Measuring a 19-day decision with a 7-day report

A $460 phone is researched. Buyers read comparisons, check storage options, wait for a paycheck, come back three times. Even inside the reports the console does provide, two fifths of attributed sales arrived after day seven, and whatever lands beyond day fourteen is not reported at all.

The brand optimized on a seven-day report. Everything measured inside that window looked like a loss except the narrowest bottom-funnel terms, so over two years the account had been pruned down to exactly those: model-name-plus-storage searches by people who had already decided. Spend was efficient and the business had stopped growing, because every campaign that introduced the brand to a new buyer was switched off within two weeks of launch for failing a test it could not pass.

We changed the measurement before changing the spend.

• The yardstick changed to total-account economics: total sales against total spend on a rolling 28 days, because the console attributes seven and fourteen days and no setting extends that.
• Two separate budgets: a discovery budget judged on new-to-brand orders and assisted revenue, a harvest budget judged on immediate return.
• Upper-funnel terms reinstated with a minimum eight-week evaluation period, protected from being cut on one bad two-week stretch.
• Switch-back tests on the discovery campaigns, eight weeks on and four off, which is the only way to see a lag this long without per-user data.

Measured as total sales over total spend, the account returns 4.8x where the seven-day attributed report had been showing 2.1x: most of that gap was always there and simply outside the window the console reports.

Revenue grew 69% and new-to-brand orders nearly tripled, because a third of the budget is now spent introducing the brand to people who have not decided yet. The old approach was not wrong about its numbers; it was reading a seven-day window on a decision that takes weeks, and concluding that demand generation does not work.