Conversion rate isn't a single number. It's a trend across multiple time horizon
By Anthony Nguyen 🟢 · June 9, 2026 · Curated by George's Blog
Conversion rate isn't a single number. It's a trend across multiple time horizons.
I use five lookback windows: 7, 14, 30, 60, and 90 days. Each one answers a different question.
7-day CVR catches sudden changes. A listing hijack, a negative review surge, a price increase from a competitor, these show up in the 7-day window first. If 7-day CVR drops 30% below your 30-day average, something broke. Go find it.
14-day CVR smooths out weekend and promotional noise. Flash deals and Lightning Deals create spikes that distort the 7-day number. Two weeks gives a cleaner signal.
30-day CVR is your operational baseline. This is what Amazon's algorithm likely weighs most heavily for ranking decisions. If 30-day CVR is declining, organic rank will follow.
60 and 90-day CVR provide seasonal context. A "declining" 30-day CVR might actually be normal if you're comparing peak season to off-season. The longer windows tell you whether current performance is truly anomalous or just cyclical.
The system expands adaptively. It starts with the 7-day window and only looks further back when short-term data is insufficient or inconsistent.
Bid decisions based on a single CVR number are guesswork. Bid decisions based on CVR trends across multiple windows are strategy.
AdsCrafted tracks your organic rank, runs keyword gap analysis, and builds rank strategies, all in one conversation → adscrafted.com