Deciding what to make
Without a creative score
The angle comes from whoever spoke last in the meeting. Nobody can say which customer objection it answers, so there is no way to tell a bad idea from a good one until money has been spent on it.
Creative score
Know Your Creative Score Before You Spend on Ads
A creative score is a rating of how likely a piece of creative is to work. Selzee scores yours against your own customers: every angle traces to a ranked objection pulled from your reviews and comments, one hook per concept is marked to test first, and the asset itself comes back broken down into hook, on-screen text, layout and pacing. Once it has spent, the same ad is graded against your CPA and ROAS targets. Built for ecommerce growth and creative teams who want the weak ad caught at the brief.
The short version
- Two different things are sold as a creative score: a prediction made before launch, and a grade measured after. They are not equally reliable.
- Selzee does not predict click-through or conversion rate from an unpublished file. Nobody has solved that reliably.
- Before launch you get the argument scored against ranked customer objections, a test-first hook pick, and a structural read of the asset.
- After it spends, the ad is graded against the CPA and ROAS thresholds written into the brief, and the verdict feeds the next one.
Without Creative Scoring vs With Creative Scoring
The cost of no scoring is not that bad ads get made. It is that nobody can say why one worked, so the next brief starts from zero. Three stages where that shows up.
Deciding what to make
With a creative score
Every angle traces to a ranked objection pulled from your own reviews and comments, and one hook per concept is marked to test first with the reason attached. Weak ideas get caught at the brief, which is the only stage where fixing them is free.
Reviewing the asset
Without a creative score
Review happens by opinion in a comments thread. The loudest note wins, feedback contradicts itself between rounds, and the thing everyone actually disagreed about was never written down.
Reviewing the asset
With a creative score
Upload the image or video and get a structural read back: the hook, the on-screen text transcribed verbatim, the layout, the focal point, the format, and for video the shot list and pacing. Now the argument is about something recorded rather than remembered.
After it runs
Without a creative score
Results are read once a month, ads are paused on a hunch, and the reason a winner won is never written down. Month six looks a lot like month one because nothing compounds.
After it runs
With a creative score
Tracked ads are graded against your own CPA and ROAS targets with explicit win and kill thresholds set before launch, plus fatigue signals. The verdict names the angle that won, so the next brief starts from evidence.
How Creative Scoring Works
-
1
Connect the data you already own
Selzee reads your reviews, comments, product data and campaign performance through the tools you already run: Shopify, Meta Ads, TikTok Ads, Klaviyo, Google Ads and review platforms including Trustpilot, Judge.me and Yotpo. Nothing gets uploaded by hand and no research doc is started from scratch.
-
2
Score the argument before anything is designed
Objections and desires are extracted from customer language and ranked by how often they actually block or drive a purchase. Concepts are built against those, so the score at this stage answers one question: is this ad arguing something your customers care about, or something the team assumed they did?
-
3
Pick which hook goes first
Each concept carries three hooks and exactly one is marked to test first, with the rationale written out. That single pick matters more than the rest of the setup, because most of a test budget lands on the opening.
-
4
Read the asset itself
Drop in the finished image or video and get its structure back: the hook, on-screen text read verbatim, layout, focal point, visual style, call to action and format. Video adds a transcript, shot list and pacing. It is a factual breakdown of what the ad does, not an aesthetic grade.
-
5
Write the win and kill threshold into the brief
Before spending starts, the brief records what success looks like: beat this CPA by this margin, within this window, at this spend. Pre-registering the threshold is what stops the post-hoc arguing about whether a test worked.
-
6
Grade it once it has spent
Tracked ads are graded against your CPA and ROAS targets and returned as a verdict rather than a dashboard: won, lost, or still running. Fatigue shows up as rising frequency against falling click-through. Those verdicts feed the next round of briefs.
What Selzee Will Not Tell You
One honest limit, because this search result is full of the opposite claim. Selzee does not score an unpublished ad for predicted conversion rate or brand recall. Predicting the performance of creative that has never run, from a file alone, is not a problem anyone has solved reliably, and a confident number there is a sales feature rather than a measurement. There is no heatmap and no attention map either, because a generated overlay implies a precision that eye-tracking hardware earns and an image model does not.
Nor does Selzee grade aesthetics. Scoring how polished an ad looks measures adherence to a template, and templates are exactly what stops an ad sounding like it came from your customers. What is left is more defensible and more useful: score the argument against evidence before you build, read the asset factually, and grade the result against real spend after. That is a feedback loop rather than a forecast, and it is the version that gets better every cycle.
Where Scoring Saves Money, Time and Wasted Spend
Creative is where the remaining leverage sits, because platform automation has taken most of the delivery decisions off the table. The gains below all come from deciding better, not from spending more.
Profit without a bigger budget
The gain comes from killing losers earlier and scaling winners sooner, not from spending more. Creative is where that leverage sits now that platform automation has absorbed most delivery decisions, which is why the same budget produces a different result when the argument improves.
Hours back from manual review mining
Reading every review, comment and survey answer for recurring objections is a full day of work per product line, repeated whenever a new angle is needed. Doing that continuously is the single biggest time saving here, and it is the part teams most often skip when they are busy.
Less spend on creative that was never going to work
An ad arguing something no customer mentioned will lose slowly and expensively, and its failure teaches you nothing because the concept was untested rather than wrong. Catching that at the brief costs nothing.
Volume that stays coherent
Scaling output usually means quality drifts, because more people make more decisions from less shared context. When every concept traces to a ranked objection with the customer quote attached, ten concepts a week stay as sharp as two.
Why Creative Carries the Result Now
Nielsen's creative-effectiveness research credits creative, not targeting or media spend, with up to 89% of a digital ad's in-market success, and Meta's own research reaches the same conclusion about high-quality creative increasing ad ROI. The cost of getting it wrong keeps climbing, with Facebook cost per lead up almost 21% year over year in 2025, per WordStream's benchmarks. That is the case for scoring creative at the point it is cheapest to change, which is before anybody opens a design file.
The input that makes a score meaningful is customer evidence, not a model's taste. Sarah Levinger, Founder of Tether Insights, names the habit most teams are missing: “I wish more brands would use the data they have and start analyzing for emotional insights instead of just transactions.” Scoring an ad against a template tells you it looks like other ads. Scoring it against your reviews tells you whether it answers a reason somebody did not buy.
Score Creative Across Every Channel
Facebook and Instagram ads
The heaviest use, because Meta is where most ecommerce spend lands and where creative fatigue arrives fastest. Static, carousel and video all get the same structural read, and tracked ads come back graded against your CPA and ROAS targets once they have spent.
TikTok and short-form video
Video analysis returns the transcript, the hook, the shot list and the pacing, which is where short-form is usually won or lost. TikTok Ads connects directly, so performance verdicts run on the same footing as Meta.
YouTube ads and video hooks
The opening seconds carry the result, and the same hook read applies: what is said, when the product appears, how fast the cuts land. A hook that works on Reels often needs a different runway here, and the breakdown makes that visible before you re-cut.
Product listings and marketplace images
Listing images are ads with less room. The on-screen text is transcribed verbatim and the focal point named, which catches the common failure of a hero image that looks good and communicates nothing at thumbnail size.
Email, landing pages and organic
The scored layer is the argument, and arguments travel. An objection that blocks a purchase blocks it in a subject line, above the fold, and in a caption. The format changes, the ranked objection does not.
Want to score something right now without booking anything? The free ad analysis tool rates pasted ad copy from 0 to 100 and shows its working, and the ads reporting tool scores every creative in a Meta or TikTok CSV export against your own account averages. For the post-launch side in depth, creative diagnostics covers win, hold and kill thresholds, and creative tracking covers keeping the verdicts attached to the right ads.
The fastest way to see this is on your own products. Book a session and we will score a live concept against your reviews and campaign data, and show you the verdict format your tracked ads come back in.
Request a DemoWhat Changes When Scoring Replaces Opinion
Selzee is used by performance marketers, creative strategists, growth managers and founders running lean ecommerce teams. The pattern they describe is consistent: research that used to take a day happens in a session, and briefs go out with the customer quote attached to every claim. We are gathering named customer stories and will publish click-through, conversion and ROAS figures as those results are verified, rather than quoting numbers we cannot yet stand behind. You will find no invented lift percentages on this page for the same reason.
-
Creative review stops being an opinion thread, because the asset read is on the table before anyone gives a note.
-
Fewer rounds spent discovering that the direction was the problem rather than the execution.
-
An angle bank that grows, because every verdict names the argument that won rather than just the ad that won.
"A score that predicts how an unpublished ad will perform is the one thing everybody wants and nobody can honestly sell. The version that works is duller: write down what you expect before you spend, then let the result settle it."
Creative Scoring FAQ
01 What is a creative score? +
A creative score is a rating of how likely a piece of creative is to work. The term covers two very different things: pre-launch scores that predict performance from the asset alone, and post-launch scores that grade an ad against its own targets once it has spent. Selzee does the second, plus a structural read of the asset before launch.
02 Does Selzee predict how an ad will perform before it runs? +
No. Selzee does not score an unpublished ad for predicted click-through rate, conversion rate or brand recall. Predicting performance from a file alone is not a problem anyone has solved reliably, and a confident number there is a sales feature rather than a measurement. What you get before launch is the argument scored against ranked customer objections, a test-first hook pick, and a structural breakdown of the asset.
03 How does the scoring actually work? +
Three layers. Objections and desires are extracted from your reviews and comments and ranked by frequency and purchase impact, and concepts are scored against those. Uploaded assets are broken down structurally: hook, on-screen text, layout, focal point, format, and for video the transcript, shot list and pacing. Tracked ads are then graded against the CPA and ROAS thresholds recorded in the brief, using deterministic rules rather than a model guess.
04 Do I get a heatmap or attention map? +
No. There is no heatmap, eye-tracking overlay or saliency map. The asset read names the focal point and describes the layout in plain language, which is the useful part of that idea without the false precision of a generated attention map.
05 Does it analyse the copy as well as the visuals? +
Yes, and copy is the part it reads hardest. On-screen text is transcribed verbatim rather than paraphrased, video returns a full transcript, and the hook is separated out because it carries most of the result. Selzee also has a free hook scoring tool that rates pasted ad copy from 0 to 100 with the reasoning shown.
06 Which formats and channels can it score? +
Static images, carousels, short-form vertical video, longer video and marketplace listing images, across Meta, TikTok, YouTube, Google and email or landing page copy. Performance grading needs a connected ad account and enough spend to be meaningful; the asset read and the argument scoring work on anything.
07 How accurate is it? +
The post-launch grade is arithmetic, not a prediction: it compares real spend, cost per acquisition and return on ad spend against thresholds you set, so accuracy is a question of whether your tracking is clean. The asset read is a factual transcription and description. Neither claims to forecast, which is precisely why neither can be wrong in the way a predicted score can.
08 Can I use it on things that are not ads? +
Yes. The ranked objections and customer language apply to email subject lines, landing page headlines, product descriptions and organic captions. Performance grading is ads-only, because it needs platform spend data to grade against.
09 How does this improve ROI over time? +
Because the verdicts compound. Each graded ad names the angle that won or lost, which becomes evidence for the next brief instead of being forgotten. Teams without that loop restate the same three angles for a year; teams with it build an angle bank, which is what stops month six looking like month one.
Start Scoring Your Creative Today
Stop finding out which ad was weak by paying for the impressions. Selzee scores the argument against your own customers before anything is built, reads the finished asset back to you in structure rather than adjectives, and grades the ad against your CPA and ROAS targets once it has spent. Book a session and we will run it on your products so you can see all three on your own data.
Request a DemoPrefer to explore on your own? Get started or browse free ecommerce tools.
Keep exploring: the free ad analysis tool, the ads reporting tool, creative tracking, Facebook ad fatigue, market testing for ads.