The Quiet Discipline of Floodlight Hygiene

I once inherited a legacy enterprise account where the dashboard reported an incredible spike in conversions — but the client’s internal sales database stayed completely flat. Auditing the back-end tagging architecture turned up the reason: multiple legacy Floodlight tags, left behind by past agencies, all firing simultaneously on a single confirmation page. Our industry loves talking about algorithmic bidding. Almost nobody audits the plumbing that feeds it.

Here’s the three-step check I now run on a schedule, not just at launch.

1. Run the CRM sanity sync (10 minutes)

I pull a raw Floodlight volume report and set it side-by-side against the client’s actual internal sales numbers or Google Analytics. A major statistical gap between the two is an immediate flag for a tracking discrepancy or a double-firing tag.

2. Calibrate the lookback window (5 minutes)

I check the active attribution settings. Accounts are frequently left running on the default 30-day click/view lookback window. For a fast-moving or low-consideration product, an overly generous view-through window inflates performance that isn’t real. I tighten the window to match the client’s actual buying cycle.

3. Audit the custom variables (10 minutes)

I check whether proprietary data points — customer tier, purchase value, product line — are passing cleanly through the custom (u-) variables. If those strings come back blank or errored, the optimization engine is flying blind and doesn’t know it.

4. Real example: paying to reach people who already bought

We managed a lead-generation campaign where automated bidding was aggressively scaling spend against a high volume of form completions. Cost-per-lead looked excellent to the untrained eye. A routine hygiene audit found that our custom variables weren’t distinguishing existing users logging into their portals from new prospects. The system had learned it could buy cheap ad space by retargeting people who were already customers — and it was doing exactly that, efficiently. Cleaning up the Floodlight logic to isolate true net-new actions saved the client nearly 40% of media spend that had been going toward preaching to the converted.

Wrapping up

None of these three checks are glamorous, and none take more than ten minutes. But skipping them means every downstream decision — bidding, budget, optimization — is being made on data that might already be lying to you. Clean data beats a clever strategy every single day.