Most online stores track dozens of metrics and act on almost none of them. Dashboards fill with sessions, pageviews and follower counts while the questions that decide whether the business grows go unanswered. The fix is not more data. It is a short, defensible list of KPIs that connect day-to-day work to revenue and margin.
Vanity metrics versus decision metrics
A useful test for any metric is simple: if this number moved, would you change what you do tomorrow? If the honest answer is no, it is a vanity metric. Total sessions, social followers, email list size and raw pageviews usually fail this test. They feel like progress but rarely change a decision.
Decision metrics share three traits:
- They tie to money, either revenue or cost.
- They can be influenced by a team’s actions within a quarter.
- They expose a problem when they move the wrong way.
Keep the vanity numbers if you like them for context, but never put them on the dashboard your team checks every morning.
The core commercial KPIs
Start with the metrics that describe the health of the trading engine. For most stores, these four carry the most weight.
Revenue per visitor (RPV)
RPV is conversion rate multiplied by average order value, expressed per session. It is the single most honest top-line efficiency number because it refuses to let you cheat. A conversion-rate “win” that quietly shrinks basket size shows up immediately in RPV, where it would hide if you tracked conversion alone. We treat it as the headline number for most optimisation work; see why revenue per visitor beats conversion rate for the full argument.
Conversion rate, segmented
A blended site-wide conversion rate hides more than it reveals. Always segment by device, by new versus returning, and by traffic source. The pattern we see most often is a desktop rate that looks healthy masking a mobile rate that is half as good, which is usually where the easy money sits. Our guide to the mobile conversion gap covers how to read this.
Average order value (AOV)
AOV reflects merchandising, bundling and the quality of cross-sell. It is also one of the few levers you can pull without spending more on acquisition. Track it alongside units per order so you can tell price-mix changes from basket-building.
Gross margin after variable costs
Revenue is a flattering number. Contribution margin after cost of goods, payment fees, shipping and returns is the one that pays salaries. Many stores discover that their best-converting category is also their thinnest on margin once returns are counted.
Acquisition and retention KPIs
Trading efficiency means little if you overpay for customers or lose them quickly.
- Customer acquisition cost (CAC): fully loaded, including agency fees and creative, not just media spend.
- Contribution-margin payback period: how many months of a customer’s margin it takes to recover CAC. Under twelve months is comfortable for most categories; longer means you are funding growth from the balance sheet.
- Repeat purchase rate and time to second order: the second order is the strongest early signal of a real customer. Watching the gap between first and second purchase often predicts retention better than any survey.
- Customer lifetime value (CLV): best paired with CAC as a ratio rather than read alone. If you have not modelled this yet, start with predicting customer lifetime value.
A frequent mistake is optimising CAC in isolation. Cheap traffic that never returns is expensive traffic in disguise.
Experience and operational KPIs
These sit upstream of revenue and explain why the commercial numbers move.
- Site search usage and search-led conversion. Visitors who search convert at several times the site average in most stores, which makes search one of your highest-leverage surfaces. Our piece on site search as a converting channel explains how to instrument it.
- Add-to-cart and cart-to-checkout rates. Splitting the funnel here tells you whether you have a product-page problem or a checkout problem.
- Return rate by category. Returns quietly erase margin and signal mismatched expectations. High returns in a category are often a merchandising or content failure, not a product fault.
- Support contact rate per order. Rising contacts per order usually means something broke upstream, in delivery promises, stock accuracy or page clarity.
Building a KPI tree, not a list
A flat list of metrics invites cherry-picking. A KPI tree forces logic. Put your one north-star outcome at the top, typically contribution margin or RPV, then break it into the inputs that drive it, then break those into team-level operational metrics.
A simplified tree might look like this:
- Contribution margin is driven by RPV, traffic volume and margin rate.
- RPV is driven by conversion rate and AOV.
- Conversion rate is driven by add-to-cart rate, checkout completion and page speed.
Now every team can see how their daily work rolls up to the number the board cares about. The merchandising team owns AOV and return rate. The performance team owns CAC and payback. Nobody is left optimising a metric that does not connect to anything.
This structure also makes prioritisation honest. When you draft an eCommerce AI roadmap, each initiative should name the node on the tree it is meant to move, with a target.
Common pitfalls
- Averaging away the signal. Blended numbers hide the segments where the problem lives. Always be ready to split by device, source and customer type.
- Tracking rates without volumes. A 40% conversion rate on twelve sessions tells you nothing. Show the denominator.
- Ignoring lag. Returns, repeat purchases and CLV mature over weeks or months. Judging them on last week’s cohort will mislead you.
- Attribution overconfidence. Channel-level revenue claims are softer than they look as tracking degrades. Treat them as directional and read our take on privacy-first attribution.
- Too many KPIs. If your dashboard has thirty numbers, it has none. Aim for a single screen that a busy founder can read in under a minute.
A practical starting set
If you are rebuilding your reporting from scratch, this set covers most stores without overwhelming the team:
- Revenue per visitor (overall and by device)
- Contribution margin and margin rate
- Conversion rate by device and source
- AOV and units per order
- CAC and contribution-margin payback
- Repeat purchase rate and time to second order
- Return rate by category
- Search usage and search-led conversion
Get these clean and trustworthy before adding anything more advanced. Sound measurement also depends on the plumbing underneath; weak tracking produces confident but wrong numbers, so it is worth reading eCommerce data foundations before you invest in dashboards.
Where this leads
Good KPIs are not about reporting more. They are about arguing better, prioritising honestly and knowing within a week whether a decision worked. A tight KPI tree turns scattered effort into compounding progress, because everyone can see what they are responsible for and how it adds up.
If you want a second pair of eyes on which numbers should sit at the top of your tree, our data insights team works with online retailers on exactly this. Get in touch and we will help you cut the dashboard down to what actually matters.