# Ad testing hypothesis template

> A copy-and-paste ad testing hypothesis template: one belief, one change, one metric and a decision rule set before you spend. Includes a filled-in example.
- **Author**: Marxx
- **Published**: 2026-11-01
- **Category**: Planning
- **URL**: https://marxx.ai/library-templates/ad-testing-hypothesis-template

---

An ad testing hypothesis is one short plan that says what you will change, what you expect to happen and why. Write it before you spend, and write the rule you will use to call the result.

Copy this into a doc or sheet. Fill in one per test.

## The template

```
AD TEST: [ID, e.g. T-014]
Dates: [start] to [end]

1. BELIEF
We think [audience] cares most about [reason]
because [evidence: review, comment, sales call, past ad].

2. CHANGE
We will change ONE thing: [hook / angle / offer / format / first frame].
Everything else stays the same.

3. PREDICTION
If we do this, [primary metric] will [rise / fall]
from [current number] to [target number].

4. METRICS
Primary: [hook rate / CTR / cost per purchase / ROAS]
Guardrail (must not get worse): [metric and limit]

5. BUDGET AND TIME
INR [amount] per version per day, for [days] days.

6. DECISION RULE (written before launch)
Scale if: [ ]
Stop if: [ ]
Retest if: [ ]

7. RESULT (fill in after)
[numbers]

8. WHAT WE LEARNED
[one sentence]
```

## A filled-in example

Say you sell hair oil at INR 549. Your best video has a [hook rate](/glossary/hook-rate) of 24% and a cost per purchase of INR 380. Hook rate is 3-second plays divided by impressions.

- **Belief:** Our reviews say hair fall in the monsoon worries buyers most.
- **Change:** Open with that worry instead of the product shot. Same video body, same offer.
- **Prediction:** Hook rate rises from 24% to 28% or more.
- **Metrics:** Hook rate is primary. Cost per purchase must stay at INR 380 or below.
- **Budget and time:** INR 700 a day per version for 5 days, so INR 3,500 each.
- **Decision rule:** Scale the new hook if hook rate is 4 points higher and cost per purchase is no worse. Otherwise keep the old hook and test a different worry.

Now the result. The new hook gets 2,900 plays in 10,000 impressions. That is 2,900 / 10,000 = 29%, above the 28% target. If cost per purchase is INR 350, you scale it. If it is INR 450, you stop. The hook got attention but drew people who did not buy.

## Tips

- Test one thing at a time. If you change the hook and the offer, you will not know which one moved the number.
- Always write the "because". A guess with a reason teaches you something even when it loses.
- Set the budget and the days first. Do not call a test on day one.
- Log every test in the [creative testing tracker](/templates/creative-testing-tracker-template). A lost test is still an answer.
- Not sure how much to spend? See [how to build a creative testing budget](/guides/how-to-build-a-creative-testing-budget).

Marxx reads your connected Meta ad account in full and tags each ad by hook, angle, format and offer. That shows what you have already tested and what you have not. See [creative intelligence](/features/creative-intelligence).

## FAQ

### How long should an ad test run?

Long enough to spend the budget and cover the days you wrote down. A single day's numbers mix luck with signal.

### Can I test two changes in one ad?

You can, but you will not know which change caused the result. Change one thing, then test the next.

### What if the test ends in a tie?

Keep the version that is cheaper to make. Then test a bolder change. A tie often means the two versions were too alike. Read more about [creative testing](/glossary/creative-testing).

---
- [More Planning articles](https://marxx.ai/library-templates/category/planning)
- [All articles](https://marxx.ai/library-templates)