Diagnosing whether a campaign's performance drop is real or noise requires reviewing three time windows side by side: last week, last month, and last three months. Consistent underperformance across all three windows indicates a genuine problem; a dip visible only in the shortest window may just be normal fluctuation.
Any drop identified this way should be cross-checked against the listing's change history before acting on the ad account alone — title edits, image changes, price changes, and stock-outs can all cause a PPC performance drop that looks like an ads problem but is actually a listing problem. This mirrors the logic of Leading Actions vs. Lagging Metrics (PPC Diagnosis Framework): an ad metric drop is a lagging signal, and the leading cause may sit outside the ad campaign entirely.
When any of the three windows (last week, last month, last three months) shows a drop, don't diagnose it as a targeting or bid problem in isolation. Cross-check the dip against the account's PPC change history first, then against the listing's own edit history — title changes, image changes, price changes, and stock-outs. A dip that lines up with a listing edit or a stock-out is a listing problem wearing PPC clothing, not an ad-targeting problem.