DM Shares Are the New Like: What Moves Reach in 2026

Two videos, same account, same week. The first got 3,100 likes and 40 sends. The second got 290 likes and 260 sends. The second one is still pulling reach a month later. The first one was finished in eighteen hours.

If you’re trying to understand what changed about the Instagram algorithm in 2026, that’s the entire story compressed into two numbers. Nothing dramatic happened to distribution. What changed is which signal the system trusts — and the signal it trusts most now is the one that costs the viewer the most to produce.

Here’s the direct answer: a send is the most expensive thing a viewer can do with your video, which is exactly why it’s worth the most. A like costs a thumb twitch. A send costs a decision about a specific human being, a small wager of social credibility, and the risk of being the person who forwards mediocre content. Ranking systems have spent the last two years quietly re-weighting toward signals that are hard to fake and expensive to give. Everything below is about how to earn the expensive one on purpose.

Why the send became the scarcest currency

Every engagement signal has a price, and the price is what makes it informative.

Likes went through classic metric inflation. When a behavior becomes free, it stops separating good work from adequate work — people double-tap while already swiping away. A ranking system that leans on a saturated signal gets worse at ranking, so it does the obvious thing and reaches for the scarcer measurements.

Saves are more expensive than likes: the viewer is claiming a future use for the video. But saves are private. They tell the platform the content has utility, not that it has social value.

A send is categorically different. To forward a video, a viewer has to hold three thoughts at once: this is good, I know someone specific who would care, and sending it makes me look like someone with taste rather than someone who spams. That third one is the filter, and it’s brutal. Most content that clears “this is good” fails at “and I’d attach my name to it.”

This isn’t reading tea leaves. Instagram’s leadership has said publicly and repeatedly that sends per reach is one of the health metrics they watch most closely for Reels, alongside watch-through and reshare likelihood. TikTok’s own published description of its recommendation system says it weights signals by strength of interest, and that one strong indicator outweighs a stack of weak ones. Both platforms are telling you the same thing in different vocabularies: cheap signals got discounted, costly signals got promoted.

We covered the broader passive-versus-active split in Ghost Engagement: Why You Get Views but No Saves, Shares, or Follows. This post is about the top of that hierarchy specifically, because the send behaves differently from every other active signal — it recruits a second viewer.

The compounding part nobody accounts for

A like is a data point. A send is a data point plus a distribution event.

When someone forwards your video into a DM thread, the recipient watches it in the highest-trust context that exists on the platform: recommended by a friend, in a one-to-one conversation, with no competing content in the frame. Completion rates in that context are not comparable to feed completion rates. And a meaningful share of those recipients then forward it again.

So a send doesn’t just tell the ranking system your video is valuable. It manufactures a small pocket of near-guaranteed watch-through, which the ranking system then reads as further evidence. That’s the loop that makes a modest video keep moving for weeks while a high-like video dies overnight.

It also means a big chunk of your real reach is happening somewhere your dashboard describes poorly. Creators call this dark social — traffic and views that travel through private messages and leave a thin trail. You can see the send count. You cannot see the thread, the recipient, or the second-order forwards. Which is why creators consistently underestimate how much of their growth came from four people with good taste.

Four reasons a human forwards a video

Across the videos we analyze, the ones that pull disproportionate sends aren’t better made than the ones that don’t. They’re built around one of four social jobs. This is the most useful frame we’ve found, because it converts an unpredictable outcome into a design choice you make before you shoot.

1. It says something for them. The viewer agrees with a position they’ve been unable to articulate cleanly. Forwarding it is cheaper than making the argument themselves. This is why sharply worded opinion content out-sends balanced content by a wide margin — not because it’s more correct, but because it’s more quotable on someone’s behalf.

2. It’s aimed at a person they can name. Not “people who procrastinate” — their friend who has been saying they’ll start a channel since March. The narrower the described person, the more likely a viewer recognizes someone real. Broad relatability produces nodding. Specific relatability produces a send.

3. It’s a piece of social currency. Genuinely surprising information, an insider detail, a thing that makes the sender look informed. The forward is a small status move: I found this first. Novelty is the load-bearing element, and it has a short shelf life.

4. It’s an invitation to do something together. A challenge, a recipe, a place, a trip, a thing that’s better with two people. The send isn’t commentary — it’s logistics. These are the most underrated sends because they don’t feel like content performance at all.

Notice what’s missing from that list: quality, polish, and effort. A beautifully graded four-hour edit with no social job attached gets admired and abandoned. A phone-shot rant that says the thing 200,000 people were thinking gets forwarded into a thousand group chats.

How to build the send in during the edit

Four concrete moves, roughly in order of how much lift we see them produce.

• Name the recipient inside the video, not the caption. “Send this to the friend who keeps buying gear instead of shooting” does more work spoken in the first eight seconds than written under the post. The caption is read by people who already finished. The line spoken mid-video reaches the people still deciding.
• Make one sentence extractable. Sends often happen because a viewer wants to transmit a single line. Write one that stands alone without the video around it, and let it sit on screen, unhurried, with nothing competing for the frame. If your best idea is buried in the middle of a paragraph of voiceover, nobody can carry it out of the video.
• Take a position instead of surveying options. “Here are five approaches to X” is savable and unforwardable. “Four of these are a waste of your time, here’s the one that isn’t” is forwardable, because now the viewer has something to agree with in front of their friend.
• End on the loop, not on gratitude. The last three seconds are where the send decision gets made. “Thanks for watching, follow for more” spends that window on you. A closing line that reactivates the specific person the video was about spends it on the send.

One caution, because this fails loudly when overdone: explicitly asking for shares works exactly once per video and only when the ask names a real category of person. Generic “share this with someone who needs to see it” has been used flat and reads as filler. The ask has to be specific enough that the viewer’s brain returns an actual face.

Sends aren’t weighted the same everywhere

The send is the strongest signal on Instagram. On TikTok it’s strong but not dominant — TikTok’s distribution leans harder on completion and depth of watch, because its default surface is a stranger feed, and its native forwarding culture is weaker than Instagram’s DM-heavy behavior.

The practical consequence: a video engineered for forwarding can underperform on TikTok if you bought the send with a slower, more conversational build. On TikTok, you still have to survive the first two seconds against a viewer with no relationship to you. We broke the platform split down in Instagram Reels vs TikTok: What Actually Changes.

There’s also a classification effect worth knowing. Who forwards your video, and to whom, is some of the sharpest evidence the platform has about which audience your content belongs to — which is why send patterns quietly shape the slot you get sorted into. That mechanic is in How Instagram Decides What Niche You’re In.

Measure it as a rate, for one week

Totals scale with distribution and will lie to you. Do this instead.

1. Pull your last fifteen videos and compute sends per 1,000 views for each. That single column is a better predictor of next month’s growth than any other number on the page.
2. Sort by it. Look at your top three and write down which of the four social jobs each one was doing. You almost certainly have a house pattern and don’t know it yet.
3. Post four videos in a week, each built for a different one of the four jobs, and change nothing else. Four isolated variables beat one redesigned account.
4. Compare send rate — not views, not likes. The video with the fewest views may well be the one to build on.

The uncomfortable version of this advice: expect your view counts to get worse before your growth gets better. Content built for forwarding is more opinionated and more specific, which means fewer people watch and more people act. That trade is the whole point.

Where Blossom fits

This is the gap Blossom is built to close. Paste a video URL — yours or a competitor’s — and you get the hook scored with the reasoning behind it, the format and tactics the video is actually leaning on, and specific suggestions for what to change, all read against a large library of analyzed videos in the same space. For send-worthiness in particular, what’s useful is seeing whether the video contains anything a viewer could carry out of it: a position, a named person, a line worth quoting. That’s visible before you post, not three weeks later in your insights.

It won’t promise you reach, and no honest tool can. What it will tell you is whether the video you’re about to publish gives anyone a reason to hand it to someone else. Run it on your last post, or start with the FAQ if you’re still working out which signal your content is failing on.

So stop asking whether people will like your video. Ask who, specifically, is going to send it — and to whom. If you can’t answer that before you shoot, the algorithm won’t answer it for you afterward.