Finding winning angles before you spend

An angle is a claim, not a format

Teams routinely confuse the two. A talking-head UGC video is a format. "You will get approved before your coffee gets cold" is an angle. Formats affect delivery cost; angles affect whether anyone cares. When a test result is filed under the format, the learning is lost — you know that UGC beat static, but not why, so nothing transfers to the next batch.

Write angles as full sentences that could be argued with. If a stranger cannot disagree with your angle, it is not an angle — it is a category description.

Where angles actually come from

Not from a brainstorm. Four sources produce almost everything worth testing, and all four are available before you spend a euro.

1. Your own support tickets and sales calls. The objection that comes up most often in month one of ownership is the objection your ads should be answering.
2. Cancellation and refund reasons. People are unusually honest when leaving. Each reason is an angle inverted — solve it in the ad and you pre-filter the cohort.
3. Competitor reviews, especially three-star ones. Five-star reviews are marketing. One-star reviews are anger. Three-star reviews contain the trade-off the market is actually making.
4. Search and store queries. The words people use when nobody is watching are the words your hook should use.

A week of review mining routinely produces twenty angles. Three of them will carry the account for a year.

Scoring before testing

Twenty angles is too many to test properly at once — you would split the budget until nothing reaches significance. Score them on three axes and test the top handful first.

• Reach — how many people in the market plausibly hold this belief? A true angle for two percent of the market cannot carry scale.
• Differentiation — can a competitor say the same sentence honestly? If yes, you are funding the category, not the brand.
• Provability — can the claim be demonstrated inside the first five seconds? Angles that require trust you have not earned yet convert badly cold.

One variable, one answer

When the shortlist goes into production, keep everything except the claim as constant as the format allows. Same length, same pacing, same call to action, same landing page. This feels wasteful — you could be testing five things at once — but a test that changes three variables produces a winner you cannot explain, and an unexplainable winner cannot be extended.

The point of angle testing is not to find one ad. It is to find the argument that works, so that the next twelve ads can be built from it with confidence.

Angles expire; the map does not

Markets move. An angle built on a price advantage dies the week a competitor matches it. Keep the angle map as a living document with a status against each entry — untested, validated, scaled, exhausted, blocked by compliance. After two quarters, that document is more valuable than the ad account itself, because it holds every conclusion you paid for.