Multi-Line for A/B Test Metrics — Experiment Variant vs Control Tracker
An A/B test multi-line graph tracks longitudinal conversion metrics for experimental treatment arms ($CR_{Control}(t), CR_{VarA}(t), CR_{VarB}(t)$) across time to confirm sustained uplift beyond day-of-week seasonality.
Tip: Paste Excel or Google Sheets rows (Ctrl+V / ⌘V) directly into any cell.
| # | Label | ||||
|---|---|---|---|---|---|
| 1 | |||||
| 2 | |||||
| 3 | |||||
| 4 | |||||
| 5 | |||||
| 6 |
Relative lift of Variant A over Control: Lift% = ((CR_A - CR_Control) / CR_Control) * 100%. Absolute basis point lift: bps = (CR_A - CR_Control) * 100.
Chart Design & Use-Case Tips
- 01Relative lift of Variant A over Control: Lift% = ((CR_A - CR_Control) / CR_Control) * 100%. Absolute basis point lift: bps = (CR_A - CR_Control) * 100.
- 02Look for parallel vertical separation where the winning variant consistently stays above Control across all weeks without crossing back.
- 03If Variant A spikes in Week 1 but converges back to Control by Week 3, the lift was likely a temporary novelty effect.
- 04Run experiments for at least 2 full weekly cycles to smooth out weekend vs. weekday traffic mix.
Frequently Asked Questions
Why plot A/B test results over time instead of just looking at a final bar chart?+
A longitudinal multi-line plot reveals whether the treatment lift held steady across every week or was distorted by a single anomalous traffic spike or novelty effect.
What does it mean if the variant lines cross each other repeatedly?+
Frequent line crossovers indicate high variance relative to effect size—meaning the difference between variants is likely statistical noise.
Can I plot just Control vs. Variant A (2 lines)?+
Yes, set Series C values to zero or clear them, and the chart will cleanly compare your two active lines.