Analytics for Online Stores: What to Track and What to Ignore

Open any analytics tool and you are handed hundreds of numbers. Sessions, engagement rate, average session duration, exit pages, new vs returning, events per session — page after page of them. The stores that grow are not the ones watching the most dials. They are the ones who have decided, in advance, which five or six numbers actually change a decision, and who ignore the rest on purpose. This guide is about that decision: for a PrestaShop store specifically, what to track, what to ignore, and where each number really lives in your back office. It is not a setup tutorial and not a tool comparison — those have their own homes, linked below, so this stays focused on the one question in the title.

Why "track everything" is the wrong instinct on PrestaShop

PrestaShop hands you data from two directions at once, and that is what makes merchants drown. On one side there is your own back office: Stats (the AdminStats controller) ships with the platform and gives you Best-selling products, Catalog statistics, Registered customer info, Visitors origin and a dozen more panels straight from your database — no JavaScript, no third party. On the other side there is whatever you bolted on the front end: GA4, a Pixel, a heatmap tool, a tag manager firing all of them. The two rarely agree, because PrestaShop's native Stats counts what hit your server while GA4 counts what fired in a consented browser — and the gap between those two is normal, not a bug.

So the first act of discipline is not adding a tool. It is deciding what question you are answering, then reading the one source that answers it cleanly. Revenue truth comes from your orders table (back office, never a tracker that ad-blockers and consent banners quietly eat). Behaviour truth — where people hesitate, rage-click, abandon — comes from front-end analytics. Mixing the two into one dashboard and treating every line as equally important is how a Monday review turns into an hour of staring with nothing decided.

The metrics that actually drive a decision

A metric earns its place only if a bad reading would make you do something different this week. By that test, most stores need six.

1. Conversion rate — but segmented, never as one number

Your store-wide conversion rate (orders ÷ visits) is close to useless on its own, because it averages away the only thing worth knowing: which traffic converts. PrestaShop's own Stats → Visitors origin (available when the native stats modules are installed and enabled) gives you a first, server-side cut of this. For the cut that actually drives spend — conversion rate by channel, by device, by landing page — you want GA4, and the specific reports and pitfalls (consent gaps, the dreaded "direct" bucket) are worth getting right. We cover exactly which GA4 numbers earn attention in GA4 for PrestaShop: the metrics that actually matter. The rule of thumb: if desktop converts at 3% and mobile at 0.8%, your "fine" 2% average is hiding a mobile problem the average will never let you see.

2. Average order value

AOV is the cheapest lever you have, because lifting it needs no extra traffic and no extra conversion — just more value per order you already win. PrestaShop surfaces the raw material in Stats → Best-selling products and in the Catalog statistics panel, and your order list exports the rest. A store at 2% conversion and €80 AOV earns the same as one at 2% and 50% more traffic at €53 AOV — except the first store paid nothing for that traffic. Cross-sells, free-shipping thresholds and bundle pricing are where AOV moves; watch it monthly, not daily, because it is noisy on small order counts.

3. Cart and checkout abandonment

This is the metric PrestaShop is unusually good at exposing, and most merchants never look. The back office keeps abandoned carts under Sell → Orders → Shopping Carts — every cart with no matching order, with the customer, the contents and the timestamp. That is your raw abandonment data, sitting in ps_cart with no order row in ps_orders. The number itself (industry talk lands around 70%, treat that as directional, not a target) matters less than the shape: a cart that dies at "add address" is a different problem from one that dies at "see shipping cost." If checkout is your leak, the fix lives at the checkout, not in the analytics — start with a GA4 setup that tracks the begin_checkout and purchase events so the funnel is even measurable, then watch where the steps drop.

One caution specific to PrestaShop in 2026: not every abandoned cart is a lost human. Automated shopping agents now add items and leave, inflating your cart count with traffic that was never going to buy. Before you panic at a rising abandonment rate, rule that out — see what Google Storebot doing in your store means for your numbers.

4. Revenue by traffic source

Not visits by source — revenue by source. Organic search routinely converts at several times the rate of cold social traffic, and email higher still, so a channel that looks small in a visitor chart can be your biggest earner. The honest version of this number needs proper attribution, which on PrestaShop means UTM-tagging every campaign link and letting GA4 assign credit rather than dumping it all into "direct." Getting that plumbing right without editing template files is exactly what Google Tag Manager for PrestaShop is for.

5. Customer acquisition cost vs first-order margin

If it costs €25 to win a customer whose first order nets you €20, you are buying losses unless they come back. PrestaShop won't compute CAC for you — the spend lives in your ad accounts — but it gives you the other half: Stats → Registered customer info and the customer's order history tell you whether buyers return. Pair the two and CAC stops being a vanity reassurance and becomes a sustainability test.

6. Repeat-purchase / returning-customer rate

The number that decides whether your whole model holds together. A store leaning entirely on first orders is renting customers from ad platforms; one with a healthy repeat rate owns them. PrestaShop's Stats → Best customers and the new-vs-returning split in your customer data give you the baseline. This is also where store-level reporting outgrows the native Stats panels — if you need cohort-style views, margin by customer, or profit (not just revenue) over time, that is a reporting job rather than an analytics one, and our Financial Revolution advanced reporting module exists precisely so you can read those numbers from the back office instead of exporting CSVs into a spreadsheet every month.

What to ignore — at least until those six are healthy

Ignoring metrics is not laziness; it is the only way the six above stay legible. These are the usual time-sinks.

None of these are "wrong" data — they have diagnostic uses once you are chasing a specific question. The trap is treating them as a daily scoreboard. They move, you feel something, and nothing changes in the store.

Numbers tell you what; you still need why

Every metric above answers what is happening. None of them tells you why a particular page leaks. For that you need to watch real sessions — where people click on something that isn't a link, where they scroll past your add-to-cart on mobile, where a form field stalls them. That is a different category of tool entirely, and on PrestaShop the two worth knowing are covered in their own pieces: Microsoft Clarity (free, heatmaps and recordings) and Hotjar (heatmaps, recordings and on-site surveys). The discipline is the same as with metrics: watch a handful of recordings a week with a question in mind, not a hundred out of curiosity.

A note on where your data lives

One decision quietly shapes everything above: whether your behavioural analytics sits on Google's servers or your own. For many EU PrestaShop stores, consent banners eat a meaningful slice of GA4 data, and data-sovereignty matters to customers and to your DPO. If that describes you, self-hosting analytics is a real option rather than a fringe one — the trade-offs (data ownership and GDPR posture vs Google's reporting depth) are laid out in Matomo vs Google Analytics. Whichever side you land on, the list of six metrics that matter doesn't change — only where you read them does.

The weekly routine that makes any of this pay off

Data you don't look at on a schedule is data you don't have. Block thirty minutes every Monday and read, in order: revenue this week vs last (trend, not panic over a daily wobble); conversion rate by channel and by device (anyone suddenly worse?); abandoned carts in Sell → Orders → Shopping Carts (a spike means something broke at checkout); best-selling products (are your winners getting shelf space?); and returning-customer rate (is the base growing or are you just renting?). Five reads, thirty minutes, and it catches problems while they are still one bad week instead of a bad quarter.

Two rules keep that routine honest. First, never decide on a small sample — fifty visitors is noise; wait for 200–300 data points (and ideally 30 days) before you conclude a change worked or didn't, or you will revert something that was actually winning. Second, always compare like to like: this week to last week, this month to last month — never a random range, because Tuesday traffic simply isn't Saturday traffic.

That is the whole craft. Analytics is not about owning the most dashboards; it is about answering the same handful of questions every week and acting on the answers. Are we making money? Where is the good traffic coming from? Where are we losing buyers? What should we do more of, and what should we stop? PrestaShop already hands you most of the raw material — in the Stats panels, in your carts and orders tables — and the front-end tools fill the behavioural gaps. The numbers were never the problem. The discipline of ignoring the ones that don't decide anything is.