Ad Relevance Score
Ad Relevance Score is a platform rating that measures how closely your ad's message and creative match the intent and expectations of the audience or keyword it targets.
Key takeaways
- Ad relevance rates your message against competing advertisers, not against a fixed standard.
- On search it reflects semantic match with the keyword plus click behaviour.
- Social platforms weigh negative signals such as hides, skips and reports heavily.
- One message per audience raises relevance; one message for many audiences dilutes it.
- Strong relevance lowers delivery cost but says nothing about commercial fit.
In depth
Ad relevance is a comparative judgement rather than an absolute one: the platform estimates how well your message matches the query or audience it was served against, then rates that match relative to other advertisers competing for the same people. On search it is inferred largely from the semantic overlap between your text and the keyword, and from whether users click when they see it. On social it leans on negative signals as well, such as hides, skips and reports, which count heavily against a weak ad.
The rating rises when one message addresses one situation. Repeating the query language, naming the segment and stating a single specific outcome all narrow the gap between what a person expected and what they got. It falls when one creative serves several audiences at once, when the hook promises something the destination does not deliver, or when the same ad runs long enough to fatigue its audience. The cost of narrowing is fragmentation: more variants, smaller learning pools and slower confidence per variant.
The practical routine is to segment first and write second. Split the audience into groups that would answer a different question about their own situation, write one ad per group, then give each a destination that repeats its promise in the opening line. A scorecard quiz keeps this honest, because the quiz topic can be worded exactly as the ad's hook, so someone who clicked a question about leaking pipeline immediately starts answering questions about their own pipeline rather than reading a company overview.
Relevance ratings are relative to the competitive set, so an unchanged ad can drop simply because a stronger advertiser entered the auction. They are also coarse: three or four bands hide meaningful differences and update slowly at low volume. A high rating is no promise of commercial fit either, since a curiosity-driven hook can match an audience perfectly while attracting nobody who buys. Read the rating as a delivery-efficiency signal rather than as a verdict on message quality.
Example in practice
How to measure it
Watch the rating in bands rather than as a precise number, and read it beside click-through rate and frequency for the same audience. A falling rating with rising frequency usually means fatigue rather than a bad message. A falling rating at stable frequency more often means the audience widened or a competitor arrived. Segment the report by audience and placement before deciding which of the two is happening.
Confirm with cost and quality together. Track cost per click, cost per quiz start or form submission, and the qualified-lead share of each variant. The pattern worth scaling is a variant whose relevance rating and qualified-lead share move upward together. A variant with excellent relevance but a weak qualified-lead share is attracting interest from people who sit outside your buying group entirely.
Common mistakes
The most common failure is scaling a winning ad by broadening its audience instead of building new creative. The same copy now runs against people it was never written for, the rating slips into a lower band, delivery is throttled and costs climb with no obvious change inside the account. When a winner needs more volume, duplicate the structure and rewrite the hook for the next segment rather than widening the targeting on the original ad.
The second is optimising the advertisement and ignoring the destination. Platforms read post-click behaviour, so an ad promising a diagnostic that lands on a pricing page produces quick exits which feed back as a weaker signal. Match the first line of the landing experience to the ad headline almost word for word, and remove any step that arrives before the thing the ad actually promised the visitor.
Frequently asked questions
How do I raise my Ad Relevance Score?
Match your ad copy and creative tightly to the keyword or audience, and send clicks to a landing experience that continues the same message. Splitting broad ads into segment-specific variants usually lifts relevance the fastest.
Where do I find my ad relevance rating?
On search platforms it appears as a component column on the keyword table, usually labelled below average, average or above average. On social platforms it sits in the delivery or quality diagnostics for each individual ad, often as a ranking against similar advertisers. Both live at ad or keyword level, so account and campaign views will not show it.
Does ad relevance affect what I pay?
Indirectly but reliably. Platforms discount delivery for ads their users respond to well, so a stronger relevance rating tends to lower the cost of the same placement and widen how often you win it. A weak rating works the other way: you keep bidding, win less often, and pay more for the impressions you do win.
How is ad relevance different from Quality Score?
Ad relevance is one component feeding the composite Quality Score on search platforms, which also includes expected click-through rate and landing page experience. On social platforms there is no composite and the relevance or quality ranking stands alone. Treating them as the same metric leads teams to fix the wrong part of the setup.
How many ad variants should I run per audience?
Enough to learn without fragmenting the data. A workable rule of thumb is two or three genuinely different hooks per audience, left running until each has accumulated enough impressions to be separated from the others. Running many more usually means every variant stays in a learning state and none of them reaches a usable conclusion.
My relevance dropped without any change. Why?
The rating is relative, so it can fall when a competitor enters your auction with stronger creative or when your audience has simply seen the ad too often. Check frequency first, then the auction insight or competitive overlap report. If frequency is high, refresh the creative; if it looks normal, the competitive set most likely changed around you.
Can a quiz landing page improve ad relevance?
It can, when the quiz topic restates the hook of the ad. Relevance is influenced by whether people engage after clicking, and a page that immediately asks the question the ad raised keeps them there instead of bouncing back. The gain comes from continuity of message, not from the quiz format on its own.