Conversion Lift Calculator
Your variant converts a bit higher — but by how much, and can you trust it? Enter both groups to get the relative lift, the absolute lift in points, and whether the gain is statistically real.
Conversion lift measures how much a variant beats the control. Relative lift — the percentage change in the rate — is what teams quote (‘a 30% lift’); absolute lift in percentage points is what actually shows up in revenue. This calculator reports both, then runs a two-proportion z-test so you know whether the lift is a real effect or just noise. A lift you cannot prove is a story, not a result.
Conversion Lift Calculator inputs and result
How to use this calculator
- Enter the control groupAdd the control’s visitors and conversions. This sets the baseline rate the variant is measured against.
- Enter the variant groupAdd the variant’s visitors and conversions. The tool computes both rates and the change between them.
- Read relative and absolute liftRelative lift is the headline percentage; absolute lift in points is the change you can multiply by volume to estimate revenue. Both matter for different reasons.
- Check the significance verdictThe p-value and verdict tell you whether the lift is real or within the range of normal noise at your confidence level.
- Size the win, then exportMultiply the absolute lift by your traffic to project the impact, then copy a share link, export CSV, or print.
RGM Expert Says
Lift is the number that gets put in the deck, which is exactly why it needs discipline. A ‘30% lift’ sounds enormous until you notice it moved a 1% rate to 1.3% — real, but small in absolute terms. We always report both numbers, because relative lift sells the story and absolute lift pays the bill, and conflating them is how teams over-promise.
The pairing we insist on is lift plus significance. A big relative lift on a small sample is the most seductive trap in optimization: it looks like a breakthrough and is often just a lucky run. The p-value beside the lift is the reality check — if the gain cannot clear the bar, it does not go in the deck as a win, it goes back into the test queue.
We also use the absolute lift to translate results into money. A two-point lift on a checkout that sees 50,000 sessions a month is a thousand extra conversions — that is the sentence that earns the next experiment its budget. Relative lift starts the conversation; absolute lift, multiplied by volume, is what finishes it.
How it works
The tool computes each conversion rate, takes the relative and absolute difference, and runs the same pooled two-proportion z-test used for significance to decide whether the lift is distinguishable from noise.
- p₁, p₂ — control and variant conversion rates.
- Relative lift — percentage change in the rate; the headline figure.
- Absolute lift — change in percentage points; multiply by volume for impact.
- p-value — two-tailed, from the pooled z-test; decides if the lift is real.
Relative and absolute lift are simple arithmetic; the significance check is the standard pooled two-proportion z-test with a normal-CDF (Abramowitz-Stegun erf) p-value. See Optimizely and Georgiev, Analytics-Toolkit on why lift needs a significance check.
A lift you cannot prove is just a number
Relative and absolute lift answer different questions, and using the wrong one misleads. Relative lift (‘+30%’) is great for comparing experiments and telling the story; absolute lift (‘+3 points’) is what you multiply by traffic to forecast revenue. A huge relative lift on a tiny base can be a rounding error in the P&L, so always read them together.
Lift without significance is the classic optimization mistake. Early in a test, the variant routinely shows double-digit lifts that vanish as the sample grows. The p-value beside the lift is what separates a result you can bank from one you are about to over-claim — which is why this tool refuses to show lift without it.
Once a lift is real, the absolute figure is your business case. A two-point lift on a high-volume funnel funds the roadmap; the same relative lift on a low-traffic page may not be worth the engineering. Translating lift into volume is how experimentation earns its seat at the budget table.
Relative vs absolute lift, side by side
The same absolute change looks very different as a relative lift depending on the baseline. This is why both numbers belong in every result.
| Control | Variant | Absolute lift | Relative lift |
|---|---|---|---|
| 1.0% | 1.3% | +0.3 pp | +30% |
| 10.0% | 13.0% | +3.0 pp | +30% |
| 5.0% | 5.5% | +0.5 pp | +10% |
| 20.0% | 21.0% | +1.0 pp | +5% |
What the experts say about lift
Report the absolute effect, not just the relative one. A 30% lift on a tiny base can be a rounding error dressed up as a breakthrough.
A lift without a confidence interval is marketing, not measurement. Always pair the number with how sure you are of it.