Conversion Rate & A/B Test Calculator
Find your conversion rate and what a better one is worth, check whether an A/B test winner is real, plan how many visitors a test needs, and see where your funnel leaks.
Or sessions, ad clicks, email opens or leads: whatever people could convert from.
Sales, sign-ups, downloads or bookings from those visitors.
What one conversion is worth to you on average. Leave empty to count conversions only.
Empty: we show a rate 10% higher than now.
For the same period. Gives cost per conversion and return on ad spend.
Two-sided is the safe default. Use one-sided only if you decided before the test that you only care whether B beats A.
Checks the split came out as planned (sample ratio mismatch).
With order value, shows what switching everyone to B is worth.
Your baseline: the rate of the page as it is now (A).
Minimum detectable effect: 20% of 2.5% means 2.5% → 3%.
2 for a plain A/B test; 3 for A/B/C, and so on.
What a higher conversion rate is worth
How your numbers are worked out
Conversion rate = conversions ÷ visitors × 100.
- 1
Conversion rate
250 ÷ 10,000= 2.5%
- 2
Revenue per visitor
2.5% × $50= $1.25
- 3
A rate 10% higher
2.5% × 1.1= 2.75%
- 4
Extra revenue
10,000 × 0.25%= 25 sales × $50= $1,250
Read more: the formulas, worked examples, A/B test significance and sample size
A conversion is any action you want visitors to take: a sale, a sign-up, a booking, a download. The conversion rate is the share of visitors who take it. This page also tests whether a change really moved that rate (an A/B test), plans how big a test must be, and finds where people drop out of a funnel.
The formula
conversion rate = conversions ÷ visitors × 100
revenue per visitor = conversion rate × average order value
extra sales = visitors × (new rate − current rate)
cost per conversion = spend ÷ conversions
return on ad spend = revenue ÷ spend
uplift = (rate B − rate A) ÷ rate A
pooled rate p = (conversions A + conversions B) ÷ (visitors A + visitors B)
standard error = √(p × (1 − p) × (1 ÷ visitors A + 1 ÷ visitors B))
z = (rate B − rate A) ÷ standard error
p-value (two-sided) = 2 × (1 − Φ(|z|))
interval for the gap = (rate B − rate A) ± z* × √(rA(1 − rA) ÷ nA + rB(1 − rB) ÷ nB)
visitors per version = (zα × √(2 p̄(1 − p̄)) + zβ × √(p₁(1 − p₁) + p₂(1 − p₂)))² ÷ (p₂ − p₁)²
days = visitors per version × versions ÷ visitors in the test a day
Φ is the standard normal distribution, z* is 1.645, 1.96 or 2.576 for 90%, 95% or 99% confidence, zα is the same value for the significance you choose (1.96 for 95% two-sided), zβ is 0.842, 1.282 or 1.645 for 80%, 90% or 95% power, and p̄ is the average of the current rate p₁ and the new rate p₂.
Worked example
Conversion rate. 250 sales from 10,000 visitors is 250 ÷ 10,000 = 2.5%, one sale for every 40 visitors. At $50 an order that is $12,500 of revenue, or 2.5% × $50 = $1.25 per visitor. A rate 10% higher is 2.5% × 1.1 = 2.75%, which on the same 10,000 visitors is 10,000 × 0.25% = 25 more sales, worth 25 × $50 = $1,250.
A/B test. A gets 250 conversions from 10,000 visitors (2.5%), B gets 300 from 10,000 (3.0%). B is 0.5 points higher, an uplift of 0.5 ÷ 2.5 = 20%. The pooled rate is 550 ÷ 20,000 = 2.75%, so the standard error is √(0.0275 × 0.9725 × (1 ÷ 10,000 + 1 ÷ 10,000)) = 0.231%. Then z = 0.5% ÷ 0.231% = 2.16 and the two-sided p-value is 0.0306. That is below 0.05, so B wins at 95% confidence (but not at 99%, where p must be below 0.01). The 95% interval for the gap is 0.5% ± 1.96 × 0.231% (unpooled), or about +0.05 to +0.95 points: B is better, but by anything from a little to a lot.
Sample size. To spot a 20% lift on a 2.5% rate (2.5% → 3.0%) with 95% confidence and 80% power: p̄ = 2.75%, so (1.96 × √(2 × 0.0275 × 0.9725) + 0.842 × √(0.025 × 0.975 + 0.03 × 0.97))² ÷ 0.005² = (0.4533 + 0.1946)² ÷ 0.000025 = 16,792 visitors per version, 33,584 in all. At 1,000 visitors a day the test takes 33,584 ÷ 1,000 = 34 days.
Funnel. 10,000 visitors → 4,000 view a product (40%) → 1,200 add to cart (30%) → 600 reach checkout (50%) → 300 buy (50%). Overall that is 300 ÷ 10,000 = 3%. The leakiest step is add to cart, which loses 70% of product viewers. If that step kept 10% more people, every later step grows by 10% too: 30 more sales.
Reading an A/B test result
| Confidence | Significant if p is below | z needed (two-sided) | z needed (one-sided) |
|---|---|---|---|
| 90% | 0.10 | 1.645 | 1.282 |
| 95% | 0.05 | 1.960 | 1.645 |
| 99% | 0.01 | 2.576 | 2.326 |
- Decide the sample size first, then wait for it. Peeking every day and stopping at the first significant result can turn a 5% false-win rate into 20–30%.
- Run whole weeks. Weekday and weekend visitors behave differently; one or two full weeks at least.
- "Not significant" isn't "no difference". It means this test was too small to tell. Look at the confidence interval: if it is wide, you need more visitors.
- More versions need stricter tests. With A, B and C you make two comparisons, so this page splits the 5% between them (Bonferroni): each must reach p < 0.025.
- Check the split. If a 50/50 test ends up far from 50/50 (sample ratio mismatch, p < 0.01), something in the set-up is broken and the result can't be trusted.
- "Chance B beats A" is a quick estimate from the same normal approximation (like a Bayesian result with flat priors), not a p-value.
The statistics use the normal approximation to the binomial distribution (the standard two-proportion z-test, the same as most A/B test tools). It works well when each version has at least about 10 conversions and 10 non-conversions; with fewer, use an exact test (Fisher's). Each visitor should be counted once and only see one version. Sample sizes are the usual fixed-horizon formula without continuity correction, so other tools may differ by a few percent.
Common questions
How do you calculate conversion rate?
Divide the number of conversions (sales, sign-ups, leads) by the number of visitors, sessions or clicks, then multiply by 100. With 250 sales from 10,000 visitors, the conversion rate is 250 ÷ 10,000 × 100 = 2.5%, or one sale for every 40 visitors.
How much is a higher conversion rate worth?
Multiply your visitors by the rise in rate, then by your average order value. Raising a 2.5% rate by 10% to 2.75% on 10,000 visitors gives 10,000 × 0.25% = 25 more sales. At $50 an order that is $1,250 more revenue, with no extra traffic.
How do I know if my A/B test result is significant?
Run a two-proportion z-test. With 250 conversions from 10,000 visitors on A (2.5%) and 300 from 10,000 on B (3.0%), z = 2.16 and the two-sided p-value is 0.0306. That is below 0.05, so B is significantly different at 95% confidence, but not at 99%.
What does the p-value mean in an A/B test?
It is the chance of seeing a difference at least this big if A and B really convert at the same rate. A p-value of 0.03 means a gap like yours would turn up by luck about 3 times in 100. It is not the chance that B is better.
How many visitors do I need for an A/B test?
It depends on your current rate and the smallest change you care about. To spot a 20% lift on a 2.5% rate (2.5% to 3.0%) with 95% confidence and 80% power, you need about 16,792 visitors per version, or 33,584 in total. At 1,000 visitors a day that takes 34 days.
Should I stop my A/B test as soon as it is significant?
No. Checking again and again and stopping at the first significant result makes false wins far more likely. Work out the sample size first, run the test for that many visitors and for at least one or two full weeks, then read the result once.
Related calculators
Made by Zulu Calculator, part of AlphaZulu LLC. Formulas are shown in full and rule figures are checked against official sources; see how we build calculators.