4 ad copy variations vs the control: −77% CPA in one week
Commentary byROASArchiveOur analysis, not theirsAgelessRx tested its current ad copy against 4 new value propositions, and after one week its top ad showed a 77% lower CPA.
The test was run with Marpipe, which published the case study. AgelessRx then raised daily spend by 6x.
- CPA: down 77% in the top performing ad after one week. The page does not say what the CPA was compared against or give dollar amounts.
- Daily spend: raised 6x after the test, per the page.
- Expectation: the SVP of Marketing, Andrew, says they would have been happy with a 10% cut.
- Not shared: spend, conversions, which copy won or what it said.
The change
Before
AgelessRx had one product shot and its current ad copy, which served as the control.
After
The team added 4 copy and value-proposition variations on the same product shot, giving 5 ads in total.
Our reading: why it likely worked
Our interpretation, not the advertiser's claim.
- With only one image available, the copy was the cheapest thing to vary, and the value proposition is often what decides a click on health products.
- A 77% drop in one week on the best of 5 ads may partly be a small-sample win. The 6x spend increase is a better sign that it held up.
- Keeping the image fixed meant any difference came from the words, which makes the lesson easier to reuse.
Caveats: Marpipe published this about tests run on its own tool, so it is vendor-reported. It is a one-week test of 5 ads with no spend, conversion counts or absolute CPA shown, and the 77% refers to the single best ad.
How to run this in your account
- Hypothesis
- If we test several new value propositions against our current copy on the same image, then CPA will fall because one message will match what buyers care about better.
- Primary metric
- CPA
- Guardrails
- Conversions per ad
- CTR
- Spend per variation
- Who should try it
- Brands with limited creative assets but room to write new angles, such as health or subscription offers. It may tell you little if each variation gets too few conversions to compare.
Who got the result
Marpipe · AgelessRx published this result (Tool vendor case study). The numbers are theirs; the summary and commentary on this page are ours.
Questions
How many ad copy variations should I test at once?
AgelessRx tested 4 variations against its control, 5 ads in total.
That was enough to find a top ad with a 77% lower CPA in a week.
Can changing ad copy alone lower CPA?
Here the image stayed the same and only the copy changed, and the top ad's CPA fell 77%.
The case study is vendor-reported and gives no absolute numbers.
How long should an ad copy test run?
This one ran for one week. Longer runs give more conversions per ad and make the winner more reliable.