A creative health score is a single number, usually on a 0 to 100 scale, that estimates how much useful life an ad creative has left. It compresses several live performance signals - click-through rate against its launch baseline, cost per thousand impressions (CPM), engagement, frequency and the number of days the asset has been running - into one rating, so that a media buyer can see at a glance which ads are still working and which are wearing out. The term exists because modern campaigns run dozens or hundreds of creative variants at once, and checking each one's trend lines by hand does not scale.

The phrase is not a standard. As of October 2026, none of the major ad platforms - Meta, Google, TikTok, Amazon Ads or Google's Display & Video 360 - ships a feature called "creative health score". The name lives in third-party creative analytics tools, agency dashboards and open-source agent playbooks, while the platforms offer adjacent scores and flags under their own names.

How a health score is built

The clearest public specification is a community playbook for Claude Code on claudecodehq.com, sourced from a GitHub repository called digital-marketing-pro; the page states it is not affiliated with Anthropic.

For every active creative, the tool collects impressions, frequency, current click-through rate (CTR), CPM or cost per click (CPC), engagement rate, days running and spend. It then establishes a baseline from the first 7 to 14 days of delivery. Each current metric is expressed as a ratio against that baseline, and the ratios are weighted:

  • CTR ratio, 30%
  • CPM ratio, 25%
  • Engagement ratio, 20%
  • Frequency or impression saturation, 15%
  • Time running relative to channel norms, 10%

The weighted result is a score where, according to the playbook, "100 is peak health and 0 is fully fatigued." Scores are bucketed into five fatigue stages: Fresh (80-100), Mature (60-79), Fatiguing (40-59), Fatigued (20-39) and Exhausted (0-19). Advertisers can supply their own triggers, such as a 30% CTR drop or a 25% CPM rise from baseline; otherwise the playbook says "industry-standard thresholds" apply, without listing them.

Consider a video ad that launched with a 1.2% CTR and an $8 CPM. Three weeks later its CTR is 0.84% (a ratio of 0.70), its CPM is $10 and frequency has climbed from 1.5 to 4. Each component is penalised by its weight, and the composite can fall into "Fatiguing" before cost per result has doubled. That early warning is the selling point.

Not every vendor points the scale the same way. Finsi's glossary defines a creative fatigue score as a 0-100 composite "where higher scores indicate greater fatigue," and treats 80 or above as the usual trigger to pause and replace. A health score of 20 and a fatigue score of 80 can describe the same ad. Anyone comparing tools needs to check the direction of the scale before reading the number.

Where it sits in the campaign workflow

A health score is an in-flight metric. It exists only once an ad has delivered enough impressions to set a baseline, and it sits downstream of the auction: the score reads outcomes but does not feed the bid. Performance agencies, in-house growth teams and creative strategists operate it on the buy side. The platforms run their own equivalents.

Meta's Ads Manager, for example, can label an ad "Creative Fatigue" or "Creative Limited" in its Delivery column. According to Jon Loomer, the first label appears when cost per result is at least twice its past level and the second when it is elevated but below double; he notes that these are simplified readings and that Meta does not fully explain how it ties a decline to repeated exposure. Third-party health scores are, in effect, attempts to see that deterioration earlier and across more signals than cost per result alone.

Pre-flight scoring, which rates an asset before money is spent, is often marketed alongside health scores but measures predicted quality rather than observed decline.

Origin and evolution

Platform-computed creative scores predate the vendor term by roughly a decade. Facebook introduced a relevance score on a 1 to 10 scale in February 2015, built from positive and negative audience feedback such as hides and reports. Jerry Daykin, then global digital and client director at Dentsu Aegis Network, told The Drum at the time that "it remains to be seen how this new score relates back to sales and brand metrics." In March 2019 Facebook said it would retire the single score by April 30, 2019, replacing it with three ad relevance diagnostics: quality ranking, engagement rate ranking and conversion rate ranking, each comparing an ad with competitors for the same audience, according to PMG.

Google took a different route. When responsive search ads arrived in August 2018, Google added Ad Strength, a pre-flight rating of the relevance, quantity and variety of an ad's headlines and descriptions, with a reporting column from early September 2018, according to Search Engine Land. Today the Google Ads Help Center describes ratings from "Incomplete" through "Poor", "Average" and "Good" to "Excellent", and says the rating "isn't used to calculate Ad Rank, Quality Score, or auction wins." Individual assets also receive post-launch performance ratings of Learning, Low, Good or Best, which Google's November 2025 responsive search ad guide tied to thresholds of 500 asset impressions and 2,000 ad impressions over 30 days.

Independent pre-flight scoring followed. CreativeX built its Creative Quality Score (CQS) around "digital suitability fundamentals" such as early branding, mobile framing, sound-off design and safe zones, mapped to platform guidance like Meta's Brilliant Basics and Google's YouTube ABCD framework. The company says about 70% of the average Fortune 500 brand's digital budget goes to low-CQS content.

The named in-flight health score is the most recent layer, emerging as automation pushed advertisers toward larger creative pools. Meta added a Creative breakdown on July 11, 2025, exposing cost per result and delivery status for each asset inside a flexible-format ad - the raw material any health score needs.

Why it matters for marketers

The case for health scoring rests on a measurement gap. An EMARKETER survey with Perion, fielded in February 2026 among 111 US marketers each managing more than $5 million in media, found that 72.1% wait for a performance decline before changing creative and only 11.7% use AI-driven performance signals to trigger a refresh, as PPC Land reported in April. Perion's chief executive Tal Jacobson called the prevailing approach "not optimization; it is a post-mortem." The survey was commissioned by a company that sells the remedy, which the coverage noted.

The stakes are visible in spend allocation. A Kroger, Vidmob and MMA Global study of 1,934 ads found that about half of historical Meta and DV360 spend went to the lower-scoring tier, with the model classifying unseen ads as high or low converters with 81% accuracy.

Limitations and disputes

The first problem is the lack of a definition. Weights, baselines and stage boundaries differ by vendor and are rarely validated publicly, so a 55 in one tool is not comparable with a 55 in another.

The second is that the baseline is a moving target. Early delivery is often unrepresentative because platforms are still in the learning phase, auction prices change seasonally, and a CPM rise in November may reflect competition rather than fatigue. A ratio against week one cannot separate those causes.

Third, platform scores that look authoritative have a mixed record. Google's help page says improving Ad Strength from Poor to Excellent yields 15% more conversions on average, citing internal data from August 15 to 20, 2025. Optmyzr's September 2024 study of more than one million ads across 22,000 accounts reached the opposite conclusion: responsive search ads rated "Average" had the best cost per acquisition, conversion rate and return on ad spend, and the company found "no clear correlation between ad performance and Ad Strength." Optmyzr's sample skews towards advanced users of its own software.

Fourth, predictive scoring remains largely unproven in controlled tests. The Kroger findings are observational, and the planned in-market test pitting high-scoring against low-scoring creative had not run when they were published. Bannerflow's attention-prediction stage, launched in October 2026, came without validation data or pricing. And when the platform's system decides delivery, a falling score may reflect the algorithm shifting spend rather than the audience tiring.

What it is not

  • Creative fatigue is the underlying phenomenon: performance decay from repeated exposure. A health score is one way of measuring it, and Meta's "Creative Fatigue" status is another.
  • Ad Strength is Google's pre-flight rating of asset completeness and relevance. It does not track decline over time and does not enter the auction.
  • Ad quality ranking is one of Meta's three relevance diagnostics, comparing perceived quality with competing ads for the same audience rather than with the ad's own past.
  • Creative quality score usually means a pre-flight compliance score such as CreativeX's CQS, which audits craft attributes before launch.

Recent developments

The platform layer is moving closer to the vendor concept. In late August 2026 Meta began showing a Creative Diversity column at ad set level, rating the variety of active ads as Low, Medium or High and labelled "estimated and in development," according to Common Thread Collective and MarketingConcurrent. Meta has not published documentation of the method. The opportunity score, launched globally in June 2025, already rates campaign optimisation from 0 to 100, though it scores setup rather than individual creatives.

Pre-flight vendors keep adding prediction, with Smartly's Creative Predictive Potential in November 2025 using computer vision and eye-tracking models. The spread of agent connectors to ad accounts, including Meta Ads AI Connectors in open beta since April 29, 2026, means scripts like the Claude Code playbook can compute health scores directly from live account data. The metric remains unstandardised, but its inputs have never been easier to reach.

Timeline

  • February 2015 - Facebook introduces a 1-10 relevance score for ads.
  • August 2018 - Google launches responsive search ads and the Ad Strength indicator; the reporting column follows in early September.
  • March 19, 2019 - Facebook announces it will replace relevance score with three ad relevance diagnostics by April 30, 2019.
  • September 9, 2024 - Optmyzr publishes its study finding no clear correlation between Ad Strength and performance.
  • June 2025 - Meta makes opportunity score available globally.
  • July 11, 2025 - Meta introduces Creative breakdown for flexible formats and AI-generated images.
  • August 15-20, 2025 - Google's internal data window behind its 15% Ad Strength conversion claim.
  • November 18, 2025 - Smartly announces Creative Predictive Potential.
  • February 2026 - EMARKETER and Perion field the creative optimisation survey.
  • April 29, 2026 - Meta Ads AI Connectors enter open beta.
  • Late August 2026 - Meta's Creative Diversity column appears in Ads Manager.
  • September 30, 2026 - Kroger, Vidmob and MMA Global publish their creative scoring research.
  • October 12, 2026 - Bannerflow launches pre-flight attention scoring.

Summary

Who: Third-party creative analytics vendors, agencies and agent-tool authors use the term; Meta, Google and creative-scoring firms such as CreativeX, Vidmob, Smartly and Bannerflow operate the adjacent platform and pre-flight scores.

What: A non-standard 0-100 composite that rates how close a live ad is to fatigue, built from ratios of current CTR, CPM, engagement, frequency and age against a launch baseline.

When: Platform creative scores date to Facebook's 2015 relevance score and Google's 2018 Ad Strength; the named health score is a 2020s vendor convention, with Meta's Creative Diversity column added in August 2026.

Where: In creative analytics dashboards and agent scripts that read platform reporting, chiefly for Meta, TikTok and Google campaigns, sitting after launch and outside the auction.

Why: Large creative pools and automated delivery make manual monitoring impractical, and most marketers still refresh creative only after performance has already fallen.