What almost never gets sent: generic aspirational content, engagement bait, and posts whose entire value is aesthetic. These can accumulate likes indefinitely and still go nowhere, which is precisely the pattern behind the most common complaint on Instagram — good likes, no growth.
There is a useful test when drafting: picture a specific person receiving this post in their DMs, and write the message that would go with it. If you can write that message in a few natural words — "this is the checklist I mentioned", "this is so you" — the post has send potential. If the only message you can imagine is "cool, right?", it does not. That test takes ten seconds and filters better than any engagement-rate benchmark.
Sends also interact with topic classification. A post gets forwarded when it is unmistakably about something a specific person cares about, and vague content is hard to forward for the same reason it is hard to classify. If your real problem is that nobody can tell what your account is for, low sends are a symptom rather than the cause — that mechanism is covered in your Instagram content fingerprint.
One caution worth stating: do not ask for sends. "Send this to a friend who needs to hear it" reads as engagement bait, and platforms have been reducing the reach of posts that solicit interactions directly. The mechanism works when the send is spontaneous. Asking for it degrades both the signal and the content.
Watch time still ranks above sends
Worth being precise here, because "sends beat likes" is easy to over-apply. Watch time comes first in the stated order, and early performance — the opening moments of a Reel — is weighted heavily.
The practical sequence is therefore:
- Watch time gets you evaluated. If people leave in the first two seconds, nothing downstream matters, because there is no audience left to send it.
- Sends get you expanded. Once the content holds attention, sends are what push it beyond your existing audience.
- Likes confirm. They correlate with the above, which is why they look predictive without being causal — and why chasing them directly rarely moves anything.
These signals also sit downstream of a separate gate. Instagram's topic controls, which reached the main feed in June 2026, let users declare which subjects they want more or less of — so signals determine ranking among content you are eligible to be shown for, which is covered in Instagram's Your Algorithm. Strong sends cannot rescue a post that never entered the candidate pool.
Two other notes from Instagram's restated signals: carousels outperform single images, and original content is prioritized while low-value reposting is penalized. Carousels are interesting here because they buy watch time and sends simultaneously — multiple frames hold attention, and a carousel with a genuinely useful final slide is highly forwardable. The originality question we covered in what counts as original content.
Check this on your own account
You do not have to take any of this on faith. Instagram Insights exposes shares and saves alongside likes, so the test is straightforward:
Pull your last twenty posts. Rank them by likes. Rank them by sends. Then rank them by reach. If the send ranking predicts the reach ranking better than the like ranking does — and on most accounts it will — you now have your own evidence, on your own audience, in your own niche.
That is worth more than any general advice, including this article. Signal weightings are described publicly in vague terms and change without announcement; your own analytics do not. If you run several accounts, you can run the same comparison across niches simultaneously and see whether the effect size differs — one of the few genuine advantages of a multi-account setup that has nothing to do with volume. Tools that keep multiple accounts running with distinct content per niche, ours included, make that kind of parallel comparison practical. But a single account and twenty minutes in Insights gets you most of the answer.
FAQ
Can I see how many sends a post got?
Yes — Instagram Insights reports shares for posts and Reels, which includes DM sends alongside other share types. It is not broken out by destination, so you cannot separate a DM send from a share to Stories. For tracking purposes the aggregate is fine: what you are watching is the trend and its correlation with reach, not the precise composition.
Do saves work the same way as sends?
Similar direction, different meaning. A save signals personal future utility; a send signals value to someone else, which carries the extra distribution effect of actually delivering your post to a new person. Both outrank likes in usefulness. If you have to choose a design target, sends have the broader reach implication — but saves are the stronger signal on some other platforms, so do not port this conclusion elsewhere without checking.
Does this mean likes are worthless?
No — likes still appear in the signal order, and they correlate with content people enjoy. The mistake is treating them as the objective. Likes are a byproduct of good content, not a lever you pull. If you optimize for sends and watch time, likes follow; the reverse is not true.