Skip to content
Documentation
All docs

Analytics Reference

Period Comparison

Compare return metrics between any two time periods to measure improvement.

Period Comparison is the toggle on the global date filter that lets you compare any two date ranges side-by-side. Every chart in the analytics hub respects it — pick two periods and every number, bar, line, and segment shows both ranges with the delta highlighted.

How to enable it

  1. Open any chart in Insights.
  2. Click the date range picker in the top right.
  3. Pick your primary period (the period you want to evaluate).
  4. Toggle Compare to previous (auto-fills the prior period of the same length) or Compare to custom to pick your own.

What you see

  • Numbers show both values (current and comparison) and the delta as an absolute change and a percentage change.
  • Trend lines overlay both periods on the same axis for eyeball comparison.
  • Bar charts show paired bars per category, with the change annotation on top.
  • Color coding — Green is improvement (lower returns, lower risk, higher savings), red is regression. Improvement and regression depend on the metric: a higher refund rate is bad, a higher savings number is good.

Useful comparisons

  • This week vs. last week — Routine pulse check. Use for on-call ops review.
  • This month vs. same month last year — Controls for seasonality. Useful for fashion stores where weekly comparisons are too noisy.
  • Last 30 days vs. the 30 days before RefundSentry was installed — The closest you’ll get to a clean before/after view of the product’s impact. Good for board updates.
  • After threshold change vs. before — Validate that a tuning decision actually moved the needle in the direction you wanted.

Caveats

Period comparison is naive — it doesn’t control for promotions, product launches, or marketing campaigns that drove a different mix of orders during one period vs. the other. A 40% jump in returns the week of your biggest sale of the year doesn’t mean RefundSentry got worse; it means more orders shipped.

For high-stakes comparisons, layer the period comparison with a filter on the same product type or customer segment. This isolates the variable you actually care about.

Next steps