Skip to main content
AriaHelpDesk
Blog

Why your analytics stopped matching reality

Your dashboard says one thing, the bank says another. Usually it is not a bug: it is consent, ad blockers and attribution windows quietly changing underneath you.

6 min readAnalyticsMeasurement

At some point most teams notice their dashboard and their accounting no longer tell the same story. Sessions look healthy, conversions look flat, and the revenue figure in the analytics tool is meaningfully below what actually landed. The instinct is to look for a broken tag. Usually the tag is fine.

Consent is doing more than you think

Once a consent banner is in place, a share of your visitors never enter the dataset at all. Depending on the market and how the banner is worded, that share is not small. Nothing is broken; you are simply measuring a subset and comparing it against a total.

The fix is not to weaken the banner. It is to know your consent rate, and to read every number as a share of the consented population rather than of everyone.

Attribution windows quietly disagree

Each platform counts a conversion against its own window and its own rules. Two ad platforms will happily both claim the same sale. Add your analytics tool, which attributes differently again, and the totals cannot reconcile by construction.

  • Decide which system is the source of truth for revenue, and say so out loud
  • Use platform numbers for optimising inside that platform, not for board reporting
  • Compare like with like: the same window, the same definition of a conversion

Client-side collection loses data

Ad blockers, aggressive privacy modes and simple network failures all drop events before they arrive. Moving critical events server-side does not solve consent, but it does stop the avoidable losses, and it makes the remaining gap explainable.

What to do about it

Start by measuring the size of the gap rather than arguing about its cause. Pick one week, take the revenue figure from your billing system, take it from analytics, and write down the difference as a percentage. That number is your baseline.

Then work through the causes in order of size. In our experience consent is usually the largest, attribution the most confusing, and blocked events the easiest to fix. You will not get to zero, and you should not try. You should get to a gap you can explain in a sentence.