01The starting pointHow they tested before
The brand had a good product, a designer, a founder with opinions and a budget of €12K a day. What it didn't have was a testing system.
- Each test ran about 5 days and cost around €800 per concept
- A "winner" was whatever had the best CTR when someone checked
- Tests shared campaigns and budget with the ads that paid the bills
- Disapproved ads sat in review for 3 days at a time
- The result was 2 new winners a month, against an account that needed five or more unique winning concepts per month to grow
The team knew creative was the bottleneck. So they made more of it. Output went up, and the number of winners stayed where it was.
02DiagnosisWhy it was slow and expensive
More ads didn't help, because the problem was never volume. Four things made every test cost more and teach less:
They tested variations, not ideas
9 of the last 12 "tests" were the same video with a different first line or colour. Those variations did not test distinct buying reasons, making the concept-level results harder to interpret.
Nobody had written down what "winning" meant
Without a spend cap or a CPA target, tests ran until someone looked. Some ran too long and burned money. Others were killed after a day, before a single purchase could mature.
Tests fought the scaling budget
New ads launched inside the same campaigns as proven ones. The algorithm favoured what it already knew, so new concepts barely delivered and stayed "inconclusive" for a week.
Launches kept slipping
Disapprovals, re-reviews and a spending limit meant that a test planned for Monday often started on Thursday. One payment failure and one spending-limit incident delayed two planned launches in the baseline month. Both were corrected before the new testing cadence began.
03ConceptsEvery concept starts with a buying reason
A concept is a reason to buy, not a visual style. We rebuilt the testing pipeline around three buying reasons, each answering a different question in the customer's head:
| Concept | Execution | The question it tests |
|---|---|---|
| Story | A creator explains on camera why people with neuropathy are switching to the cream | Does a relatable story make the purchase easier? |
| Ingredients | Static showing each key ingredient with a 60-day risk-free offer | Do ingredients plus a risk-free trial answer the main objection? |
| Night use | Product static built around one everyday pain moment and a low-commitment first try | Does a specific moment make the product feel worth trying? |
Swipe the table sideways to see every column →
Writing down the question changes the review. Instead of "which ad got the most clicks", the team now asks "what did we learn about why people buy".
04BriefingThe brief every concept goes through
Every ad starts on the same one-page brief. It takes ten minutes to fill in, and it removes most of the reasons ads fail or get disapproved.
- Buying reason
- Relief without covering up the discomfort
- Proof we can show
- Creator holding the jar, ingredients named on camera
- Opening (0–3s)
- The hook line and the creator on screen in the first frame
- Demonstration
- Why nerves get overstimulated, what is in the cream, how it is used
- CTA + destination
- "Tap the link and grab the official cream" → product page offer
- Not allowed
- Before/afters, invented testimonials, claims the product page does not support
05CreativeRound one
The first round: three concepts with two openings each, plus a static for every video. Here's what ran and what happened.
"This is how thousands of people with neuropathy are aging their nerves backwards."
€480 spend, 20 new customers, €24 CPA against a €32 target. Scale verdict after the one-day click window matured.
"Beat nerve pain. Try it risk-free for 60 days."
€600 spend, 18 purchases, €33.33 CPA; 36% three-second-view/impression rate and 1.7% link CTR. Rework the offer and ingredient explanation; this diagnosis is a hypothesis.
"If sitting for more than 10 minutes wrecks your back… try it for a night."
€240 spend and 4 new customers (€60 CPA), above 1.5× the €32 target. Stopped after maturation; the link between the pain moment and the product needed revision.
06RulesBudgets and verdicts, written before launch
Each concept got a hard cap of €600, so a round of three cost at most €1,800. The CPA target and the review date were agreed upfront, allowing for the brand's normal purchase delay of 1 day.
| Verdict | Rule | Next step |
|---|---|---|
| Scale | CPA at or under €32 with at least 15 purchases | Moves to scaling; the test record remains linked to the concept |
| Rework | Attention is there (hook rate, CTR) but CPA is above target | Same reason to buy, new opening or new destination |
| Kill | At the matured review: no purchases at the cap, or CPA above 1.5× target | Off, logged, angle parked |
| Inconclusive | Delivery too low to judge | Relaunch once. No forced winner because the review date arrived |
Swipe the table sideways to see every column →
Clicks and watch time help explain a result. Only purchases at an acceptable cost decide one.
07The rebuildWhat made testing faster
The shorter cycle came from several changes working together. Better briefs meant fewer weak tests. Written rules meant faster verdicts. A setup that didn't stall meant tests actually started when they were planned.
| Area | When they ran it in-house | After the rebuild |
|---|---|---|
| Concepts | Variations of one video | Distinct buying reasons, each with a written question |
| Briefs | Loose direction to the designer | One-page brief with proof, opening, destination and limits |
| Verdicts | Best CTR when someone checked | Spend cap, CPA target and review date set before launch |
| Campaign setup | Tests inside scaling campaigns | A dedicated testing campaign with its own budget |
| Ad accounts | One account, shared spending limit | Our agency accounts, with backups ready |
| Compliance | Disapprovals found after launch | Checked before launch, from 6 to 0 disapprovals a month |
Swipe the table sideways to see every column →
Funding ran through top-ups, scheduled before each Friday launch so tests started on time.
The test cycle went from 5 days to 2.5. Sharper briefs, clear rules and fewer delays each took a piece of that.
2 to 5 winning concepts a month. 40% less per test.
This team reduced average test spend from €800 to €480 and cut the testing cycle from 5 to 2.5 days. Send me how many concepts you test each month, your average spend per test and your target CPA. I'll review your testing structure, spend caps and winner criteria.
Review my testing setup →08CadenceThe weekly rhythm
The same five steps, every week. The review date never overrides the purchase delay.
- Monday
Review matured results
Verdicts on last week's tests, based on purchases that have had time to come in.
- Tuesday
Write the briefs
New concepts from open questions; reworks from last week's near-misses.
- Wednesday–Thursday
Produce
3 concepts × 2 openings, plus statics.
- Friday
Compliance check and page match
Every ad checked against the product claims and the exact page section it lands on.
- Friday evening
Launch
Into the testing campaign, with caps set and the balance topped up for the weekend.
09IterationIterating on winners: one change at a time
A winner isn't finished. It's the best-understood idea in the account, so that's where the cheapest next win usually sits. Each new version changes exactly one thing: the opening, the creator, the demonstration or the CTA.
| Version | What changed | Spend | CPA | Verdict |
|---|---|---|---|---|
| STORY_V01 | Original | €480 | €24 | Scale |
| STORY_V02 | New opening: the creator starts with the burning and tingling at night. | €600 | €33 | Rework |
| STORY_V03 | Different creator, same script | €240 | €60 | Kill |
| STORY_V04 | Static version of the best frame | €600 | €50 | Kill |
Every row is saved with the asset, the dates, the market and the offer. When a creator leaves or a new buyer joins, the learning stays in the account.
10ResultsResults
| Measure | Before | After | Change |
|---|---|---|---|
| Cost per concept test | €800 | €480 | −40% |
| Test cycle | 5 days | 2.5 days | −50% |
| Winners per month | 2 | 5 | ×2.5 |
| Daily spend | €12K | €45K | ×3.75 |
| CPM | €14.00 | €8.40 | −40% |
| Profit per month | €181K | €1.38M | ×7.62 |
| Purchase CPA | €42 | €31 | −26.2% |
More winners meant the scaling campaign always had fresh, proven ads to put budget behind. That, more than any single ad, is what let daily spend grow.
Profit is reported after media spend, product and operating costs, creative production, our fees and the allocated account setup cost. ROAS, CPM and CPA use media spend only. Profit increased by €1.199M across the 30-day comparison windows, from €181K to €1.38M.
Every launch had a buying reason, a spend ceiling and a purchase-based verdict; variations were kept separate from new concepts.
11TakeawaysWhat to take from this
- More ads won't fix a testing problem. Distinct buying reasons will.
- Write the rules before you spend. Cap, CPA target, review date. No forced winners.
- Give tests their own room. Separate campaign, separate budget, no competing with what already works.
- Remove the waiting. Check compliance before launch and make sure nothing in the setup stalls a test once it is planned.
- Iterate one variable at a time, and keep a log the next person can read.