The only 5 numbers an ecommerce store needs to check in GA4 each week — where they live, and what to decide after you look

In the previous post (You installed analytics. Now what?) I covered why ecommerce analytics ends up meaning nothing, and a 15-minute weekly routine to fix that. This post is the detail. I’ll narrow the numbers you check in GA4 each week down to five, show you where each one lives, and say what to decide once you’ve looked.

It’s written for stores that send ecommerce events (purchase, add_to_cart) to GA4. The major carts — Shopify, WooCommerce, Colorme and the like — send them through their standard integration or official plugin. If yours doesn’t, numbers 3 through 5 in this post will be empty, so check your cart’s GA4 settings first.

The five numbers at a glance

Revenue is three numbers multiplied together.

Revenue = ① Sessions × ② Conversion rate (CVR) × ③ Average order value

And when ② moves, you can narrow down the cause by splitting “where are people leaking out” into two stages.

④ Product page → added to cart (product appeal, price, stock) ⑤ Added to cart → purchased (checkout friction, shipping, page speed)

Those five are the weekly check. Everything else in GA4 — engagement rate, event count, average engagement time — only needs a look when you’re hunting for the reason one of the five moved.

① Sessions — “how many people came”

Where it lives in GA4: Reports → Acquisition → Traffic acquisition. Look at the “Sessions” column, with “Session default channel group” as the row.

How to split it against last week: by source (channel). When total sessions go up or down, check which of Organic Search, Direct, Email, Paid or Organic Social moved. Looking only at the total ends in “it dropped a bit, I guess.” Splitting by channel almost always pins it to a single reason: “that was the week we didn’t send the newsletter,” or “we paused the ads.”

What to decide after you look: for the channel that moved, answer “was that change intended?” Only an unintended drop — search traffic falling two weeks in a row, say — is a candidate for next week’s to-do.

⚠ One caveat. If you’re putting custom values like sns or x in utm_medium, GA4 drops those into “Unassigned.” If Unassigned is more than a few percent, the problem isn’t your analytics — it’s how the links you send out are tagged.

② Conversion rate (CVR) — “what percentage of visitors bought”

Where it lives in GA4: in the same Traffic acquisition report, there’s a Session key event rate column (formerly “Session conversion rate”). If you’ve set purchase as a key event, that column is your conversion rate. If you have several key events set up, pick purchase from the selector at the top right of the column.

How to split it against last week: by device. Reports → User → Tech → Tech details, set the row to “Device category,” and look at the same column.

The reason is simple: about half of all ecommerce CVR problems take the form “mobile alone is dramatically low.” A total CVR of 2.5% is often really 5% on desktop and 1.5% on mobile. If you only ever look at the total, you will never see that gap.

What to decide after you look: if mobile CVR is less than half of desktop, next week’s candidates narrow to two — mobile page speed, or how much typing mobile checkout demands. If the gap is small, the problem isn’t CVR; go look at ④ and ⑤.

③ Average order value — “how much per order”

Where it lives in GA4: GA4 has no “average order value” metric. You divide “Purchase revenue” by “Transactions” yourself. Both appear in the cards at the top of Reports → Monetization → Ecommerce purchases, or at session level in Explore. Alternatively, “Average purchase revenue” in Explore gives you a per-user figure (which is slightly different from per-order).

How to split it against last week: new vs returning. In Explore, use “New / returning” (newVsReturning) as the dimension and put revenue and transactions side by side.

What to decide after you look: if AOV fell and the share of new customers rose, that’s your acquisition working — new people are making a “trial” purchase. That isn’t bad news. The decision isn’t “raise AOV,” it’s “how do we get new customers to buy a second time.” If instead it’s returning customers whose AOV fell, bundles and your free-shipping threshold are the candidates to revisit.

⚠ Be careful about what “revenue” means. If your store sends the full amount paid, including tax and shipping, in the purchase event, GA4’s revenue will come out higher than the “sales” figure in your cart’s admin. Line up the definitions before you compare the two (this post goes into detail).

④ Product page → added to cart — “do the products look appealing”

Where it lives in GA4: Reports → Monetization → Ecommerce purchases. Rows are products (item name); the columns include “Items viewed,” “Items added to cart” and “Items purchased.” Look at “added to cart ÷ viewed” per product.

How to split it against last week: by product. Compared with the average across all products, find one product where this ratio is unusually low and one where it’s unusually high.

What to decide after you look: a product with plenty of views that isn’t getting added to cart points to one of three things — price, photos and copy, or an “out of stock” notice. Conversely, a product with a high add-to-cart rate but few views is one where more exposure alone would grow revenue. This number can decide the order on your homepage and which product goes in the newsletter.

⑤ Added to cart → purchased — “are we losing people at checkout”

Where it lives in GA4: Reports → Monetization → Purchase journey. It shows sessions and abandonment rate at each step: session start → view product → add to cart → begin checkout → purchase. Write down “purchase ÷ add to cart” every week.

How to split it against last week: by device (same as ②). The Purchase journey report can be split by device category at the top.

What to decide after you look: when this ratio drops, the cause is almost always “something got added mid-checkout.” A shipping cost appearing, account registration becoming mandatory, a change of payment methods, a new delivery-date field. Cross-check against what you changed on your own site the previous week and it usually comes down to one thing.

The five on one sheet

Each week, just fill in this table. Copying the numbers into a spreadsheet beats saving GA4 views — it’s the only way the comparison with last week is guaranteed to be there.

Keep “next week’s change” to one thing. With two or more, four weeks from now you won’t know which one worked.

Honestly, though

Filling in this table means walking through five GA4 reports every week, doing the division for ③ by hand, and building an Explore report for each split. It’s 20 minutes once you’re used to it — but the moment you skip a busy week, “last week” is gone and the comparison breaks.

NextRise Analytics puts these five numbers and their splits on one screen from the start. Sessions, CVR, average order value and transactions sit across the top with a comparison to the previous period, and the Acquisition, Pages, Revenue and Day × Hour tabs give the breakdown for each split. And the equivalent of the “next week’s one change” column at the right of the table comes out automatically every morning as a card. Each one says which split, and how far off it is — “Mobile shoppers are buying at a low rate (mobile 4.4% / desktop 14.0%)” — so you no longer have to fill in this post’s table yourself.

You don’t need to stop using GA4. Adding one line of tag starts measuring from today, and you can import the period before installation from GA4 once. Start by opening the sample store dashboard and finding where this post’s five numbers show up.