Run your own numbers.
Three calculators from client work, left out in the open. No signup, nothing you type leaves your browser.
Four numbers you already have become a year of revenue under three lifts, and a check on whether your traffic could even detect them.
| 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 |
Check first
The weeks-to-detect on the Conservative row. If +5% needs twenty weeks, you get two real answers a year.
What it means
Revenue, not profit, assuming twelve live months. Plan against Likely; treat Ambitious as upside.
Monday
Likely delta × your margin − programme cost. If Conservative can't conclude inside a quarter, buy traffic, not tests.
The formula and a worked example → Every scenario, the detectability shortcut, and the four things it will not model.
Plan sizes the test before it runs. Analyze says whether the finished result is real.
Variant B vs Control
SignificantVariant B is beating control by 15.0% relative. At 95.0% confidence this is a real difference, not noise. Ship it.
Check first
Plan tab: the days to runbefore a developer touches the theme. Analyze: the interval, not the headline lift.
What it means
Significant ≠ big, and “not yet” ≠ “it lost”. Inconclusive at the planned sample means the effect was smaller than you sized for. Write it down.
Monday
Put the stop date where people can see it, then don't peek. Stopping at the first 95% is the most expensive habit in CRO.
The formula and a worked example → Sample size, the z-test, the Bayesian read, and the three ways programmes fail.
Free shipping is a discount on every qualifying order. This prices six thresholds against having none at all.
| Threshold | Order value | Contribution / order | vs no threshold |
|---|---|---|---|
| $50.00 | $57.17 | $29.49 | -$3.91 |
| $60.00 | $58.26 | $30.69 | -$2.71 |
| $65.00 | $59.00 | $31.41 | -$1.99 |
| $70.00 | $59.87 | $32.20 | -$1.20 |
| $80.00 | $62.00 | $34.02 | +$0.62 |
| $85.00 Best | $63.26 | $35.05 | +$1.65 |
Directional, not prescriptive. Category, price sensitivity and how you present the threshold all move the real numbers. Chapter 6, the checkout chapter, walks through the trade in more depth.
Check first
The contribution column, not order value. On the defaults, four of six thresholds lose money.
What it means
A hypothesis: the model assumes top-off buyers stop dead on the bar and nobody abandons over it.
Monday
Pull the one input stores guess, orders that already covered delivery, from your reporting. Rerun, then ship the winner for a full month.
The formula and a worked example → The full model, the worked table, and everything it deliberately ignores.
Stay sharp on ecommerce CRO
3 short examples, 2 CRO tips, 1 teardown. Every Tuesday.
Questions about the calculators
Which one should I open first?
If nobody has agreed that testing is worth paying for, the ROI calculator. If that argument is settled and a test is about to be built, the significance calculator, on the Plan tab. The shipping threshold is a separate question that happens to come up on nearly every store.
Is anything I type here saved or sent anywhere?
No. All three run entirely in your browser. There is no request carrying your numbers, nothing stored, and no email field in front of any of them. Reload the page and your inputs are gone.
Why is there no email gate?
Gating a calculator is the exact behaviour this site argues against everywhere else. We run a conversion agency and some readers eventually hire us; that is the whole arrangement. A tool that tells you not to test is more useful to us than a lead who was never a fit.
What do all three assume?
Binary outcomes: a visitor either converted or did not, so revenue per visitor needs a different test. US dollars, though the arithmetic is currency-agnostic if you read the figures as units. And that they are models, not measurements: each one names its own simplifying assumptions on its own page.
Can I use these on a client store or in a deck?
Yes, including commercially, and you do not have to ask. Every formula is printed on the deep page beside each tool, so you can show your working rather than cite us.
One of these gave me a number I cannot act on. Now what?
That is usually the tool working. A twenty-week test duration, or a threshold that loses money at every level, is a real answer that saves you the build. If you want a second opinion on your own numbers, the free page review is a Loom and three fixes, not a sales call.
Want this done on your store?
Twenty minutes on your store and your numbers. We tell you what is leaking, what to fix first, and whether we are the right people to do it. No deck.
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