Conversion rate: calculate it properly, then raise it

Definition and calculation: the formula, and three traps

“Taux de conversion” (conversion rate) is searched 3,600 times a month in France, “taux de conversion e-commerce” 110 times. The formula fits on one line: conversion rate = conversions ÷ visits × 100. A hundred purchases for ten thousand visits gives 1%.

The formula is simple; what you feed into it is what skews everything. Three traps come up on almost every site we audit:

• Counting visitors instead of visits, or the other way round, depending on the month: pick one base (sessions in Google Analytics 4) and never change it.
• Mixing sources: a visit from a brand campaign converts far better than a visit from a blog post. Calculate the rate per source and per page type.
• Forgetting what the tool cannot see: a visitor who declines measurement, or a purchase that does not fire the purchase event, drops out of the calculation. Check once a month that the purchases in your payment tool match those in your analytics tool.

Why “average conversion rates” tell you nothing

Articles that rank average rates by industry contradict one another, because they mix sites of every size, baskets of a few euros and of several thousand, brand visits and discovery visits. A rate that looks low can be excellent for an expensive product bought after several visits; a rate that looks high can hide traffic made up almost entirely of existing customers.

Our own sites are a case in point: on a small sample of 351 sessions measured in October 2026, we count a purchase rate of roughly 1.1% per session. The sample is too small to serve as a benchmark, and that is not the point: what matters is the gap between your rate before and after each change.

So the right question is not “is my rate average?” but “where am I losing visitors, and how many?”. A funnel answers it: product view, add to cart, checkout started, purchase.

Find the leaking step: the four-step funnel

Google Analytics 4 documents the recommended ecommerce events, including view_item (product view), add_to_cart, begin_checkout and purchase. Once set up, they show the step where you lose the most people. Each step has its typical causes:

1. 1
   Few product views compared with visits: visitors are not landing on the right page, or the landing page leads nowhere. Rework the title, the first sentence and the links to products.
2. 2
   Few add-to-carts compared with views: the product page does not convince. Price missing or shown late, poor photos, no proof, doubts about delivery.
3. 3
   Few checkouts started compared with carts: costs discovered late, a mandatory account, a complicated cart.
4. 4
   Few purchases compared with checkouts started: a form that is too long, a missing payment method, a technical error, a lack of trust at the moment of entering the card.

The levers that raise the rate, in order of impact

The checkout is almost always the first gold mine. Baymard Institute estimates that an average large ecommerce site can gain 35.26% in conversion by fixing only the checkout usability issues documented in its testing. Among the abandonment reasons given by the US shoppers surveyed, extra costs that are too high come first (40%), ahead of delivery that is too slow (20%), not trusting the site with card details (19%) and being forced to create an account (18%).

Speed comes right after, especially on mobile, where 61% of measured visits happen on our sites. In the “Milliseconds Make Millions” study Google commissioned from Deloitte across 37 brands, a 0.1 second gain on mobile came with 8.4% more conversions in retail. It is a commissioned study, not a peer-reviewed paper, but the direction is consistent.

1. 1
   Show the total price early: delivery costs and lead times visible on the product page, not discovered at checkout.
2. 2
   Remove the mandatory account: guest checkout, email offered rather than imposed.
3. 3
   Shorten the checkout: one click to a hosted payment page, Apple Pay and Google Pay, no unnecessary field.
4. 4
   Speed up the buying pages: light images, no superfluous script, a button visible without scrolling on mobile.
5. 5
   Rewrite the product pages that already get traffic: an answer in one sentence, real photos, verifiable proof.
6. 6
   Add true trust signals under the button: lead time, returns, secure payment.

Service websites: the conversion rate of a quote request

For a tradesperson, a practice or an agency, the conversion is a quote request, a call or a booking. The calculation is the same: enquiries ÷ visits. You just need to count all three channels, because many customers call without filling in the form.

The levers run in parallel: one page per service and per area, the answer in the first sentence, a tappable phone number at the top of the page, a form of three to five fields, real projects and reviews, and a response time that is stated and kept.

Measure the effect of a change without fooling yourself

A rate that moves from one week to the next proves nothing: seasons, campaigns and chance are enough to make it vary. Compare periods of the same length, on the same traffic source, and note the date of each change.

A/B testing is the clean method, but it needs enough conversions to reach a conclusion. On a small site, fix what is obviously broken first (a button invisible on mobile, a hidden fee), then watch the trend over several weeks.