Meta Scaled 350%. Once the Real Number Was Visible, Google's Own Search Campaign Started Lying About Why.
CertifyMe.Net, an online forklift certification platform, plus sister site Lasersafetycertification.com, both under the same account, same 60-day window.
Side-by-side tracking of Meta and Google that caught PMax re-billing demand Meta created.
Planned to separate cold-search from PMax's pooled remarketing before scaling further.
The Setup
CertifyMe.Net runs a Meta campaign that's been quietly working for about 60 days. Nothing fancy, mostly AI-generated creative (a forklift racing down an F1 track, that kind of thing), but effective for a blue-collar audience that isn't actively searching for "forklift certification near me" so much as scrolling past it in a feed.
From May 1 to June 30, spend on that campaign scaled 350%, from $7,300 to $32,000. The in-platform CPA barely moved (up just 6%, from $187 to $199) with 126 new imported purchases. By every account-native signal, scaling Meta was working cleanly.
The Break
Nothing changed in the Google account during this window: zero spend change, no structural edits, the same setup that had been stable for six months. And yet its numbers moved anyway. CPM up 88%, clicks up 100% (20,000 to 40,000), conversions down 22%, and conversion rate down a full 61%.
An account nobody touched got measurably worse at the exact moment a different platform's spend went up.
The Diagnosis
The mechanism, not just the outcome: this account runs Performance Max, which pools search, YouTube, Discover, and Display into one bidding pool. As Meta drove more people to the site (top-of-funnel, not yet searching), PMax started re-classifying those visitors as its own "engagement" the moment they later Googled anything adjacent to the brand, then bid to re-serve them cheap remarketing inventory dressed up as new demand.
The tell is in the metric itself. PMax's clicks stayed flat, but its "interaction" count and CPV spiked, meaning it was counting views and re-engagements, not genuine new clicks, as growth. Per a direct confirmation from a source inside Google, PMax's "intent" signal is defined as having been to your website before, meaning the campaign's own targeting logic is explicitly built to find and re-bill people another channel already reached, not to find people who haven't been found yet.
The sister account, Lasersafetycertification.com, is the clean control. It runs plain Search only, no PMax, and over the same period its clicks, impressions, spend, and conversions all stayed flat and stable, because there was no pooled remarketing engine sitting underneath it to reabsorb Meta's traffic and relabel it. Same portfolio, same 60 days, same Meta scale-up in the background. The only variable that changed the outcome was which Google campaign type sat on top of the account.
The Fix
This is a live account still in motion, not a closed-out before/after. But the diagnosis itself points to the fix: separate genuine cold-search demand capture (plain Search) from PMax's pooled remarketing behavior before scaling Meta further, so Meta's real prospecting lift doesn't get silently reabsorbed and re-billed by a campaign type that's structurally designed to claim credit for traffic it didn't originate.
Why It Matters
This is the same trust deficit AxiaOS's other case studies expose, but with a sharper mechanism than usual: a named, confirmed targeting rule, not a vague overclaim or a modeling artifact like an accrual window or a duplicate pixel. PMax defines "intent" as prior site visits, and that rule turns a channel built to look like prospecting into one that's mostly re-billing someone else's demand.
Scale Meta without knowing this, and the Google side of the account will report success (more clicks, more "interactions") in the exact same window its actual conversion performance is quietly collapsing. Two platforms are watching the same visitor. Only one of them was telling you what actually happened.
Owning the layer that watches both is the only reliable way to know which platform is really driving growth, consistently, rather than renting a dashboard from either one and hoping its incentives line up with yours.