Ecommerce KPIs are the indicators that tell you whether an online store is selling well, at what cost and at what quality. They are a short list of numbers, chosen because each one leads to a decision: what to promote, where to spend, what to fix. The common problem is not having too few. It is having plenty with no way to connect them. This guide starts with a simple way to do that, then moves to the ten that matter, each with its formula and what to check when it gets worse.
The most useful way to read KPIs: a tree
Sales are not a number on their own. They are the product of three. Sales = sessions (visits) × conversion rate (the share who buy) × average order value. When sales move, one of those three levers is the cause, and knowing which one tells you where to act.
An example with made-up numbers. Last month: 20,000 sessions, a 2.0% conversion rate, so 400 orders, with an average order value of €55: sales of €22,000. This month: 22,000 sessions, a 1.7% conversion rate, so 374 orders, with an average order value of €58: sales of €21,692.
Sales fell 1.4% even though traffic grew 10%. Looking only at sales, it seems a slightly weak month. Looking at the tree, the point is clear: conversion dropped 15%, while average order value went up. The extra traffic is probably less qualified, or something in the site or checkout slowed purchases down. The first check is not advertising or prices. It is working out which source and which device saw conversion fall.
The ten KPIs: formula and first check
This table summarizes the core indicators. For each one you get how it is calculated and what to look at first if it gets worse.
| KPI | Formula | If it gets worse, check |
|---|---|---|
| Net sales | Sales − discounts − refunds | Discounts applied and refunds by product |
| Number of orders | Orders in the period | Traffic and conversion, before sales |
| Average order value (AOV) | Sales ÷ orders | Product mix, discounts, free shipping threshold |
| Conversion rate | Orders ÷ sessions | The source and device where it dropped |
| Customer acquisition cost (CAC) | Marketing spend ÷ new customers | Channels where cost rose and traffic quality |
| Cart abandonment rate | 1 − (orders ÷ carts created) | Shipping costs, payment options, checkout steps |
| Repeat purchase rate | Customers with 2+ orders ÷ total customers | Time between first and second order, and entry products |
| Customer lifetime value (LTV) | Margin per order × average orders per customer | Repeat rate and margin, not just sales |
| Refund and return rate | Refunds ÷ sales | Individual products, not the average |
| Margin by product | Price − cost − shipping − fees | Discounts and products that sell a lot at low margin |
The calculation many skip: what a customer is worth
CAC alone says nothing: is €30 per customer a lot or a little? It depends on what that customer is worth. The calculation has to run on margin, not on sales, because sales don't pay the costs.
An example with made-up numbers. An average order is worth €58. Take off the cost of the product (€23), shipping paid by the store (€5) and payment fees (€2): that leaves €28 of margin per order. If each customer places 1.5 orders on average, their value at margin is €42. With a CAC of €30, each customer brings €12 beyond the cost of acquiring them.
Two practical consequences. First: with €28 of margin per order and a CAC of €30, the first order does not cover the acquisition cost. You need the second, and repeat purchase rate becomes the KPI that decides whether advertising works. Second: if you use sales instead of margin, the customer's value looks like €87 (58 × 1.5) and the picture looks far better than it is.
There is no ideal ratio that fits everyone. The useful reference is your own: compare a customer's value at margin with what it costs to acquire them, channel by channel, and see how long it takes to break even.
How to read a KPI without being misled
- Beware of small numbers. Ten orders against twelve is 20% growth, but it is noise. Four hundred against 420 is 5% and starts to mean something. The lower the volume, the longer the period you should look at.
- Always compare. Against the previous period and the same period last year: seasonality moves numbers more than anything you do.
- Segment. An average hides differences. Look by channel, by product and by new versus returning customer.
- Read them in pairs. Conversion and average order value, CAC and LTV, sales and refunds. A number that improves at the expense of its partner is not an improvement.
- Keep definitions stable. If "sales" means different things in two reports, the discussion shifts from the data to the definitions. Write them down once.
Which KPIs to pick by stage
| Stage | Track | Postpone |
|---|---|---|
| Starting from scratch | Net sales, average order value, conversion, repeat purchase | LTV, margin by channel |
| Growing with advertising | CAC, margin by product, cart abandonment | Cohort details |
| You have a customer base | Repeat rate, LTV, refunds by product, share of returning customers | Indicators nobody uses to decide |
A simple rule: an indicator joins the group when it has a reader and a question. If nobody uses it to decide, it is a statistic.
Where the data lives and why that is a problem
Orders sit in the ecommerce platform, traffic in analytics tools, advertising spend in ad platforms, costs in an ERP or a spreadsheet. That is why many teams rebuild KPIs by hand each week, and why the KPIs that need several sources, such as CAC and true margin, are the most neglected. It works for as long as someone has the time.
KPIs and questions to your data
Once the numbers are lined up, the most useful questions are follow-ups. For example: how does average order value differ between new and returning customers? Which products have the highest refund rate? How have sales changed compared with the same period last month? With a tool you can ask in plain language, the step from the number to the "why" is shorter. One caution stands: automatically generated answers should be checked. And a useful note on limits: tools that work only on store data don't know your advertising spend, so a KPI like CAC also needs data from the advertising platforms.
Where Opalca View fits
Opalca View is a set of ready-to-use dashboards for sales, products and customers, with automatic reports. Connect your Shopify store and the indicators for sales, average order value, products and new versus returning customers are already set up. The Growth and Enterprise plans also include Ask Opalca, to ask your data questions with AI. Shopify is the connector that is live today; more connectors are on the way.
Have a question or want to talk it through? Contact us. You can see the plans on the Pricing page. For a broader guide to reading your store's numbers: Shopify dashboard: which metrics to track.
Frequently asked questions
How many KPIs should an ecommerce business track?
Few. Four to start (net sales, average order value, conversion, repeat purchase), then add the others when they serve a decision. A short view you read every week beats a huge dashboard nobody opens.
What is the most important KPI?
It depends on the stage. A store growing through advertising often watches conversion and CAC, while one with an established customer base watches repeat rate and margin. Either way, no KPI should be read alone: the tree of sales, conversion and average order value is a good way to connect them.
How often should they be checked?
Sales, orders and conversion every week. LTV and repeat purchase rate make more sense over longer periods, usually monthly or quarterly.
Why is margin more important than sales?
Because sales don't pay the costs. Two products with the same sales can have very different margins, and pushing the wrong one grows revenue without growing profit.
How do I calculate CAC with several channels?
Divide each channel's spend by the new customers it brought, and then look at the overall average cost too. The average can hide a very expensive channel offset by a very cheap one.