A testing program,
not a hope that one ad works
Most accounts run three creatives for a month and call whichever spent the most the winner. We test hooks in isolation, with enough sample size to trust the result, and keep a steady supply of new creative so a winner never gets to fatigue before a replacement is ready.
Where the money leaks today
The most common mistake is launching several new creatives at once inside the same ad set, where the platform’s own delivery algorithm picks a favourite within hours based on early signals, long before there is enough data to trust that choice. This looks like testing but is closer to a coin flip with extra steps. The second failure is calling a test too early, declaring a winner after a day or two when the sample size has not yet reached anything close to statistical significance.
The third is a slow creative pipeline: once a genuine winner is found, running it unchanged for weeks until performance visibly drops, instead of having the next test ready before fatigue sets in.
A testing program only works if losing tests are allowed to lose; the discipline is resisting the urge to call a test early because one variant looks promising, since early calls are the single most common way a testing program degrades back into guessing with extra steps. A losing test is not wasted spend if it is logged properly, since knowing what does not work for a specific audience is genuinely useful information for the next round.
What we do
- Isolated testing. Hooks, formats and angles tested with proper separation so the result reflects the creative, not the platform’s early-delivery bias.
- Significance checked, not eyeballed. A test is called a winner only once sample size and duration support that conclusion, with an early-call warning if a team is tempted to decide too soon.
- A steady production cadence. New creative produced on a schedule that outpaces typical fatigue windows, so there is always a next test ready.
- A shared framework across platforms. Winning hooks and angles adapted across Meta, TikTok and other channels where the underlying idea transfers, instead of testing from zero on each platform.
- A logged history. Every test recorded with its result and reasoning, so institutional knowledge about what works for your brand accumulates instead of resetting every month.
What we need from you
- Product photography, footage or willingness to let us produce new creative, including UGC-style options.
- A fast approval process, since a testing program needs a steady supply, not an occasional batch.
- Enough budget and traffic to reach statistical significance in a reasonable window; very low-spend accounts may need longer per test.
- Agreement on what counts as the primary success metric before testing starts.
- Patience to let a test run its full course rather than calling it early.
How we measure
We report which variant won, by how much, and how confident that conclusion is, following a proper significance check rather than an early read. Where sample size allows, results are broken down by audience segment, since a variant that wins overall can lose for a specific segment, a detail easy to miss in an aggregate view.
Monthly reporting includes the full test log, so you can see the pattern across months, not just the latest result, and understand which angles reliably work for your brand over time.
We also track how long a confirmed winner keeps performing after the test ends, since this tells us how aggressively the next batch of creative needs to be scheduled.
Price and timeline
| Option | Price | What’s included | Timeline |
|---|---|---|---|
| Launch or audit | from $900 | Testing framework built, first batch of isolated creative tests launched. | 2 weeks |
| Monthly management | from $900 / month | Ongoing production cadence, significance-checked results, monthly test log report. | monthly, no lock-in |
| Full control, handover to your team | from $2,200 | Testing framework and significance methodology documented and handed to your team. | 3 to 4 weeks |
Related
This page sits under our Performance marketing and Brand and creatives work. See also Meta ads management, Tiktok ads management, Landing page testing program for adjacent paid-channel services.
For the automation side of reporting and budget decisions, see ab test analysis. For real numbers behind these benchmarks, read the sports-nutrition sales x2.7 case study and Thailand D2C store audit and rebuild case study.
Ready to see what this would look like for your account? Get in touch and we will look at your current setup in the first call.
FAQ
How much does this cost?
Both launch and monthly management start from $900, scaling with how much new creative is produced each month.
How long until we see results?
The program runs continuously; a first clear, statistically sound winner typically emerges in 3 to 4 weeks depending on traffic volume.
What ad budget do we need alongside this?
Enough traffic to reach significance within a reasonable window matters more than a specific dollar figure; we check your current volume in the first call and size the test plan accordingly.
What do you need from us?
Usable photography or footage, a fast creative approval process, agreement on the primary success metric, and patience to let tests run their full course.
How do you report on performance?
Monthly reporting on which variant won, the confidence behind that result, and a segment breakdown where sample size allows, plus a cumulative test log.