The ecommerce KPIs that matter most for CRO are conversion rate, revenue per visitor (RPV), average order value (AOV), profit per visitor and add-to-cart rate. Track all five, but judge most A/B tests on revenue per visitor, and switch to profit per visitor whenever a test touches price, discounts, shipping or bundles.
The reason is simple: a change can lift one number while quietly hurting another. The tests below, all from the LiftArchive database, show how often that happens.
Here is the short version of how the five metrics fit together:
| Metric | Formula | What it tells you | Main blind spot |
|---|---|---|---|
| Conversion rate (CVR) | Orders ÷ sessions (or visitors) | How many visits end in a purchase | Ignores how much each order is worth |
| Revenue per visitor (RPV) | Revenue ÷ visitors, or CVR × AOV | Money earned per visit | Ignores margin and discounts given away |
| Average order value (AOV) | Revenue ÷ orders | Size of the typical order | Ignores how many people buy |
| Profit per visitor | Gross profit ÷ visitors | What each visit is worth after costs | Needs cost data and more traffic to measure |
| Add-to-cart rate | Sessions with a cart addition ÷ sessions | Early buying intent | Carts are not orders |
Conversion rate: the metric everyone knows
Conversion rate is the share of visits that end in an order. Shopify defines it as "sessions that completed checkout / sessions" in its ShopifyQL sessions reference, so a shopper who visits three times and buys once counts as one conversion out of three sessions.
For context, Littledata's benchmark of 436 Shopify stores put the median conversion rate at 1.4%, with the top 20% at 2.7% or higher and the top 10% at 3.7% or higher for July 27 to October 24, 2026. Work out your own number with the conversion rate calculator.
Conversion rate is easy to understand and quick to reach significance, because it is a yes/no outcome. That is why so many tests in our database report it. Dave Diederen saw +15.61% CVR after changing a product title to a benefit-led "Achieve [outcome] with [product]" format, a change that should not affect order size much, so CVR is a fair read.
The weakness shows up when a test changes what people buy, not just whether they buy. Elevate A/B Testing reported that cutting a price by 30% produced 2x the conversion rate. That sounds like a clear win, but conversion rate alone cannot tell you whether the store made more money. You need the next metrics for that.
Revenue per visitor: the best default test metric
Revenue per visitor is total revenue divided by the number of visitors in the test. It is the same as conversion rate multiplied by average order value, so it captures both "did they buy" and "how much did they spend" in one number.
A worked example: a store gets 10,000 visitors, converts 2% of them (200 orders) and has a $60 AOV. Revenue is $12,000, so RPV is $1.20. If a test lifts CVR to 2.2% but drops AOV to $52, revenue falls to $11,440 and RPV to $1.14. The conversion rate went up and the store earned less.
RPV also makes small wins easy to price. Elevate A/B Testing moved the second product photo into first position in the gallery and saw RPV rise from £3.32 to £3.60, worth +£785 in 10 days.
Sometimes CVR and RPV move together, which is reassuring. Kasper Dolk removed a countdown timer from checkout and measured CVR +25.61% and RPV +26.18%, an estimated +€150k per month. Because both moved the same way, the result is easy to trust.
When they split, RPV is usually the better guide. Dave Diederen tested shipping price levels and found that a $9.49 shipping fee beat the control by +43.81% RPV. Dave also saw +9.17% RPV after removing the free-shipping threshold and charging shipping on every order. A test judged only on conversion rate could easily have rejected changes like these, because charging for shipping is the kind of thing teams assume will cost orders.
AOV: order size, not a goal by itself
Average order value is revenue divided by orders. Littledata's Shopify benchmark puts the median AOV at $110, with the top 20% at $298 or more over the same 90-day window. Our AOV benchmarks page covers this in more detail.
AOV is the lever behind bundles, upsells and free-shipping thresholds. Elevate A/B Testing saw +8.47% RPV from listing the biggest pack first in the bundle selector, and Milos (CROguy_) saw +15.26% RPV, about +$28,912 per month, when testing a leather strap against a mystery box as the cart drawer upsell.
The trap is treating AOV as the target. You can raise AOV by making small orders harder to place, and lose more in conversion rate than you gain in order size. Report AOV in every test so you understand why RPV moved, but do not declare a winner on AOV alone.
Offers that look generous can also backfire on revenue. Milos changed a 2-pack offer to "Buy 2 Get 1 Free" at the same price and saw -52.38% RPV, a loss of about $104,679 per month. Test offer framing the same way you test anything else.
Profit per visitor: the metric for price and discount tests
Profit per visitor is gross profit divided by visitors. Gross profit here means revenue minus product cost, and ideally minus discounts, shipping you pay for and payment fees. It answers the question owners actually care about: is this visit worth more money after costs?
RPV is fine when a test only changes layout or copy, because margin per order stays the same. Once a test changes price, discounts, free gifts or shipping, revenue and profit can move in opposite directions.
Go back to the 30% price cut that doubled conversion rate. As an illustration only (we do not have that store's cost data), imagine a $100 product that costs $40 to make. At full price, each order earns $60 gross profit. At $70, each order earns $30. Twice the orders at half the profit per order leaves profit per visitor flat, even though CVR doubled and RPV rose 40%.
Some practitioners report profit directly. Kasper Dolk tested showing a testimonial against no testimonial and measured +7.22% profit per visitor, with a 94.4% probability to beat control. If your testing tool supports product costs, use this metric for any price test.
Also watch for revenue that arrives later. Dave Diederen introduced subscriptions with the subscription option pre-checked and saw -9.17% RPV upfront but +1,295% subscription share. A short test window counts the lower first order and misses the repeat orders, so subscription tests need a longer-term value view.
Add-to-cart rate: an early signal, not the finish line
Add-to-cart rate is the share of sessions where a visitor added at least one item to the cart, per the ShopifyQL definition. Littledata's benchmark puts the median Shopify add-to-cart rate at 6.8%, with the top 20% above 12% and the top 10% at 17% or higher. Our add-to-cart rate benchmarks page has more context.
Because far more people add to cart than buy, this metric collects data faster. That makes it useful for diagnosing where a funnel leaks: a healthy add-to-cart rate with a weak conversion rate points to the cart and checkout, while a weak add-to-cart rate points to product pages.
The risk is optimizing for clicks instead of orders. Elevate A/B Testing found that a pink add-to-cart button beat green, 11.49% vs 10.81% add-to-cart. That is a real difference in carts, but whether it becomes more revenue depends on what happens after the click. Treat add-to-cart rate as a supporting metric and confirm wins on RPV.
Which one to use as the test metric
Pick one primary metric before the test starts and write it into your hypothesis. Changing the metric after you see the results is one of the most common A/B testing mistakes.
- Default: revenue per visitor. Use it for layout, copy, images, navigation, trust elements and most product page tests.
- Profit per visitor: for anything that changes price, discounts, free gifts, bundles or shipping fees.
- Conversion rate: when order value cannot change much (single-product stores, fixed-price items) or when traffic is too low to detect an RPV change. RPV varies more from visitor to visitor than CVR, so it needs more traffic for the same confidence. Check with the sample size calculator.
- Add-to-cart rate: as a secondary metric to explain results, or as a primary metric only on very low-traffic pages, with the result confirmed later.
- AOV: always as a secondary metric, never as the winner on its own.
Report the others as guardrails. A winning RPV test with a falling conversion rate is worth a second look, and the same change can behave differently across stores. Ruben (Rubendgrf) tested a countdown timer across several brands and saw about +10% CVR on a low-priced brand, flat on a high-priced brand and -9% CVR on another.
@DaveDiederen
@kasper_dolk@kasper_dolk
@elevate_ab@DaveDiederen
@CROguy_For more detail on primary and guardrail metrics, read our guide to A/B testing metrics.