Free tool
CRO ROI calculator
What a testing programme is worth on your store. Honest ranges, not single-point projections, with a check on whether your traffic can actually detect the lift you are modelling.
| Scenario | CR lift | Annual revenue | Delta | Detectable in |
|---|---|---|---|---|
| Conservative | +5% | $2,159,136 | $102,816 | 22 weeks |
| Likely | +10% | $2,261,952 | $205,632 | 6 weeks |
| Ambitious | +20% | $2,467,584 | $411,264 | 1 week |
Stay sharp on ecommerce CRO
3 short examples, 2 CRO tips, 1 teardown. Every Tuesday.
What this calculator does
It turns four numbers you already have (monthly sessions, conversion rate, average order value, and any average order value you expect to give up) into a year of revenue under three lift scenarios. Then it does the thing most ROI calculators skip: it checks whether a test on your traffic could actually detect each lift, and reports that in weeks.
Use it before you commission a programme, before you renew one, and when a board or a partner asks what conversion work is worth. It is also the fastest way to find out that your store is too small to test, which is a cheaper thing to learn here than four months into a retainer.
The maths, in the open
No modelling secrets. The revenue side is arithmetic; the detectability side is the standard sample-size shortcut for a 95% confidence, 80% power test.
The revenue
baseline = monthly sessions × 12 × (conversion rate ÷ 100) × AOV
scenario CR = conversion rate × (1 + lift)
scenario AOV = AOV × (1 − AOV drag ÷ 100)
scenario revenue = monthly sessions × 12 × (scenario CR ÷ 100) × scenario AOV
delta = scenario revenue − baseline
The three scenarios are fixed relative lifts on conversion rate: +5% conservative, +10% likely, +20% ambitious. They are applied to the whole year, which is the optimistic part of the model and worth remembering when you read the delta.
The detectability check
absolute MDE = conversion rate × lift
visitors per variant = 16 × p × (1 − p) ÷ (absolute MDE)²
weeks = (2 × visitors per variant) ÷ (monthly sessions ÷ 4.33)
p is the baseline conversion rate as a decimal, and 4.33 is the average number of weeks in a month. The 16 is the usual shortcut constant for 95% confidence and 80% power. It lands within about 10% of the exact figure; for the precise number, use the significance calculator.
A worked example
These are the numbers the calculator loads with: 120,000 monthly sessions, a 2.1% conversion rate, $68 average order value, and no AOV drag. Baseline annual revenue is $2,056,320.
| Scenario | CR lift | Annual revenue | Delta | Detectable in |
|---|---|---|---|---|
| Conservative | +5% | $2,159,136 | $102,816 | 22 weeks |
| Likely | +10% | $2,261,952 | $205,632 | 5 weeks |
| Ambitious | +20% | $2,467,584 | $411,264 | 1 week |
How to read the result
The awkward row is the conservative one. A +5% lift is worth $102,816 a year on this store and takes 22 weeks to prove. That is two tests a year. The modest outcome is the one your traffic can least afford to measure, which is why programmes on mid-size stores chase bigger changes rather than safer ones.
Plan on the Likely row. Conservative is what a year of mostly inconclusive tests looks like. Ambitious is a good year on a mature programme. Budget against the middle and treat the top as upside.
Put your own margin on it. The delta is revenue, not profit. Multiply by your contribution margin, then subtract what the programme costs, before you compare it to anything else you could spend the money on.
What it will not do
Subtract costs. Nothing here nets off agency fees, developer time or testing tools. It is a revenue model, not a return model, despite the name.
Model the ramp. The delta assumes the lift is live for all twelve months. In practice you ship winners across the year, so year one usually lands below the number and year two above it.
Model decay or interaction. Some winners fade, some only apply to one traffic source, and two winners rarely add up cleanly. The model treats a lift as permanent, sitewide and independent.
Hold seasonality. Sessions, conversion rate and AOV are all fixed at whatever you typed. If your December is triple your February, run it twice and read both.
The three inputs that matter
Traffic sets the ceiling on what you can measure. Conversion rate sets the baseline. AOV sets the value of every incremental order. Move any of them and the whole calculation moves with it.
The "Likely" scenario is the one worth planning around. "Ambitious" is what a good year looks like across a mature programme. "Conservative" is what happens if most tests come back inconclusive at your traffic, which is what actually happens on most stores under 50,000 sessions a month.
For the fuller argument on what a testing programme actually costs and returns, and how to size one honestly, read Chapter 1 and Chapter 4 of the free course. Then use the significance calculator to check whether the tests you are modelling can conclude.
All three calculators, live on one page → Significance, programme ROI and the free-shipping threshold, with what to do about each.
What founders ask before they book.
Why does this tool show a range, not a single number?
Because CRO outcomes are noisy. Any "your annual lift = $487,213" calculator is confidently reporting fiction. A conservative-to-ambitious range is the honest shape of the number.
Where do the lift assumptions come from?
From a deliberately conservative planning assumption, not from our own best year. The model assumes roughly 1 in 5 tests wins, the rate a typical programme sees, with winners clustering in the 3–8% range on the metric they targeted. Compounding across a year lands most stores in the 10–20% total range, which is where the "Likely" scenario sits. Our own published record is better than that; projecting your store off our best engagements would be the kind of forecast this site argues against.
What is the AOV drag input for?
Guardrail. A conversion win that lowers average order value can be a net loss on revenue per session. Setting a small AOV drag models that trade honestly.
Why the "Detectable in" column?
Because a 5% lift at 30,000 monthly sessions is genuinely undetectable in any reasonable window. The annualised value is theoretical if you cannot actually run the test. Chapter 3 spends real time on this.
Does the output subtract what the programme costs?
No. Every figure here is gross incremental revenue, not profit and not return on investment net of fees. Subtract your own agency, developer and tooling costs, and your gross margin, before you call anything a return.
Want us to run the programme?
Same maths, done on your account. Free 20-minute call, and if the numbers above do not work at your traffic we will say so.
Got it. Now pick a time.
Open the calendar →Twenty minutes on your store and your numbers. If we cannot help we will say so.
- 437+ stores built
- 2,000+ tests shipped
- No call needed to start