A folder called “good ads” tends to become a museum. Interesting things enter. Very few useful experiments leave.
Give each saved creative a short research note instead. The note should explain what you think the ad is doing and how you could test the idea without copying its execution.
Six fields worth keeping
Audience: Who appears to have the problem? Be specific about the situation, not just age or occupation.
Trigger: What moment makes the problem relevant now?
Promise: What outcome does the creative suggest?
Mechanism: What does the product visibly do to produce that outcome?
Proof: Is there a real demonstration, a testimonial, a claim or merely attractive editing?
Next step: What is the viewer asked to do, and where does the destination take them?
Keep source URL, advertiser identity, observation date and any visible market filter beside these fields. If a detail is not available, do not fill it with inference disguised as metadata.
Add the skeptical line
Write one sentence explaining why the creative might attract the wrong customer. A spectacular transformation can bring curiosity from people who have no need for the ongoing product. A clever joke can hide what the app actually does.
Then inspect whether the landing page and onboarding continue the same promise. A clear ad followed by a confusing first session is not a coherent acquisition path.
Turn one observation into one test
For a hypothetical meal-planning product, you might test “use what is already in the fridge” against “plan the week’s meals.” Keep the audience and destination comparable so the result can teach you something about the message.
Use your own footage, copy and product evidence. The competitor’s ad supplied a research question, not a reusable creative licence.
A good research note ends with an original, measurable experiment. If it ends with “make something like this,” take the ad apart one more time.