Every client conversation this quarter has eventually arrived at the same sentence: our engagement is down. It usually shows up with a screenshot attached and a tone somewhere between confusion and blame. And the screenshot is real — comment counts are falling almost everywhere we look.

Here’s the part that keeps getting missed. Views are holding. Watch time is holding. Shares are climbing, in some cases steeply. The audience didn’t leave. They stopped talking in the one room where we can hear them.

The comment section stopped being the room

Look at what’s actually moving. Portfolio data published this summer across dozens of brands found comments per post down roughly a quarter on TikTok and about a sixth on Instagram, while shares per post rose 45% on TikTok and 12% on Instagram — with views and watch time flat or up over the same period. Consumer research points the same direction from the other side: people who had a bad experience with a brand were about as likely to have one as ever, but the share who posted about it publicly fell from half to under a quarter in three years.

Two datasets, opposite ends of the transaction, same finding. Public performance is contracting. Private circulation is expanding.

That’s not apathy and it’s not a reach penalty. It’s a venue change. The conversation moved into DMs, group chats, close-friends stories, and text threads — rooms with no dashboard attached to them. When someone sends your video to three people, that’s the highest-trust action available to them on the entire platform, and it leaves almost no fingerprint on the post itself.

So most brands are currently being graded on the exact metric their audience walked away from.

Why “engagement is down” is usually a brief problem

Once you see the migration, the diagnosis changes. The content isn’t underperforming. It’s aimed at the wrong behavior.

Think about what most brand social is engineered to produce:

• A question at the end, asking for a reply

• A hot take, calibrated to be mildly disagreed with

• A poll, a caption prompt, a “tell us below”

• A trend participation post, which by design asks the viewer to recognize something rather than pass it on

All of that is built to harvest comments. It’s competent work aimed at a shrinking action. Meanwhile the growing action — the forward — has almost never been briefed for directly.

And forwarding is a different job. Nobody sends a video because a brand asked them to. They send it because sending it does something for them: it’s useful to a specific person, it makes them look like they know something, it settles an argument, it’s funny in a way that requires no setup, or it says something they wanted to say and couldn’t phrase. The send is a small act of social generosity, and the sender takes on a little risk when they make it. Content has to earn that.

The send test

This is the question we now put at the top of the brief, before anyone writes a word: Who sends this, to whom, and what do they get for sending it?

If the team can’t answer all three parts in one sentence, the idea isn’t finished. Not bad — unfinished. It goes back.

What that looks like in practice:

• Name a real recipient, not a persona. Not “millennial homeowners.” The friend who just closed on a house and is overwhelmed by the punch list. Specificity is what makes a thing forwardable; a piece written for a segment is written for nobody in particular and gets sent to nobody in particular.

• Write the ending first. Comment-bait ends with a question. Sendable content ends with a payoff — the thing the sender gets credit for delivering. If your last three seconds are a logo and a CTA card, you’ve spent the most valuable real estate in the edit on the least valuable content.

• Move the payoff out of the caption. Anything that only makes sense with the caption attached breaks the moment it’s forwarded, because half the time the caption doesn’t travel with it. The video has to stand alone.

• Cut a version with no brand for the first eight seconds. Not every time. But when the value is genuinely in the information, front-loading the logo taxes the send — people hesitate to forward something that reads as an ad, even when they found it useful.

• Reweight the scorecard. Share rate becomes the headline number, saves second, comments as texture. Do this before the client asks, not after.

• Stop grading the calendar and start grading the outliers. That same portfolio study found 30 posts drove half of all social value out of roughly 5,000 published in a year. Volume isn’t buying what people think it’s buying. Find what got sent, and make more of that.

What this changes on set

The measurement shift is easy to write about and hard to act on, because the actual change happens during production — long before anyone opens a reporting deck.

It changes what we shoot. Sendable content usually needs a genuine piece of information, a real demonstration, or an actual joke, and none of those get captured by accident. They get scheduled.

It changes the edit. If the payoff has to land in the last three seconds, the edit gets structured backwards from that beat, and the temptation to trail off into a soft outro disappears.

It changes coverage. A piece that works for one specific recipient is a piece that fails for everyone else, which is fine — but it means shooting three or four variants of the same core truth, aimed at different people, rather than one broad version aimed at all of them. Same day, same setup, different doorways.

And it changes review. The trend window is now visibly shorter than most approval processes. If a culturally reactive piece takes three rounds of review, the moment has passed before it ships. Reactive work needs a different, faster lane than campaign work, with the approvals pre-negotiated.

What we’d actually do next

If you take one thing from this: pull your last quarter, sort by share rate rather than engagement rate, and look at the top ten. Our experience is that roughly half of them will be posts nobody on the team was particularly proud of, and a few of the ones everyone loved won’t be there at all.

That gap is the brief. It tells you what your audience thinks is worth passing to someone they know, which is a far more honest signal than anything they’re willing to type in public.

Final thoughts

Quiet is not the same as gone. The most valuable thing a person can do with your content is hand it to someone whose opinion they care about, and that action has never been louder or harder to see. Brands that keep chasing the public metric will keep concluding they need to post more. Brands that start making things worth forwarding will post less and matter more.

We’d rather make thirty things that get sent than five thousand that get scrolled.