Metrics & KPIs

Engagement Rate vs Impressions: Reaction vs Raw Delivery

In short: Both describe how a social or video placement performed, but impressions is a fixed count and engagement rate is a ratio built on top of it. Impressions counts every ad serve with one consistent definition; engagement rate divides engagements by impressions or by reach, depending on the platform, and bundles very different actions - a like and a save - into one number. Impressions tells you scale on any objective; engagement rate is only meaningful on awareness or engagement-style placements and says little about sales. Use impressions to confirm delivery, use engagement rate only as a directional read on creative resonance, not as a stand-in for conversions.

By the AdFlint research team · Fact-checked against current Google and Meta platform behavior · Last reviewed July 2026

Engagement Rate

Engagements divided by impressions or by reach, depending on the platform, showing what proportion of delivery produced some recorded interaction.

Both halves are defined differently by every platform, so the metric only carries meaning within one account over time. It also mixes shallow taps with genuine signals like saves and shares. The common misreading is presenting it as performance for a conversion campaign; high engagement and low sales coexist constantly, especially with entertainment-led creative.

Full definition

Impressions

The number of times an ad was served or rendered, counting every appearance including repeat appearances to the same person.

It counts deliveries, not people and not attention. Platforms differ on when one registers, some on render and some on entering the viewport, so totals are not directly comparable across channels. The usual misreading is reading it as reach: a campaign with heavy frequency posts large counts while touching a very small audience over and over again.

Full definition

Side by side.

The differences that actually change what happens in your account.

	Engagement Rate	Impressions
Denominator	Impressions on some platforms, reach on others - it varies.	None - it is the denominator other rates are built from.
What counts as one unit	A bundle of different actions - likes, comments, shares, saves, expansions - counted the same.	One ad serve, a single consistent event type.
Consistency across platforms	Low - every platform defines both halves of the ratio differently.	High relative to engagement rate, though viewability rules still differ by platform.
Sensitivity to creative format	High - polls, carousels, and video invite easy taps that inflate the rate regardless of message quality.	None - a raw serve count doesn't care what format ran.
Relationship to sales	Weak and unreliable - high engagement and low sales coexist constantly with entertainment-led creative.	None claimed - it never represented outcomes, only exposure.
Where it's meaningful	Awareness and engagement-style social or video placements.	Every campaign objective and placement.
Summing across periods	Cannot be summed - must be recalculated from each period's own totals.	Sums cleanly across campaigns and time periods.
Failure mode	Presenting it as performance on a conversion campaign when it measures something else entirely.	Reading a large total as engaged attention rather than plain exposure.

What actually separates them.

Engagement rate's own denominator varies by platform - impressions in some reports, reach in others - so the same raw engagement count produces different rates depending on which tool computed it; impressions has no such ambiguity, it's a fixed count.

Engagement rate blends heterogeneous actions into one number, so a comment and a photo expansion often count the same, while impressions counts one consistent event type with no blending involved.

Impressions is tracked on every campaign objective; engagement rate is most meaningful on awareness or engagement-style placements and becomes noisy or near-meaningless on Search or a pure conversion feed.

Impressions is a supply-side count that doesn't care about creative quality; engagement rate is highly sensitive to format, since polls, carousels, and video invite easy taps that can inflate the number independent of message strength.

Impressions sums cleanly across campaigns and periods for reporting scale; engagement rate cannot be summed the same way, it has to be recalculated from each period's own totals.

Which one should you use?

Use Engagement Rate when

You're judging which creative variant resonates emotionally with the audience.
You're reading sentiment around a launch or brand moment.
You're comparing how organic a paid social post feels.
You're deciding whether a post is a candidate to boost further.
You're monitoring community reaction to a sensitive campaign.

Use Impressions when

You're confirming delivery scale before reading any rate built on top of it.
You're computing CPM.
You're verifying budget is pacing as expected.
You're setting a frequency cap.
You're comparing raw exposure across placements before layering on any engagement read.

Common questions.

Does a high engagement rate mean the campaign is working?

Not necessarily. High engagement and low sales coexist constantly, especially with entertainment-led or curiosity-driven creative that invites easy taps without moving anyone toward a purchase. Treat it as a resonance signal on awareness-style placements, not as evidence the campaign is driving business results.

Why is my engagement rate different in two different reports?

Check the denominator - one report may be dividing by impressions and another by reach, and those produce different rates from the same raw engagement count. Confirm which base each report is using before comparing them.

What actually counts as an engagement?

It's a platform-specific bundle that typically includes likes, comments, shares, saves, and sometimes clicks or expansions, all counted as equally valuable even though they aren't. Check the platform's own definition before drawing conclusions from a swing in the number.

Is engagement rate useful for a Search campaign?

Not really. It's built for social and video placements where reactions and shares are native behaviors; Search ads don't have an equivalent interaction surface, so the metric either doesn't apply or carries little signal there.

Or stop choosing between them.

AdFlint picks the setting, writes the ads, and keeps optimizing inside the Google and Meta accounts you already own.

Try AdFlint free See how it works

Related comparisons

All comparisons