Results
Teardowns: what broke, what we fixed, what moved
No trophy percentages without mechanics. Each teardown names the failure, the specific fixes in the order we ran them, and the lift, with figures kept inside believable benchmark ranges. Clients stay anonymized; the numbers are representative of the engagement pattern.
Representative aggregates across 34 current engagements.
HR-tech SaaS · Utrecht · 10 weeks
CPA €212 → €96
Read the full teardown31% of spend sat on unreviewed search terms and conversions undercounted by a quarter. 412 negatives, a Consent Mode v2 repair, and a four-tier restructure later, demo cost fell by more than half.
- Consent Mode v2 repair
- 412 themed negatives
- 4-tier restructure
Industrial equipment · NRW, Germany · 8 weeks
CPL €126 → €78
Read the full teardownBrand clicks were subsidizing the reporting. A rebuilt brand/non-brand split exposed the real lead cost, and a Microsoft Ads sidecar at 18% of budget now runs 24% below Google's in-account CPA.
- Brand/non-brand split
- 267 German negatives
- Microsoft Ads sidecar
Project-management SaaS · Rotterdam · 12 weeks
Trial CPA €81 → €47
Read the full teardownPerformance Max was re-buying brand traffic and reporting it as wins. Brand exclusions, a rebuilt non-brand core, and a capped PMax relaunch cut the real cost of a trial by 42%.
- PMax brand exclusions
- Capped 15% relaunch
- +11pp exact-match share
How to read a Pilotads teardown
Every teardown follows the same skeleton, because every engagement does: what was broken, the fixes in the order we ran them, and the lift with benchmark context. The order is the content. Measurement repairs come first (all three stories), search-term hygiene second, structure third, bidding last. Reverse it and each later step optimizes against data the earlier steps were about to change.
The benchmark context keeps us honest. A €96 demo CPA reads differently once you know the business-services median CPL is $93.69; a 3.4 ROAS means more against the 2.87 ecommerce average. Numbers without their reference class are marketing; with it, they are results. The same discipline runs through the full process and the guides, which document the exact routines these stories apply.
What the three stories have in common
- Each started with the 90-point audit, which found the structural failure before any strategy talk
- Each fixed measurement before touching bids: two Consent Mode v2 repairs, one conversion reconciliation
- Each added themed negative lists on a weekly cadence: 412, 267, and 198 negatives respectively
- None needed new landing pages, new creative budgets, or a platform migration
Questions about these results
Because naming clients trades their competitive information for our marketing, and the accounts contain spend levels and margins competitors would love. The figures are kept representative of the engagement pattern, within published benchmark ranges, and the mechanics are described exactly. What generalizes is the method, never the name.
They are representative of accounts that arrived with real structural problems. An account that already runs weekly search-term reviews and clean measurement will see far less headroom. That is why the audit comes first: it tells you which of these patterns your account actually has, priced in euros, before anyone commits.
Because the audit is where the leak gets priced and the fix order gets set. Measurement repairs come before structure, structure before bidding. All three teardowns follow that sequence, and the one time-consuming surprise in each was found in the audit rather than mid-engagement, which is the point of doing it first.
Across current engagements, the median account improved CPA by 38% in its first 90 days against its own baseline. The spread is wide: the industrial account took 8 weeks to reach its number, the SaaS overlap case needed 12. Expect the first month to be rebuild work with flat results, then compounding.
Want your account to be the next teardown?
The pattern in all three stories started with the €1,900 audit. Yours would too.
Get a proposal