Work with me
Want to review creatives for a client campaign?
Get in touch. I work with a small number of clients on paid media strategy, creative direction, and performance audits.
Get in touch →Upload paid ad creatives — images or video. Claude AI scores each one across six dimensions and returns a ranked report with specific weak signal flags tiered by impact: what to pause, what to iterate, what to scale. Works across Meta, Google, and TikTok formats. No account connection required.
Every creative is scored against the same rubric every time. Video ads include audio transcript extraction — so the scoring covers what the viewer hears, not just what they see. Weak signals are tiered by impact: what to pause, iterate, or scale.
Each creative gets scored 1–10 on each dimension. Anything 6 or below flags as a weak signal with specific notes. For video, the composite is role-weighted: each dimension is drawn from the frame where it matters most.
Does it stop the scroll and make the viewer feel something? Scores both pattern interrupt and emotional resonance. Interrupting without connecting caps at 6.
Is there a clear focal point? Is the reading order obvious without effort?
Is the core message understood in under two seconds? For video, also scores sound-off legibility — whether the message lands with audio muted. A 1–3 means the creative depends entirely on voiceover.
Coherent colours, fonts, and tone. Does it read as a brand or as noise?
Absence of overused formats: talking-head UGC with no hook, plain white product shots, generic CTAs, logo-first layouts. A high score means the format is distinctive and low-fatigue-risk.
Does the proposition, value, and next step land clearly by the end of the creative? Scored on the closing frame for video — the moment a viewer should know exactly what they're being asked to do.
Three frames extracted per video — opening, midpoint, closing — plus an audio transcript via Whisper. Scoring covers what the viewer hears, not just what they see. The composite is role-weighted across frames.
Scores are AI-generated using a consistent rubric. Reproducible and documented — but should be used alongside campaign performance data, not as a replacement.
The rubric applies the same logic to Meta, Google, and TikTok creatives. Platform-specific attention patterns are not currently scored separately — that’s a future scope item.
Not a real-time audience measurement tool. It scores creative signals, not ad performance. A creative can score well here and still underperform if targeting is the problem.
Most creative review processes are subjective. A senior marketer looks at a set of ads and gives a verdict based on experience and instinct. That works — until you’re running 30 variants across four campaigns and need a consistent framework for prioritising which creatives to pause, which to iterate on, and which to scale.
This tool applies a consistent six-dimension rubric to every creative you upload. Each image or video frame is scored on hook quality, visual hierarchy, copy clarity, brand consistency, originality, and offer clarity. For video, an audio transcript is extracted and used alongside the visual frames — so the scoring reflects what the viewer hears, not just what they see. The composite gives you a ranked view across a batch. The weak signal flags — tiered by impact — tell you exactly what to fix.
Creative quality also feeds directly into paid media efficiency. A creative that fails to hold attention drives up your effective CPM and frequency thresholds — you end up spending more to reach the same number of people before results decay. Identifying the specific weak signal (a low-quality hook, an unclear headline, a fatigued format) is the difference between a brief that fixes the problem and one that just asks for something “more engaging”.
What makes a paid creative fail? Not production quality. It’s stopping the scroll, communicating the offer in under two seconds, and avoiding the formats audiences have learned to ignore. The tool is built around those three problems.
Creative fatigue describes the point at which an audience has seen a format, style, or message often enough that engagement drops — not because the product is less appealing, but because the creative has lost its ability to interrupt. Common signals include generic UGC formats with no hook, logo-first compositions, plain white product shots without context, and overused CTA phrases like ‘Shop now’ or ‘Learn more’.
A manual review depends on the reviewer’s frame of reference and changes between sessions. An AI-scored rubric applies the same criteria every time, returns a documented rationale for each score, and flags specific weak signals rather than general impressions. The output is the same whether you score 5 creatives or 50.
The weak signal flags are designed to feed directly into briefs. If Hook Quality scores 4 with a note about ‘no pattern interrupt in the first frame’, that’s a specific instruction for the next creative — not a general note about ‘making it more engaging’.
Check which dimension is pulling the composite down. A low Hook Quality score points to the opening frame. A low Copy Clarity score points to the headline or body copy — or, for video, a lack of captions. A low Originality & Fatigue Risk score means the format itself may need to change, not just the copy or image. A low Offer Clarity score means the proposition isn’t landing by the close.
Bidding, structure, conversions, audiences and six other areas — severity-ranked, in under 60 seconds.
Free Google Ads Audit → Landing pageCore Web Vitals, message match, conversion architecture, trust signals. Real Lighthouse, not estimates.
Free Landing Page Audit →Conversion tracking issues distorting your performance data? Read the guide.
Fix Google Ads Conversion Tracking →Get in touch. I work with a small number of clients on paid media strategy, creative direction, and performance audits.
Get in touch →Creative is one lever. The account and the landing page are the other two.
Audit your Google Ads →