The metric that was lying by 40% for a year

The metric that was lying by 40% for a year

A media site's whole strategy rested on pageviews, and they'd been climbing nicely. Ad revenue per pageview, oddly, kept falling. The two numbers didn't agree.

The clue was in the real-time report: a single article showed two pageview events firing milliseconds apart. A botched tag migration had double-fired the analytics snippet on 40% of templates for nearly a year. Pageviews were inflated; the falling revenue-per-view was the truth leaking through.

They deduplicated the tag and rebased the historical numbers.

'Real' pageviews dropped 28% overnight — painful — but engagement metrics like scroll depth and time-on-page suddenly matched the revenue. Planning finally ran on honest numbers.

Lesson: when a volume metric and a revenue metric disagree for months, suspect the volume metric first.