You installed analytics. Now what? — 3 reasons GA4 does nothing for most ecommerce stores

Ask an ecommerce store owner “Do you use analytics?” and most will say “It’s installed.” The GA4 tag went in from the cart’s settings page. They open it now and then. There are numbers on the screen. But if they’re honest, they’ve rarely decided anything because of it.

That’s not laziness. I kept analytics “installed but unused” on my own store for years. To be precise: I’d open it, look at the pageview count, and close it. Over and over. This post explains the three reasons analytics settles into a state where it does nothing for an ecommerce business, and then gives you a 15-minute weekly routine to get out of it.

Reason 1: “Looking” became the goal

Analytics is usually treated as a tool for looking. It’s actually a tool for deciding. The screen answers nothing unless you arrive with a question.

You open it thinking “how did this month go?” Sessions are up — nice. Sessions are down — worrying. You close it. Used this way, analytics is a weather forecast: you can watch it all day and nothing you do changes.

Having a question looks like this:

• Revenue dropped versus last month. Was it fewer visits, a lower conversion rate, or a smaller average order? Which one?
• The new product page — what percentage of people who see it add to cart? Higher or lower than the others?
• If I paused ads, which traffic source’s revenue disappeared?

One question narrows the screens you need to one or two. No question means scrolling through all of them and closing the tab without deciding anything.

Reason 2: Too many metrics, and the ones that move revenue are buried

GA4 is a general-purpose tool. It’s built to work for media sites and apps as much as stores, so the numbers that matter for ecommerce sit next to numbers that don’t, at equal weight. Engagement rate, event count, average engagement time — all meaningful, none of them directly tied to revenue.

Ecommerce revenue reduces to three numbers multiplied together:

Revenue = Sessions × Conversion rate × Average order value

If revenue moved versus last month, one of these three moved. Look at those three first, pick the one that moved, then split it by a dimension: traffic source, device, product, landing page, or day-of-week × hour.

For example, if conversion rate fell, split it by mobile vs desktop. If only mobile is dramatically low, the cause is almost certainly page speed or how much typing checkout demands. If average order value fell, split it by new vs returning customers. If new customers grew and AOV shrank, that may not be bad at all — it might mean your acquisition is working.

Three numbers and five dimensions. That’s genuinely all an ecommerce store needs to watch. GA4 can produce them, but they live on separate screens and you repeat the same clicks every time. I think that’s the single biggest reason people stop opening it.

Reason 3: The loop of notice → try → check never closes

This is the biggest one.

Say you do have a question, look at the numbers, and notice “mobile is slow, that’s probably why mobile conversion is low.” You compress the images. For most stores, the improvement ends here — because nobody checks whether it worked.

When you don’t check, two things happen. If it worked, you never find out, and you miss the chance to apply the same fix to other pages. If it didn’t, all that remains is the memory of “we did that,” so the next time the same problem appears you dismiss it with “we already tried.” Either way, the motivation to keep using analytics erodes.

Closing the loop means:

1. Notice — split a number by a dimension and find a gap versus what you’d expect (mobile CVR is a third of desktop)
2. Decide — form one hypothesis and pick one change (compress product images to bring LCP down)
3. Try — do it, and write down the date
4. Check — four weeks later, look at the same number in the same dimension (did mobile CVR rise? is the rise bigger than chance would explain?)

When step 4 runs, your hypotheses in step 2 get sharper next time. That’s what “analytics that means something” looks like. Conversely, a store that only does steps 1–3 drifts back to “installed but unused” no matter how powerful the tool.

A 15-minute weekly routine

Before switching tools, here’s a routine you can run in the GA4 you already have. Same weekday, 15 minutes.

First 5 minutes: compare the three numbers to last week. Sessions, conversion rate, average order value. Pick the one that moved. If all three moved, pick the one with the largest effect on revenue.

Next 5 minutes: split that number by one dimension. Source, device, product, landing page, or day × hour. Just one. Stop as soon as you find the dimension with the biggest gap.

Last 5 minutes: decide one change and write it down with the date. “Sept 10 — compressed product images on mobile.” Then put a reminder on the calendar for the same weekday four weeks out, to look at the same number in the same dimension.

That’s it. The first week you can’t “compare to last week,” so just start by writing down this week’s numbers.

Honestly, though

Keeping this routine up every week is hard. I broke it many times on my own store. Walking through five GA4 screens, splitting by dimension, digging out the number from four weeks ago — in a busy week, it always slips.

So I built something that runs the look → decide → try → check loop on the tool’s side. Every morning, NextRise Analytics splits your numbers by dimension, finds the gaps against expectation, and turns them into cards: where to fix, how, and roughly how much it’s worth. Each recommendation carries an estimate you can reproduce by multiplying the numbers on screen (here’s how we calculate it). And four weeks later it puts the number from when it flagged the issue next to the current one, and checks whether it worked — automatically.

If you’re on GA4 today, keep it. Adding one line of script starts measuring from today, and if you want, you can import the period before installation from GA4 once, so year-over-year is there from day one.

Start without signing up: open the sample store dashboard. At the top of the recommendations column there’s a card saying something like “the checkout pass-through rate flagged on Aug 6 has gone from 66% to 81%.” That card is everything this post was trying to say.