Core concepts

Attribution & settled data

Why recent days keep changing, and how MerchDash keeps automation from acting on incomplete numbers.

When a shopper clicks your ad today and buys tomorrow, Amazon credits the sale back to the day of the click. That is attribution, and it means the last few days of your data are always incomplete and still changing. MerchDash is built around this fact.

The two-day settled window

Most late attribution lands within two days. MerchDash therefore treats the most recent two UTC days as unsettled: rolling windows in automation (LAST 7 DAYS, LAST 30 DAYS) and YESTERDAY automatically exclude them, so a rule never pauses a keyword for having “no orders” when the orders simply have not been attributed yet.

WindowWhat it reads
YESTERDAYThe latest settled attribution day, currently two UTC days behind today.
LAST 30 DAYSThe last 30 settled days (ending two days ago).
TODAY, N DAYS AGOExact calendar days, deliberately unlagged, partial data, useful when you explicitly want the freshest numbers.
FROM x DAYS AGO TO y DAYS AGOExact offset windows, also unlagged.

The 14-day restatement window

Amazon can restate sales up to about 14 days back. That is why every sync re-pulls the trailing 14-day window (see How data sync works): restated days are overwritten in place, and your archive converges on Amazon's final numbers. Beyond ~14 days, data is stable.

Why your numbers may differ from the Amazon console

  • Recency. The console shows live-but-partial numbers for recent days; MerchDash's settled windows intentionally exclude them. Compare like for like: a settled MerchDash window against the same fixed dates in the console, a couple of days after they end.
  • Sync timing. MerchDash data is as fresh as the last sync (roughly every 12 hours). The dashboard shows exactly when that was.
  • Time zone. MerchDash days are UTC.
Rule-writing takeaway. Use settled windows (LAST N DAYS, YESTERDAY) for anything that acts on conversions, pausing, harvesting, negating. Reach for TODAY only when partial same-day data is genuinely what you want, such as an intraday budget brake.