E-commerce website conversion rate: what makes it vary from page to page

The e-commerce benchmark available is 2.4% conversion into a qualified lead or a sale for retail and e-commerce (Ruler Analytics, 2026), and 4.8% of paid clicks converted in e-commerce SEA (Odiens, 2025): two different measurements, one on qualified leads and sales, the other on bought clicks alone. They place your figure without saying whether it is good, which is settled by the criteria for a good conversion rate.

For the e-commerce director, the e-commerce website conversion rate is played out first in the intent of the queries captured, then on the category page and the product page; the funnel weighs at the end of the journey.

What the conversion rate of an e-commerce site measures

The conversion rate sets the number of orders against the number of sessions, over a given period. The numerator is the order, not the buyer: a customer who orders twice in the month counts twice. Three points change everything on a store:

• The unit is the order, not the product: an order of three items counts as one, and it is the average order value and the margin that count the other two.
• The scope is the catalogue: hundreds or thousands of pages each have their own rate, and their average resembles none of them.
• The denominator is the session: it is the "session conversion rate" in GA4, the convention across the whole cluster.

The formula is simple: (number of orders ÷ number of sessions) × 100. If you sell smartphones and over the month of January you record 250 orders for 10,000 sessions, your conversion rate is 2.5%.

Keep this definition constant: a change of unit or of tagging opens a new series, to be read beside the old one rather than recalculated. The variants in vocabulary ("conversion rate", "transformation rate") are set out in the definition of the conversion rate.

Reading the figure before judging it: by channel, by device, by category

Google Analytics (or your reporting tool) calculates this rate and how it changes over time, provided you declare the order as a conversion. You still have to read it broken down, along five cuts:

• Source/medium: SEO, SEA, e-mail.
• Device: mobile, desktop, tablet.
• New against returning: first purchase or a customer coming back.
• Product category: the cut specific to retail.
• Country: delivery, prices and payment methods change.

Then cross it with the average order value and the margin: the objective is to optimize profitability, not only the conversion percentage.

The retail funnel: where the loss occurs

A single rate never says where the loss occurs. For that you need the retail funnel, in its order: product view, add to basket, checkout started, purchase. The intermediate steps are micro-conversions, tracked separately and outside the rate.

On 10,000 product views, 800 adds to basket, 400 checkouts started and 200 purchases, you read three passages (8%, 50%, 50%) and the most costly step is plain to see. A rate per product view of 2% would never have shown it, and it does not compare with the rate per session.

Add to those the refunds, which take away orders already counted, and the margin: a rise in purchases on heavily returned or low-margin products reads as a victory in the funnel and as a loss in the accounts.

Product category: the cut specific to retail

A store selling accessories at fifteen euros and devices at two thousand does not have one rate, it has two, and their average describes neither. Price and length of consideration vary far more from one category to another than from one month to the next: a product page is therefore compared with its category, not with the site average.

Yet three different measurements carry the name "rate of a category", and they do not give the same result:

• Sessions that entered on the category page, set against the orders that followed: it judges the page itself and ignores the buyers who entered through a product page.
• Sessions that viewed at least one product page in the category, set against the orders: it measures interest in the range, whatever the way in.
• Sessions set against the orders containing at least one item from the category: it measures what the range sells, whatever the way it was found.

Choose one, write it beside the figure, and do not change it again: the gap between those three readings is of the same order of magnitude as the gaps you are trying to interpret.

The breakdown by category also explains an observation often misread: the overall rate falls while every category improves, because the traffic mix has shifted towards the least converting categories. This mix effect, and how to read the overall rate in general, are covered in the site conversion rate.

Traffic or pages? The fork that decides what comes next

When the rate falls, one question decides what comes next: has the traffic changed, or have the pages? The most expensive mistake is to redo product pages when the problem is upstream.

On the traffic side, the causes are known:

• less qualified traffic: campaigns too broad, ranking on informational queries when the page expects a buyer;
• a stock shortage on the products that carried the orders;
• a rise in delivery charges;
• a gap between what the page promises in the search results and what it delivers once opened.

Analyse the fall by channel and by device first of all: it is the fastest test. This dimension weighs heavily on a store, since the share of e-commerce traffic coming from Google organic search is 43% (SEO.com, 2026), a measurement the SEO statistics document.

Two comparisons are then enough to settle between traffic and pages:

That all your content should deteriorate in the same month is unlikely; a common step in the funnel is far more likely, and is checked in a few minutes.

Some of the gaps cannot be fixed from the content: the order funnel, usability and speed belong to the product team, UX and development, with whom the division is settled following the method of conversion rate optimisation. These levers weigh on the end of the journey; the qualification of the traffic and the product page decide before.

Category pages and product pages: the content that decides

The funnel only concludes: a store converts first because the visitor has found the right product, understood what they were buying and known where they stood on price, delivery time and returns. Those answers are written on your pages, four types of which carry most of the work, each with its own intent and its own measurement:

What a category page must carry

This is the page that receives the most organic traffic in a store, and the most profitable lever in the catalogue: the visitor roughly knows what they want, not yet which one. A category page aligned with purchase intent carries three things:

• Its selection: featured products, default order, sorting and filtering criteria that match the way buyers decide, not the way the catalogue is arranged internally.
• Text that helps decide: a few lines on what the category covers, what sets the ranges apart and what to decide on, rather than an introduction written for search engines.
• Its internal linking: towards high-margin products, towards the comparison that helps choose and towards the buying guide where one exists.

What a product page must carry

The product page is the decision page: is this really the product, what sets it apart from the one next to it, when will it arrive, and what if it does not suit? It answers through concrete elements:

• a description that describes instead of praising;
• specifications comparable from one page to another;
• customer reviews and sufficient visuals;
• delivery, return and guarantee information displayed on the page rather than in the site footer.

On reviews, one source states that adding them to a product page doubles the conversion rate (SiteW, 2026). That order of magnitude, published without details of method, says where to put the effort, not what your catalogue will return.

On your own site, compare product pages equipped with reviews or a video against comparable product pages without them, at similar price and traffic mix, on enough orders for the gap to be readable.

To these is added structured data (price, specifications, reviews, availability), which serves display in the results as much as pick-up by generative engines. On thousands of pages, what can be industrialized is the template: mandatory fields, format of the specifications, markup. What stays human is the top of the catalogue, the products that make the numbers.

Category, product page, funnel: three levels to read separately

Each level has its own measurement, its own warning signal and its own action, and they do not belong to the same teams:

Prioritizing: which catalogue pages to start with

Four criteria designate the pages to start with, and it is their crossing that decides:

• the qualified traffic received;
• the margin potential;
• the gap between the rate of the page and that of its category;
• the minimum volume of orders below which a page rate means nothing.

The third criterion is the most useful and the least used. A product page converting at 1.1% when its category is at 2.8% deserves a closer look, but the gap does not prove that its content is at fault: within one and the same category, price, availability and traffic mix differ from one product page to another, and each of the three can open a gap of that size. Check them before concluding anything about the page.

Conversely, a product page at 1.1% in a category also at 1.1% says nothing: it is the category that needs looking at. That gap prioritizes a whole catalogue without opening a single page: it designates what needs examining, not what needs rewriting.

The fourth criterion avoids chasing noise. On a page that records a handful of orders, one order more or less moves the rate by several points: set a floor volume of orders before any reading, and handle what falls below it by product family.

The order of treatment follows from the rest:

• Categories first: a category fix benefits every product page it feeds.
• Then the product pages with a large gap and a high margin: they are the ones that bring in the most per fix.
• The rest by template: fixing the model rather than the pages is the only way to move thousands of URLs without devoting a whole team to it.

Generative AI in the buying journey: comparison, citation, reassurance

E-commerce is particularly affected by generative AI: consumers now use ChatGPT and Perplexity to compare products, read reviews and obtain personalized recommendations before buying. The visitor arrives further along in their decision. Three effects follow:

• Product comparison by AI: engines generate comparison tables from multiple sources, and product pages with structured data are cited more often in them.
• Reassurance through the content of the page: when a buyer asks the AI about the reliability of a site, the answer draws on what your pages display (reviews, guarantees, returns policy).
• Product SEO and citability: pages optimized for SEO are also the ones most cited by AI; FAQs, buying guides and structured comparisons serve both.

These three points converge on one and the same move, an inexpensive one: structuring product data in schema.org. Declaring price, availability, specifications and reviews properly changes nothing about the look of the page, but makes it usable by an engine, generative or not.

Tracking extends what you already do: isolate the traffic coming from answer engines and look at the conversion rate of the product pages they cite. Attributing an order to that channel is an attribution problem in its own right.

Placing yourself: your history by category first, the sector benchmark second

What is a good conversion rate for an e-commerce site? It depends on your business model, your sector and your categories, and it is economic first of all: it is good if it makes the acquisition cost compatible with the margin per order and if it improves against your history. To place yourself, proceed in this order:

• Your history, at constant definition, category by category, over at least twelve months: that is where a gap of a few tenths of a point becomes interpretable.
• Your segments: channel, device, type of page, with the same cuts from one report to the next.
• Published benchmarks, last, with an identical denominator, as an order of magnitude and never as a target.

An outside benchmark is looked for by sector, never across all sectors: that is what the conversion rate by business sector brings together. For benchmarks by channel, SEA included, see the average website conversion rate. In both cases, check what the source measures and on what population: a value counting qualified leads does not compare with a value counting orders.

Continue reading

Depending on what you need to do with your rate once it is broken down, five pages take over:

• Your pages are prioritized and you are launching the project: where to start, in what order and how to measure the effect of what you change, see optimizing the conversion rate.
• You are setting up the tracking apparatus: channel, entry page, device, campaign and indicators coming from generative engines belong to the conversion rate KPIs.
• You have to attribute your orders to the organic channel: saying what share of them belongs to SEO, and not only how many you record, is a matter of SEO conversion.
• Your categories are not ranking: the problem is upstream of the page, in the architecture and the visibility of the catalogue, which e-commerce SEO deals with.
• You do not yet have a tool that calculates your rate by category: the setup and how to read the reports are detailed in Google Analytics.

To produce and enrich category pages and product pages across a whole catalogue (templates, comparable specifications, reassurance content), the content production module absorbs the workload that page-by-page tracking cannot carry.

Frequently asked questions

What is a good e-commerce conversion rate?

The published benchmark is 2.4% conversion into a qualified lead or a sale for retail and e-commerce (Ruler Analytics, 2026). For paid clicks alone, e-commerce SEA converts 4.8% of clicks (Odiens, 2025). These two values place your rate without setting a target: a good rate makes your acquisition costs compatible with your margin and improves against your history.

How do you calculate the conversion rate of an e-commerce site?

The conversion rate is calculated by dividing the number of orders by the number of sessions, then multiplying by 100: 250 orders for 10,000 sessions give 2.5%. An order containing several items counts as a single conversion. Keep this definition constant so that the series stays comparable from one month to the next.

What is the average conversion rate of an e-commerce site?

The first benchmark is your own history, category by category, because it is the only one whose method you control. The level to expect depends on the model, the price, the length of consideration and the share of visitors who came to find out. A published benchmark is then looked for by sector, with an identical denominator.

Why is my conversion rate falling while traffic is rising?

Most often, no page has moved: it is the composition of the traffic that has changed. Campaigns too broad, informational queries, a stock shortage on the best sellers or a rise in delivery charges bring the rate down. A reading by channel, by device and by category isolates the breaking point.

What are the best SEO levers for improving the e-commerce conversion rate?

The most effective SEO levers are category pages aligned with purchase intent and product pages enriched with reviews, comparable specifications and structured data. Internal linking towards high-margin products and reassurance content complete the set. The effect is checked segment by segment.

How do you segment the conversion rate in order to decide better?

The conversion rate is segmented by source, device, new against returning, product category and country, then crossed with the average order value and the margin. On a store, the cut by category is the most decisive: it alone gives each product page a relevant reference point.

Which KPIs should be tracked for an e-commerce site?

The essential KPIs for an e-commerce site are the conversion rate, the basket abandonment rate, the average order value, the customer acquisition cost, the customer lifetime value, the product return rate and revenue per visitor. Each is segmented by acquisition channel to inform budget allocation.