Why Your Engagement Dropped (And What It Usually Isn't)

The short answer

In the large majority of cases, engagement drops because the content got worse, the audience got saturated, or the account grew and diluted the percentage. Shadowbanning is almost never the explanation, though it does exist in narrow forms. Before you conclude the platform is punishing you, check whether your last twenty posts were as good as the twenty before them.

That's uncomfortable, which is exactly why the algorithm explanation is so popular.

The real common causes

Audience saturation on a format

You found something that worked. You did it again, correctly, because that's the sensible response. You did it eleven more times.

The people who loved it the first time have now seen the shape of it. The surprise is gone, and surprise was doing a lot of the work. Watch time softens, the system distributes less, and the numbers fall — without anything about your execution getting worse.

This is the single most common cause of a gradual decline over six to ten weeks. It feels like a penalty because there's no obvious mistake to point at.

Growth diluting the rate

If your following grew 40% over three months and your average likes stayed flat, your engagement rate halved on paper. Nothing broke. You added people who followed for one specific video and don't care about the rest, which is what growth actually looks like.

Any large spike brings in low-intent followers. The rate recovers slowly, or it doesn't and that's fine, because absolute numbers are what matter commercially.

Seasonality

Real and consistently underestimated. Late December through early January. August in the UK and much of Europe. Major sporting events and elections, which eat attention across every feed. The first week back after a long holiday, when everyone's clearing a backlog.

Before diagnosing anything, check the same period last year. Half of the panicked drops that land in our inbox are August.

You changed something and forgot

New editor. New caption style. Started using a template. Shortened everything to fifteen seconds. Switched from filming in your kitchen to a proper setup, which sounds like an improvement and sometimes reads as sterile. Look at what changed in the two weeks before the drop started. There's usually something.

Posting cadence collapse

You went from four a week to one. Or you posted eleven times in five days to make up for a gap and burned through the audience's tolerance. Both produce a decline, and the second is nastier than people expect.

The myths

Shadowbanning as first explanation

The term describes something real in a narrow sense: platforms do restrict content that's borderline against their guidelines from appearing in recommendation surfaces, and they don't always tell you. Instagram has account status tools that will say so. TikTok will often notify you.

What shadowbanning is not is a general-purpose explanation for a 30% dip in likes. If you're not posting anything near the guidelines, it almost certainly isn't happening to you.

The tell: a genuine restriction usually produces a cliff, not a slope. Reach goes from normal to near-zero on non-follower surfaces, immediately, and stays there. A gentle six-week decline is not that.

"The algorithm hates me"

Ranking systems don't hold grudges. They don't have a memory of you personally in that sense. Each video is scored on how people respond to it, largely fresh.

The version of this that's true: if your recent videos have poor watch-through, the system has less reason to push the next one hard out of the gate. That's not hatred, it's the same mechanism working correctly, and it reverses as soon as a video performs.

"They're suppressing my niche"

True at the margins occasionally. Mostly a story people tell to avoid the content conversation. If the niche were suppressed, the whole niche would be down, and that's easy enough to check.

How to diagnose properly

Work in this order. It's roughly cheapest-to-most-expensive.

Establish what actually dropped. Views? Watch time? Likes per view? Followers per view? "Engagement dropped" is four different problems and they have different causes. Falling views with steady watch-through is a distribution issue. Steady views with falling watch-through is a content issue.
Set the window properly. Compare rolling 30-day periods, not last week to the week before. Short-form performance is dominated by outliers, and one big video makes a fortnight look great.
Check the same period last year. Seasonality, as above.
List everything that changed. Format, length, cadence, editing, subject matter, who's editing, what device you film on.
Check platform account status. Both Instagram and TikTok expose this now. Two minutes, rules out the restriction question definitively.
Look at your retention graphs on the last ten videos. If the drop-off point moved earlier, you have a hook problem, and that's specific enough to fix.
Ask whether the audience is the same. Big growth spike in the recent past? Then the denominator changed.

The test that resolves most of it

Post three videos that are deliberately unlike your recent run. Different format, different length, different opening. Not better necessarily — different.

If one of them performs normally, your account is fine and you had a format fatigue problem. If all three land flat too, the issue is broader and worth investigating further. This is a fortnight of work and it answers the question more reliably than any amount of forum reading.

When it genuinely is a platform change

It happens. Platforms reweight things, push a new surface, change how a format is distributed. When they do, the signature is distinctive:

It's abrupt rather than gradual — a specific week.
It affects a whole content type, not just you.
Other accounts in your niche describe the same thing with the same timing.
It often correlates with a product announcement.

If three of those four are true, you're probably right. The correct response is still not to wait it out, but to work out what the platform is now favouring and test into it. Platforms broadly reward whichever format they're currently trying to grow.

A useful discipline: don't allow yourself the platform explanation until you've eliminated the content one. It's the harder conversation and it's usually the correct one.

Where this fits

Diagnosing this properly means keeping records — of what you posted, what changed, and how it performed — which almost nobody does while also making the content. CORE handles repurposing and channel management, and a fair part of that is maintaining the boring history that makes questions like this answerable rather than speculative. If you can't currently say what your average watch-through was three months ago, that's the gap.