Turn a brief into three testable angles and check the claims
What you'll make
Three distinct ad angles, each with a hook and supporting line, and a claim log showing what you verified and how.
What you need
- Any AI writing assistant (for example ChatGPT)
- The Page Turner Books one-page brief (Practice Projects)
- 25 minutes
The idea
An angle is the reason you lead with. The same product can be sold on surprise, value, convenience or belonging. A good test compares angles, not just different wordings of the same angle.
AI writing tools are fast at generating options, but they can produce confident statements that aren't true — invented statistics, made-up awards, guarantees the business never offered. Anything that states a fact about the business, the product, a number or a competitor is a claim, and it needs checking against a source you trust before it goes public.
Only paste information you're allowed to share into public AI tools. The brief here is fictional, so it's safe to use.
Worked example
Copy this prompt into an AI writing assistant, with the brief pasted below it:
- Paste this prompt: “You are helping plan an ad test. From the brief below, write three different angles for the Mystery Book Box. Each angle must have a different core reason to buy. For each, give: angle name, one-line hook (max 10 words), one supporting sentence. Do not add facts, numbers or claims that are not in the brief. If you need a fact that isn't there, write [CHECK].”
- Read the output. Check that the three angles are actually different reasons (for example surprise, value and gifting) and not three wordings of one idea.
- Highlight every claim: numbers, prices, delivery promises, comparisons, “best”, “most loved”, anything about customers.
- Create a claim log with four columns: claim, where it came from, how you checked it, keep/change/remove.
- Check each claim against the brief. Anything not in the brief is removed or rewritten — even if it sounds plausible.
- Choose two of the three angles to test, keeping the offer and audience the same (see the Performance Marketing lesson).
Expected result
Example output after review (fictional):
| Angle | Hook | Supporting line | Claims checked |
|---|---|---|---|
| Surprise | You won't know the book until it arrives | A hand-picked mystery novel, wrapped by our staff | “hand-picked”, “wrapped by staff” — both in brief ✔ |
| Value | Three books for ₹799 | One mystery, one classic, one staff pick | price and contents — in brief ✔ |
| Gifting | The gift for someone who has everything | Add a handwritten note at checkout | “handwritten note” — in brief ✔. Removed AI line “loved by 5,000 readers” — not in brief ✘ |
Common mistake
Trusting a specific-sounding number because it looks professional (“rated 4.9 by 2,000 readers”).
How to fix it: Ask: where would I show someone proof of this? If you can't point to the source, remove it. Specific-sounding numbers need the most checking, not the least.
Practice task
Run the same prompt in a second AI tool or a fresh chat. Compare the two sets of angles: which ideas appear in both, and which claims appear in only one? Add any new claims to your log and check them.
Self-check
The AI wrote “Free delivery across India in 2 days.” The brief says “free delivery within the city”. What do you do?
Show the answer
Change it to match the brief — “Free delivery within the city” — and log it. The AI expanded the promise; publishing it would make an offer the business doesn't make.
Hinglish recap
AI se ideas lo, par har claim khud check karo. Jo number ya promise brief mein nahi hai, use hata do — chahe kitna bhi professional lage.
More lessons for AI for Digital Marketing are on the way — the study plan shows what to learn next and links to free resources in the meantime.