Almost every article about ad testing is written for advertisers with statistical significance available to them. A clinic doing 30 patients a month does not have it, and running a formal A/B test on that volume produces confident conclusions from noise.
Here is what we do instead.
Accept what you cannot know
At 30 conversions a month, the difference between a 4% and a 5% conversion rate takes many months to detect reliably. By then the market has moved, the season has changed, and you have tested something that no longer applies.
At local volumes, you cannot detect small differences. So stop looking for them and only test things large enough to be obvious.
Test big, not small
Worth testing: a different offer. A different format entirely — static versus a clinician on camera. A fundamentally different angle — price-led versus credential-led. A different landing experience — a page versus a short qualification sequence.
Not worth testing at this volume: button colors. Headline word order. Image crops. Three near-identical variants.
The first list produces differences big enough to see through the noise. The second produces numbers that move and mean nothing.
Use upstream metrics as an early signal
Conversions are scarce. Clicks, click-through rate and landing page views are not, and they arrive days earlier.
They are not the outcome, and optimizing to them is the classic error that produces cheap worthless leads. But as a kill signal they are useful: creative with a very poor click-through rate is not going to be rescued by a good conversion rate downstream. Cut it early and spend the budget on something that might work.
Run fewer things, for longer
A common local account has eight ads splitting a small budget, none of which accumulates enough delivery to leave the learning phase.
Two or three at a time, running long enough to matter, beats eight running briefly. It feels slower. It is faster.
Let the account tell you what it already knows
Before testing anything, look at what has already run. Most accounts contain months of unexamined evidence: which angles produced bookings, which placements delivered, which times of day converted.
We have repeatedly found the answer to "what should we test?" sitting in an account's own history, unread.
The thing that beats all of it
Test the offer. The offer is the campaign — two ads with the same image and different offers will teach you more in a week than a month of creative variants against a fixed offer.
Creative testing is the fun part. Offer testing is the part that changes the business.
If your account has eight active ads and a small budget, consolidating to three is usually the first improvement, before any new creative is made.