SEO

How ad relevance score works: fix costs, not numbers

Discover how ad relevance scores impact costs and visibility. Learn quick fixes to optimise performance and improve your ROI.

Ad relevance scores and Quality Scores estimate how well your ad matches what a user wants, and platforms use those estimates to adjust how much you pay and whether your ad shows at all.

How ad relevance score works: fix costs, not numbers, Juicy Designs

TL;DR: Quick Answer

Discover how ad relevance scores impact costs and visibility. Learn quick fixes to optimise performance and improve your ROI.

Ad relevance scores and Quality Scores estimate how well your ad matches what a user wants, and platforms use those estimates to adjust how much you pay and whether your ad shows at all. Three levers control the outcome: creative relevance (does your message match the intent?), targeting (are you reaching the right audience with the right ad group structure?), and post-click experience (does the landing page deliver what the ad promised?). Get all three right and your costs drop without touching your bids.

Before diving into platform mechanics, here are the three fastest actions to reduce CPM or CPC right now:

  1. Match your headline to the search term or audience pain point, a one-word mismatch between ad copy and user intent tanks expected click-through rate (CTR).
  2. Split any ad group with more than 10-15 keywords into tighter thematic clusters, this single structural fix often moves ad relevance from "Below Average" to "Average" within days.
  3. Change the H1 on your landing page to mirror your ad headline, message mismatch is the most common cause of a poor landing page experience rating, and it takes minutes to fix.

Quick wins you can act on today:

  • Swap a generic headline ("Learn More") for a specific benefit tied to the keyword.
  • Pause any ad set with a frequency above 3.5 and refresh the creative.
  • Run a five-second speed test on your landing page; anything above three seconds on mobile hurts conversion rate ranking.
  • Check whether your ad group keywords share a single clear intent, if they don't, split them.

Key Takeaways

Ad relevance scores are diagnostic tools, not performance targets, the goal is lower CPM and CPC, not a perfect score.

Point Details
Scores are diagnostic, not targets Use Quality Score and Meta diagnostics to find what to fix, not to chase a number.
Three levers control outcomes Creative relevance, audience targeting, and landing page experience each map to a specific diagnostic.
Fix order matters Address quality ranking first (fastest CPM impact), then engagement, then conversion ranking.
Timing expectations are real Allow 3-14 days for most fixes to register in diagnostics; landing page changes take longest.
One change at a time Changing multiple variables simultaneously destroys your ability to attribute what worked.

What is an ad relevance score and how does it differ across platforms?

The phrase "ad relevance score" is a broad, informal term covering two distinct platform systems: Google's Quality Score and Meta's Ad Relevance Diagnostics. Both estimate how well an ad serves the user, but they operate at different levels, use different scales, and diagnose different problems.

Google Quality Score lives at the keyword level. It runs on a 1-10 scale and is built from three components: expected CTR, ad relevance, and landing page experience. Each component carries a status of Above Average, Average, or Below Average. Quality Score is a keyword-level diagnostic designed to show you where the user experience is weakest, not to be optimised as a number in isolation.

Meta Ad Relevance Diagnostics replaced the old 1-10 relevance score in 2019. Instead of a single number, Meta now gives you three comparative rankings at the ad level: Quality Ranking, Engagement Rate Ranking, and Conversion Rate Ranking.

Both systems are primarily diagnostic signals, not auction inputs in the traditional sense. They tell you what to fix. The auction mechanics that actually change your cost operate separately, though they draw on the same underlying signals.

Dimension Google Quality Score Meta Ad Relevance Diagnostics
Level Keyword-level Ad-level
Scale 1-10 numeric Comparative labels (Above/Average/Below)
Components Expected CTR, Ad Relevance, Landing Page Experience Quality Ranking, Engagement Rate Ranking, Conversion Rate Ranking
Auction impact Feeds into Ad Rank (bid × quality × context) Feeds into "user value" in Meta's auction
Where to find it Keywords tab in Google Ads UI Ads Manager columns (customise columns)
Primary use Diagnose search ad performance Diagnose social ad performance
  • Google scores update continuously but are most stable after sufficient impression volume.
  • Meta diagnostics require a minimum number of impressions before labels appear; low-volume ads may show no status.
  • Neither score is a direct bid modifier, both influence cost indirectly through auction mechanics.

How does Meta calculate relevance and what do the diagnostics tell you?

Meta's three-diagnostic model replaced the single relevance score because one number couldn't tell you why an ad was underperforming. The three rankings each point to a different layer of the ad experience, and Meta's diagnostics compare your ad to others competing for the same audience to determine where you're weaker than the competition.

Quality Ranking

This ranking reflects how Meta estimates the perceived quality of your creative and copy relative to ads targeting the same audience. Low quality signals include excessive text overlays, clickbait-style copy, low-resolution images, and creative that feels mismatched to the audience. A "Below Average" quality ranking almost always points to a creative problem, not a targeting or offer problem.

Engagement Rate Ranking

This measures your ad's predicted engagement rate (reactions, comments, shares, link clicks) against competing ads. A low engagement ranking typically means the hook isn't landing, either the creative isn't stopping the scroll or the audience is wrong for the message. Tightening the audience or changing the opening frame of a video are the two fastest fixes here.

Conversion Rate Ranking

This ranking compares your predicted conversion rate to other ads with the same optimisation goal and audience. A "Below Average" conversion ranking with average or above-average quality and engagement is a clear signal that the problem is post-click: the landing page, the offer, or the checkout flow. Meta estimates relevance from expected positive and negative feedback from the targeted audience, and conversion behaviour is one of the strongest feedback signals in that model.

How Meta compares ads: Each ad is benchmarked against others targeting the same audience over a recent window. Meta's comparison window is commonly around 35 days, so labels reflect your competitive position in that period, not an absolute truth about your ad's quality. An ad that was "Above Average" last month can slip to "Average" if competitors improve their creative.

Where to find diagnostics in Ads Manager: Go to Ads Manager, select the Ads view, click "Columns," choose "Customise Columns," and search for "Quality Ranking," "Engagement Rate Ranking," and "Conversion Rate Ranking." Add all three. You can also add them to saved column sets for faster access in future audits.

Practical fixes mapped to each diagnostic:

  • Below Average quality ranking: Refresh creative assets, remove text-heavy overlays, test a new visual format (video vs. static), and review copy for clickbait signals.
  • Below Average engagement ranking: Test a new hook in the first three seconds of video, narrow the audience to a more specific segment, or try a different ad format (carousel vs. single image).
  • Below Average conversion ranking: Audit the landing page for message match, reduce form fields, improve page load speed, or reconsider the offer itself.

Pro Tip: When two or more diagnostics show "Below Average" simultaneously, use a Diagnostic Matrix before touching anything. Map the combination (e.g., Below quality + Average engagement + Below conversion) to a likely root cause. Changing creative when the real problem is the landing page wastes your testing budget and delays recovery.


How does Google Ads calculate Quality Score?

Google's Quality Score is a keyword-level diagnostic that runs on a 1-10 scale. Google recommends using Quality Score as a diagnostic filter to identify which part of the user experience needs attention, rather than chasing the number itself. The three components each carry a status of Above Average, Average, or Below Average, and each one points to a different fix.

Expected CTR

This component predicts how likely your ad is to get clicked when it appears for a specific keyword, based on historical performance for that exact search term. It accounts for ad position and other factors, so it's a normalised signal. A "Below Average" expected CTR usually means your ad copy isn't compelling enough for the intent behind the keyword, or your extensions aren't adding enough value.

Fastest fixes: rewrite the headline to include the keyword's core intent, add sitelink and callout extensions, and test responsive search ad combinations that surface higher-CTR variants.

Ad relevance

This component measures how closely your ad copy matches the keyword's intent. A "Below Average" status here almost always means your ad group is too broad, keywords with different intents are sharing an ad that can't speak to all of them well. The fix is structural: split the ad group into tighter thematic clusters where every keyword maps cleanly to a single ad message.

Grouping keywords by search intent before building ad groups is one of the highest-leverage structural moves in Google Ads. A keyword clustering tool can speed up this process significantly, especially for accounts with large keyword lists.

Landing page experience

This component reflects how useful, relevant, and fast your landing page is for someone who clicked the ad. Google considers content relevance (does the page deliver what the ad promised?), navigation ease, and page speed. A "Below Average" landing page experience score often coexists with a high bounce rate and low conversion rate.

Fixes: align the page H1 with the ad headline, remove unnecessary navigation that pulls users away from the conversion action, and compress images to improve load time. For a detailed breakdown of what makes a landing page convert, the anatomy of a high-converting landing page covers the structural elements that move both Quality Score and conversion rate.

Where to find Quality Score in the UI: In Google Ads, go to the Keywords tab. Click "Columns," then "Modify columns," and add "Quality Score," "Exp. CTR," "Ad relevance," and "Landing page exp." A dash (", ") in the Quality Score column means the keyword hasn't accumulated enough impressions for a reliable score, focus on keywords with data before drawing conclusions.

Pro Tip: Quality Score is calculated based on exact-match impressions for that keyword, not broad or phrase match variants. If a keyword has very low volume, its score may be unstable. Prioritise fixing Quality Score on your highest-spend keywords first, that's where the CPC savings are largest.


Meta diagnostics vs Google Quality Score: what actually differs for advertisers?

Running campaigns on both platforms means you're working with two different diagnostic languages. Understanding the practical differences helps you prioritise fixes and avoid applying Google logic to Meta campaigns (or vice versa).

Dimension Meta Ad Relevance Diagnostics Google Quality Score
Level Ad-level Keyword-level
Scale/Labels Above Average / Average / Below Average (Bottom 35%/20%/10%) 1-10; components: Above/Average/Below
Diagnostics available Quality, Engagement Rate, Conversion Rate rankings Expected CTR, Ad Relevance, Landing Page Experience
Auction impact Feeds "user value" score in Meta's auction alongside bid and estimated action rates Feeds Ad Rank: bid × quality × contextual factors
UI location Ads Manager > Ads view > Customise Columns Google Ads > Keywords tab > Modify Columns
Primary fix levers Creative refresh, audience tightening, landing page/offer Ad copy, ad group structure, landing page speed and relevance

Auction impact in practice: Ad Rank on Google combines your bid, auction-time quality signals (including expected CTR, ad relevance, and landing page experience), and contextual factors. Higher quality can win a better position at a lower CPC than a competitor who bids more but has weaker quality signals. On Meta, your ad's "user value" score, which incorporates the diagnostic signals, competes alongside your bid in the auction. Both platforms reward relevance with lower costs, but the mechanism differs.

When to prioritise which platform's diagnostics:

  • If your campaign objective is lead generation or e-commerce conversion, check Meta's Conversion Rate Ranking and Google's landing page experience component first, post-click problems are the most expensive to ignore.
  • If you're running awareness campaigns on Meta, Engagement Rate Ranking matters more than Conversion Rate Ranking.
  • For Google Search campaigns, ad relevance and expected CTR are the fastest levers because they directly affect Ad Rank and impression share.

Platform-specific actionables:

  • On Meta: read all three diagnostics together before deciding what to change. A single "Below Average" on conversion ranking with strong quality and engagement means the landing page is the problem, not the ad.
  • On Google: start with the keyword that has the highest spend and the lowest Quality Score. Fix its ad group structure first (ad relevance), then the copy (expected CTR), then the landing page.

Which signals actually move relevance scores?

Both platforms draw on a set of measurable signals to calculate their diagnostics. Knowing which signals carry the most weight helps you prioritise what to monitor and what to fix first.

Strongest signals (move diagnostics fastest):

  • Click-through rate: The single most responsive signal on both platforms. A CTR lift from a headline change can shift expected CTR status within days on Google and influence engagement ranking on Meta within the comparison window.
  • Negative feedback (hides, "Not Relevant," reports): On Meta, users who hide or report your ad send a strong negative signal that directly suppresses quality ranking. High frequency is the most common driver of negative feedback, creative fatigue, audience overlap, and overly broad ad groups are the usual culprits.

Medium-speed signals:

  • Engagement rate (reactions, comments, shares): Moves Meta's Engagement Rate Ranking over the 35-day comparison window. Consistent low engagement across an ad set suggests the hook or format isn't right for the audience.
  • Conversion rate and post-click behaviour: Bounce rate, time on page, and conversion completions feed both Meta's Conversion Rate Ranking and Google's landing page experience component. These signals take longer to accumulate because they depend on sufficient conversion volume.

Slower signals:

  • Landing page speed and content relevance: Google crawls landing pages and factors in both speed and topical relevance. Improvements here take days to weeks to register in Quality Score. For content relevance principles that apply directly to ad copy and landing page message-match, the underlying logic is consistent across both platforms.
  • Account-level historical performance: Google's Quality Score draws on the account's history for a keyword, not just recent campaigns. A new account or a new keyword starts without a score until enough impressions accumulate.
Signal Platform Speed of impact Tracking column
Click-through rate Google & Meta Fast (hours to days) CTR in both UIs
Negative feedback / hides Meta Fast (within comparison window) "Negative Feedback" in Ads Manager
Engagement rate Meta Medium (days to weeks) Engagement Rate Ranking
Conversion rate Google & Meta Medium to slow Conversion Rate Ranking / Landing Page Exp.
Landing page speed Google Slow (days to weeks) Landing Page Exp. component status
Creative freshness / frequency Meta Medium Frequency column in Ads Manager

A note on learning phase noise: During the learning phase (typically the first 50 optimisation events on Meta, or the first week of a new Google campaign), diagnostic signals are unstable. Avoid making structural changes during this window, you'll be reacting to noise, not signal. Wait for the learning phase to complete before reading diagnostics as reliable indicators.


A priority checklist for improving ad relevance step by step

Work through these fixes in order. Start with the highest-priority items, they move the most important signals fastest and protect your budget while you test.

  1. Audit diagnostic labels and identify the lowest-ranking signal (15 minutes). Pull Quality Score components or Meta diagnostics into your reporting view. Note every "Below Average" status. Prioritise any keyword or ad with two or more "Below Average" labels, these have the largest cost impact.

  2. Split broad ad groups into tighter thematic clusters (1-2 hours). If any Google ad group has keywords with different intents sharing one ad, split them now. This single fix addresses ad relevance component failures and often moves the status from "Below Average" to "Average" within a few days. On Meta, separate audiences with meaningfully different demographics or interests into distinct ad sets.

  3. Refresh creative for ads with Below Average quality or engagement ranking (1-3 days to test). Write at least two new headline variants and one new visual. Change one variable at a time so you can attribute the improvement. For Meta, prioritise the first three seconds of any video ad, that's where engagement is won or lost.

  4. Fix landing page message match and load speed (1-2 weeks to register in scores). Align the page H1 with the ad headline. Remove friction from the conversion path. Compress images and eliminate render-blocking scripts. For landing page best practices that apply to South African audiences and mobile-first traffic, page speed is especially critical given variable mobile network conditions.

  5. Resolve audience overlap and frequency issues on Meta (1-3 days). Use Audience Overlap in Meta's Audiences tool to identify overlapping ad sets. Consolidate or exclude audiences to reduce the same user seeing multiple ads from your account, this directly reduces negative feedback rates.

  6. Monitor diagnostics for 3-7 days after each fix before scaling (ongoing). Scaling a campaign before diagnostics stabilise locks in a bad signal at higher spend. Wait for at least one full comparison window on Meta (or 50+ conversion events) before increasing budget.

Testing guidance: Change one variable per test. If you change the headline, the visual, and the audience simultaneously, you won't know which fix moved the diagnostic. Run bulk variation testing only when you have enough volume to reach statistical significance within your test window.

Pro Tip: When a fix moves a diagnostic from "Below Average" to "Average," don't stop. "Average" means you're no longer penalised, but "Above Average" is where cost advantages compound. Keep iterating until you're consistently above the competitive benchmark.

Larger jumps can reset the learning phase and destabilise diagnostics you've already improved.


How long do changes take to affect scores?

Setting realistic timelines prevents the most common mistake in diagnostic-led optimisation: changing something, waiting 24 hours, seeing no movement, and changing something else. Most signals need time to accumulate before they register in scores.

Hand turning hourglass symbolising campaign timing

Change type Expected signal movement Timeframe
Ad copy / headline update Expected CTR (Google), Engagement Ranking (Meta) 48-72 hours for initial shift; stable after 7 days
Creative refresh (Meta) Quality Ranking, Engagement Rate Ranking 3-7 days within the 35-day comparison window
Landing page H1 and speed Landing Page Experience (Google), Conversion Ranking (Meta) 3-14 days
Ad group restructure Ad Relevance component (Google) 3-7 days after sufficient impressions
Audience tightening (Meta) Engagement Rate Ranking, Quality Ranking 5-10 days
Account-level trend improvement Overall Quality Score average 2-6 weeks

Metrics to track after each fix: CPM and CPC move first, often within 48-72 hours of a quality improvement on Meta. CTR shifts are visible within days on Google. Conversion rate and cost per action (CPA) take longer because they depend on conversion volume. Diagnostic label changes lag behind the underlying signal shifts by a few days.

Sample size matters: On Meta, diagnostics are unreliable below a few thousand impressions per ad. On Google, a keyword needs enough exact-match impressions before Quality Score stabilises. Prioritise fixes on high-volume keywords and ads where the data is sufficient to read the signal accurately.

Reporting cadence: Review diagnostics weekly, not daily. Daily checks during a test period create pressure to act on noise. A weekly cadence gives signals time to stabilise and gives you a cleaner before-and-after comparison. Set a minimum experiment length of 14 days for landing page changes and 7 days for copy changes before drawing conclusions.

  • Avoid pausing and restarting campaigns mid-test, this resets impression counts and delays diagnostic stabilisation.
  • Use Google Ads' "Experiment" feature or Meta's A/B test tool to isolate variables and measure impact cleanly.
  • Document every change with a date and the diagnostic status at the time of the change, this creates an audit trail that makes attribution straightforward.

Agency case example: diagnostics-led optimisation in practice

A retail client running Meta campaigns came to Juicydesigns with stagnant ROAS and rising CPMs. The initial diagnostic read was: Below Average quality ranking, Average engagement rate ranking, Below Average conversion rate ranking. Two of three diagnostics were flagging problems, which pointed to both a creative quality issue and a post-click problem, not an audience problem.

Hands sorting ad creative materials in studio

The diagnosis matrix applied: Below Average quality + Average engagement + Below Average conversion = creative needs refreshing AND the landing page is losing conversions. The engagement ranking being average ruled out an audience mismatch as the primary driver.

Interventions and timeline:

  • Week 1: New creative assets produced (clean visuals, no text overlay, benefit-led copy). Old ads paused.
  • Week 2: Landing page H1 updated to mirror ad headline; form reduced from seven fields to three; page load time cut from 4.8 seconds to 2.1 seconds on mobile.
  • Week 3: Diagnostics monitored without further changes to allow signals to stabilise.
  • Week 4: Quality ranking moved to Above Average; Conversion Rate Ranking moved to Average.

The outcome: CPM dropped materially within the comparison window, and CPA improved as the conversion ranking recovered. Industry audits of similar diagnostic-led fixes report CPM reductions in the 25-60% range when quality ranking moves from Below Average to Above Average in the same auction pool. These are observed industry ranges, not a guarantee for every account.

Juicydesigns clients across Google Ads and Meta campaigns achieve an average ROAS of 4.8x, with diagnostic-led optimisation being a consistent driver of that outcome.

Pro Tip: When two diagnostics are "Below Average," fix the creative first. Creative changes are faster to implement and test than landing page rebuilds, and a quality ranking improvement often lifts CPM within the first comparison window, giving you a faster signal that the fix is working before you invest in larger structural changes.


Common mistakes advertisers make with relevance scores

Most diagnostic errors come from misreading what the score is telling you, or from acting too fast on too little data.

  • Optimising the number, not the experience. Chasing a Quality Score of 10 or an "Above Average" label on every diagnostic is a distraction. The goal is lower CPC and better ROAS, not a perfect score. Google explicitly recommends using Quality Score as a diagnostic filter, not as a performance target.
  • Changing multiple variables at once. If you update the creative, restructure the ad group, and change the landing page in the same week, you won't know which fix moved the diagnostic. One change at a time is slower but produces learnings you can actually use.
  • Misattributing conversion problems to creative. A Below Average conversion ranking with Above Average quality and engagement is a landing page problem, not a creative problem. Refreshing the ad creative in this scenario wastes budget and delays the real fix.
  • Ignoring "Average" as a red flag. "Average" means you're not being penalised, but it doesn't mean you're winning. In competitive auctions, "Average" on conversion ranking can still mean you're losing placements to competitors with stronger post-click experiences.
  • Acting during the learning phase. Diagnostics during the learning phase are noisy. Making structural changes before 50 optimisation events (Meta) or sufficient impression volume (Google) means you're reacting to incomplete data. Common causes of unstable diagnostics include creative fatigue, audience overlap, and overly broad ad groups, but these are only readable after the learning phase completes.
  • Ignoring the three-diagnostic combination on Meta. Reading each Meta ranking in isolation leads to wrong fixes. Third-party analysis consistently shows that "Below Average (Bottom 35%/20%)" statuses have the largest cost impact and should be prioritised first, but the combination of all three tells you where to look.

The score is a symptom, not the disease

Here's the honest take on how most advertisers use relevance scores: they watch the number, feel good when it goes up, and panic when it drops. That's backwards.

The platforms designed these diagnostics to tell you where the user experience is breaking down. A Quality Score of 6 on a keyword that's generating strong ROAS at an acceptable CPC is not a problem. A Quality Score of 8 on a keyword that's burning budget with zero conversions is. The number is context-dependent, and optimising it in isolation is one of the most common ways advertisers waste testing cycles.

What actually matters is the component behind the score. A "Below Average" landing page experience on a high-spend keyword is worth fixing urgently because it's costing you on every click. A "Below Average" ad relevance on a low-volume keyword in a niche ad group might not be worth the restructuring effort if the keyword is converting well anyway.

The Meta diagnostic model is, in some ways, more honest than Google's 1-10 number. Three separate rankings force you to think about three separate problems. Advertisers who read them together and map them to a root cause before touching anything consistently outperform those who react to individual labels in isolation.

The other thing most guides won't tell you: "Average" is often good enough. In most auctions, you don't need to be "Above Average" on every diagnostic to win placements at a competitive cost. That's the threshold where costs spike and delivery suffers. Getting from "Below Average" to "Average" is the high-value move. Getting from "Average" to "Above Average" is a nice-to-have.

Spend your optimisation time where the diagnostic is red, not where it's yellow.


Sources

These are the canonical references for platform behaviour and tactical optimisation. Bookmark the platform docs, they update when platform mechanics change.


FAQ

What is an ad relevance score and how does it work?

An ad relevance score estimates how well your ad matches the intent and experience of the audience it targets. On Google, this is called Quality Score (1-10, keyword-level); on Meta, it's expressed as three comparative diagnostic rankings at the ad level.

How does Google Ads measure ad relevance?

Google measures ad relevance as one of three Quality Score components, showing a status of Above Average, Average, or Below Average based on how closely your ad copy matches the keyword's intent. The fix is almost always structural: tighten ad groups so each keyword maps to a specific, relevant ad.

How do you use Meta's ad relevance diagnostics?

In Ads Manager, add Quality Ranking, Engagement Rate Ranking, and Conversion Rate Ranking as columns under Customise Columns. Read all three together: a Below Average quality ranking points to creative issues, engagement ranking to audience or hook problems, and conversion ranking to landing page or offer issues.

How do you increase ad relevance on Google or Meta?

Match your ad copy tightly to the keyword or audience intent, split broad ad groups into tighter thematic clusters, and align your landing page headline with your ad message. These three fixes address the most common causes of Below Average statuses on both platforms.

Does a higher relevance score always mean lower costs?

Not automatically. Higher relevance improves your position in the auction, which can lower CPC and CPM, but the relationship is indirect. The goal is to eliminate "Below Average" statuses on high-spend keywords and ads, that's where the cost savings are most reliable.

Related reading: Improving your Google Ads Quality Score, What Google Ads costs in South Africa.

Cobus van der Westhuizen

Founder & Digital Strategist, Juicy Designs, Pretoria

Cobus founded Juicy Designs in 2015 and has spent over a decade marketing South African businesses across automotive, entertainment, professional services, retail and insurance. He personally oversees SEO strategy for Juicy Designs client accounts and reviews every article published on this site for factual accuracy and current market relevance.

  • Founder of Juicy Designs, established 2015
  • 64+ South African clients, 4.9-star Google rating
  • Google Ads certified practitioner
  • Google Analytics 4 certified
  • Specialist in SEO, paid media & conversion-focused web design
  • Reviewed and updated June 2026