Meta Ad Relevance DiagnosticsQuality Ranking, Engagement Rate Ranking and Conversion Rate Ranking

Ad relevance diagnostics are Meta's three per-ad quality metrics — Quality Ranking, Engagement Rate Ranking and Conversion Rate Ranking — that compare one ad against the other ads that competed for the same audience. Each is reported as Above Average, Average, or one of three Below Average tiers. They explain why an ad is expensive; they are not themselves a target to optimise.

The three diagnostics at a glance

Two properties of that table do most of the damage in practice. First, every diagnostic is relative: the same creative can read Above Average against one audience and Below Average against a more contested one, without anything about the ad changing. Second, they are reported only at the ad level — there is no ad set or campaign version of these numbers.

What the rating tiers actually mean

Meta reports five distinct values, and the three Below Average tiers are not interchangeable:

• Above Average — in the top 45% of comparable ads.
• Average — between the 35th and 55th percentile.
• Below Average (bottom 35% of ads) — a soft flag, usually fixable with an edit.
• Below Average (bottom 20% of ads) — a structural creative problem.
• Below Average (bottom 10% of ads) — replace the ad rather than edit it.

The tier is the actionable part. Bottom 35% on Engagement Rate Ranking is a hook or thumbnail problem. Bottom 10% on the same metric means the creative is not competitive in that auction at all, and iterating on it wastes the budget you would spend testing a genuinely different angle.

Quality Ranking

Quality Ranking reflects how people respond to the ad beyond clicking it: hiding it, reporting it, or selecting "why am I seeing this". Meta also treats several creative patterns as low-quality attributes in their own right — engagement bait ("tag a friend", "comment YES"), deliberately withheld information that forces a click, and sensationalised language.

What to change when it is Below Average

• Remove engagement-bait phrasing from the primary text and any on-image copy.
• Deliver the promise in the ad itself instead of gating it behind the click.
• Cut superlatives and countdown urgency that the landing page does not honour.
• Check the audience: a broad, poorly matched audience produces hides that have nothing to do with the creative.

Engagement Rate Ranking

Engagement Rate Ranking is Meta's estimate of how likely someone is to engage with the ad relative to competing ads. It is the diagnostic most directly controlled by the first frame, the first line of copy, and the format — which makes it the one that responds fastest to a creative change.

What to change when it is Below Average

• Rebuild the first three seconds. See hook formulas for the first three seconds.
• Test a different format — static against video, single image against carousel — before rewriting copy again.
• Confirm the ad is not simply worn out. The symptom pattern overlaps heavily with creative fatigue.
• Check placement-level breakdowns: one bad placement can drag the aggregate down.

Conversion Rate Ranking

Conversion Rate Ranking compares expected conversion rate among ads sharing the same optimisation goal. It is the diagnostic least likely to be a creative problem. A Below Average value here frequently points past the ad — to the landing page, the offer, the price, or a broken event signal.

What to change when it is Below Average

• Verify the conversion event is firing and deduplicated between pixel and Conversions API.
• Match the landing page to the promise in the ad — mismatched offers show up here first.
• Check mobile load time and above-the-fold clarity for the specific landing page in that ad.
• Confirm the optimisation goal matches the event you actually care about.

Reading the flags together

A single flag is ambiguous. The combination is what localises the problem:

Why the diagnostics affect cost

Meta's auction does not award impressions to the highest bid. It ranks by total value, which combines the bid with estimated action rates and ad quality. Relevance signals feed the quality component, so a weak diagnostic raises the bid required to win the same impression. That is the mechanism behind the familiar symptom: nothing about the budget changed, but CPM rose and delivery narrowed. The Meta ads auction page covers how those components combine, and Meta Andromeda covers the retrieval stage that runs before ranking.

One caution worth stating plainly: the size of that cost effect varies by account, vertical and auction, and any single percentage figure quoted for it — including ones you will find widely repeated — should be treated as an illustration rather than a measured constant.

When the diagnostics are not the answer

• Below roughly 500 impressions the values are absent or unstable. Do not act on them.
• During the learning phase they move for reasons unrelated to the creative.
• On a profitable ad a Below Average flag can be safely ignored. Cost per result is the objective; these are explanatory variables.
• Across different audiences the values are not comparable, because the peer group changes.

Turning a flag into a test

A diagnostic tells you which layer is weak; it does not tell you which variant wins. The practical loop is to change one layer at a time — hook, visual, or offer — and let the same diagnostic confirm the direction before scaling. That is the workflow described in diagnostics-led creative testing, and it is what AdRiseLab automates: generating structured creative variants that differ on one dimension at a time, so a Below Average Engagement Rate Ranking becomes a testable hypothesis instead of a guess. See how it works.

Related concepts