Most websites and ad accounts come to us with history. Previous agencies, previous managers, previous owners. Each one leaves a little behind, and nobody’s dashboard flags it. Here are three leftovers we found in August, what each one was costing, and the questions worth asking about your own.
1. The pixel nobody owned
A resort group was about to ask a former property manager to confirm that their tracking had been removed from the group’s websites. Before sending that email, we checked.
It had not been removed. Four of the group’s five hotel sites were sending every visitor’s page view to a Facebook pixel that nobody on the client side or our side could account for. The pixel was not in the website itself. It was inside an older tag container, the kind of holding pen marketers use to load tracking code, and that container also carried ad tags from two other platforms. The fifth site ran only the newer container, and it was clean. That one site was the control that proved the rest could be cleaned too.
The bigger finding was not the pixel. It was that nobody on the client’s side held access to those older containers. Whoever does can change what runs on live booking websites without touching the websites at all. The residue predates the current owners, and the fix is not to rip the containers out, because some of what they carry may still be doing a job. The sequence is to find the owner, take ownership, then empty them and re-check.
2. One hotel was buying the other one’s guests
Two sister hotels, one owner, two ad accounts. A monitoring tool flagged a small amount of one hotel’s search spend going to people who were searching for the other hotel’s name.
We read the account directly instead of trusting the flag. The leak was about six times larger than the tool had seen. On the search side, every dollar of it traced back to a single broad-match keyword, where roughly half the spend was buying the sister hotel’s name. The tool could only see keyword suggestions, so it missed the two Performance Max campaigns, where about a third of all clicks were people looking for the sister property.
We blocked the sister hotel’s name on all three campaigns, spelled out each misspelling separately, because blocking one spelling does not block the others, and paused the broad keyword. And we said one more thing in the report, because it mattered: the sister hotel’s own account was already spending to its budget cap, so the clicks we blocked do not automatically land there. The fix was still right. A guest searching for one hotel should not land on the other. But it was not an argument for a bigger budget, and we did not let it become one.
3. Every lead was counted three times
A custom-home builder asked a simple question: does anybody monitor our Google Ads account? Nobody did, so we reviewed five months of it that afternoon.
Three separate conversion actions were set up to fire on the same contact form. One form submission counted as three leads. The account’s reported cost per lead was about a third of the real number, and the client’s own CRM agreed with the real number, not the ad platform. Worse, the automated bidding was being fed three signals per lead, so it was optimizing toward a distorted goal. The account’s change log showed no human edits in thirty days.
We set one conversion action as the real one and made the other two observe only, added the negatives the search terms were asking for, built a brand campaign, and put the account on our weekly pacing check. The budget recommendation was “fix first, then raise.” More money on a bidding system that counts triple would have scaled the distortion, not the results.
Three questions for your own accounts
- Who holds the tag containers on your website, and what is inside them? If the answer is “I’m not sure,” a third party can change what runs on your site today.
- Are you bidding on yourself, or on a sibling? Read the actual search terms, not the tool’s summary. Performance Max hides a lot.
- Does each lead count once? Compare the ad platform’s conversion number to your own inbox or CRM for the same month. If they disagree by 2× or 3×, your cost per lead is fiction and your bidding is learning from it.
We do this check on every account and site that comes to us with history. Most of what we find was nobody’s fault. It was just never looked at.