E-commerce KPIs: ten metrics, one table, one decision a month

A metric is only worth something if it triggers a decision

Analytics dashboards show dozens of numbers. Most of them change no decision at all. A good KPI answers a precise question (“is my shop being found?”, “are my visitors buying?”, “does each sale make money?”) and tells you what to do when it drops.

The ten metrics below follow the equation of a profitable website: qualified visitors × conversion rate × average order value × margin, minus the acquisition cost.

Visibility: are you being found?

1. 1
   Search impressions: how many times your pages appear in Google (Search Console) and Bing (Bing Webmaster Tools). A sudden drop points to an indexing problem or a Google update.
2. 2
   Organic search clicks: the visits you earn without paying. Divided by impressions, they give the click-through rate: a low rate on a well-ranked page points to a title or description that does not make people click.
3. 3
   Traffic sources: Google, Bing and its partners, AI assistants, social networks, advertising, direct visits. On our selling sites, 13% of measured visits came from an AI assistant in October 2026.

Conversion: are your visitors buying?

1. 1
   Conversion rate: orders ÷ sessions. It is the most watched metric, and the most misleading when read on its own: it drops mechanically when you gain less qualified visitors.
2. 2
   Checkout abandonment rate: the share of started payments that are not completed. The Baymard Institute puts the average cart abandonment rate at 70.22%, an average of several dozen studies; your own trend matters more than that average.
3. 3
   Mobile share: sessions and orders on mobile. If mobile brings most of the visits but few of the orders, fix the mobile journey first.

Value: does each sale make money?

1. 1
   Average order value: revenue ÷ number of orders. French national benchmark: €62 in 2025 according to FEVAD, all sectors combined.
2. 2
   Margin per order: price minus the cost of the product, the shipping you pay for and the payment fees. This, not revenue, tells you whether the shop makes money.
3. 3
   Repeat purchase rate: the share of customers who order again within six or twelve months. A rising rate justifies a higher acquisition cost.

Cost: how much does a customer cost?

1. 1
   Customer acquisition cost: marketing spend ÷ new customers, channel by channel.
2. 2
   ROAS and break-even ROAS: ad revenue ÷ ad spend, compared with 1 ÷ margin rate. Below that threshold, the campaign loses money on the first sale.

The reading mistakes that lead to bad decisions

• Reading a rate without its volume: a conversion rate that doubles on ten visits proves nothing. Wait for enough orders before drawing conclusions.
• Comparing periods that cannot be compared: a sales month against a quiet month, a week with a bank holiday against a normal week. Compare with the same month of the previous year or with equivalent weeks.
• Believing all traffic is measured: Google Analytics only counts visitors who accept measurement. The real sales in your payment tool are the reference; the analytics tool gives the proportions and the journey.
• Adding up the sales claimed by each ad platform: Google Ads, Meta and your analytics tool can all claim the same order. Their total then exceeds your real sales.
• Judging SEO over one week: Google says the effects of a change can take from a few hours to several months. Read visibility over four to eight weeks.
• Looking at revenue without margin: higher sales of low-margin products, driven by advertising, can lose you money.

The monthly table: one page, one decision

Put these ten numbers in a simple table (a spreadsheet or a reporting tool), one month per column, and review it on a fixed date. Always read it in the same order:

1. 1
   Visibility first: if impressions collapse, nothing else matters until you have found out why.
2. 2
   Conversion next: if visits rise and orders do not, the page or the checkout is the blocker.
3. 3
   Value and cost last: if orders rise but margin does not, review the price, the basket or acquisition.
4. 4
   One action a month on the metric that slips the most, then measure its effect the following month.