The method
How the map reads ads, and where it stops
How the map reads your ads
- Pick your category and the customer decision the ads should win: a first order, a switch or a repeat order.
- Rank up to three buying questions your ads must answer. They are your priorities, not research.
- Paste your ads. Each one is read for cue words, and every label shows the words it came from. You can change any label.
- The map counts reasons to buy, groups look-alike ads into ideas and names the priorities no ad answers, with a direction for each.
The worked example, a hypothetical pet-food brand
Six ads: four say what is in the food, as a video, a static, a UGC clip and a carousel. One is a subscription offer and one names the picky eater. The map reads six ads and three reasons to buy. Two of three first-order priorities, proof and the first box, have no ad.
The eight buying questions
The map sorts every ad by the question it answers for the customer. The wording changes with the category; the eight questions stay the same.
- Is it for me?
- The person and the problem. Answered by naming who it is for in the first seconds.
- Does it work?
- The result. Answered by real footage or documented results, never a generated reaction.
- What is in it?
- Ingredients or materials. Answered by close, readable detail from the real pack or spec.
- How does it fit my day?
- Use and convenience. Answered by the real moment of use, at real speed.
- Is it worth it?
- Price and value. Answered by honest maths you can document on the day the ad runs.
- Can I trust it?
- Reviews, credentials and guarantees. Answered by real customers and sources you may publish.
- What arrives?
- The first order. Answered by opening the real box and showing what is in it.
- Why switch?
- The comparison. Answered by a fair, documented difference from what the buyer uses now.
Why count reasons to buy, not formats
Meta's public guidance to advertisers is to “diversify your creative, test, and let the system optimize”. The diversity a buyer notices is a different reason to buy, not the same message in a new format. Meta spokesperson, Marketing Brew, 28 September 2026
A new crop, caption or voice-over on the same argument is a variant. It keeps a working ad fresh, but it asks the buyer the same question. A new concept answers a different one, such as what arrives in the first box or why to switch.
The map shows which of the two you have been making. For planning concepts and variants together, read the creative testing matrix guide.
What the map does not do
- It does not predict results, score ads or compare you with other brands.
- It reads the words you give it, not your videos or images, so describe what an ad shows when the text alone does not say.
- It cannot see your account, spend or results. Concentration on one question can be the right choice for a launch or a single offer.
- Cue words miss things. An ad with no clear cue stays unknown until you label it.
Questions about creative diversity
What is creative diversity in ads?
Creative diversity means your live ads give buyers different reasons to buy, not one reason in many formats. The map measures it plainly: how many of eight buying questions your ads answer, and how many separate ideas sit behind them.
How many different concepts should we run?
No published number fits every account, and Meta's public guidance does not give one. Start from your priorities: each buying question you ranked should have at least one concept, and each concept a few variants to test.
What is the difference between a concept and a variant?
A concept answers a buying question in a particular way. A variant changes the opening, caption, crop or format and keeps the same answer. The map counts ads that share a buying question and the same evidence or wording as one idea.
Does the tool see our ad account or files?
No. You paste or type the ads, and the text stays in your browser. There is no login, no ad-account connection and no upload. A public result card holds counts only.
Is this a performance score?
No. It shows coverage, not results. It cannot tell you which ad will win, and an ad that repeats a strong argument can still be the right ad. Use it to decide what to brief next, then test it.
For AI agents and developers
POST JSON with category, decision, priorities and up to 12 ads to /api/tools/blind-spot-map and get the same map back: labels with the cue words behind them, ideas, blind spots and directions. Version blind-spot-map/1.
The endpoint evaluates only. It stores nothing, sends nothing and creates no enquiry. A GET on the same URL describes the input and output.