TL;DR

Most agricultural and farming supplies eCommerce businesses have GA4 installed. Almost none have it configured to record purchases accurately. Standard GA4 installation records sessions and pageviews. It does not record orders, revenue, or checkout completions. Those require specific configuration and, more importantly, testing. The result is ad spend allocated against data that does not reflect real orders, channels credited for revenue they did not generate, and Google's bidding algorithm pointing in the wrong direction. This article covers what is typically broken, three checks to run today, and what reliable GA4 data should actually tell you.

What GA4 tracks by default — and what it does not

Installing GA4 gives you session data, pageviews, and a handful of automatic behavioural events: scroll depth, outbound clicks, file downloads.

It does not give you purchase tracking. Not unless someone has specifically configured it.

GA4 uses an event-based model. Every meaningful commercial action on your site has to be sent as a named event with the correct parameters attached. A purchase event needs to carry order value, product data, and a transaction ID. A basket event needs to carry product and price. None of that happens automatically on most eCommerce platforms. It requires configuration, and it requires testing after that configuration has been applied.

Most agricultural eCommerce stores were connected to GA4 during initial setup. Sessions started recording and it appeared to work. Nobody ran a test transaction to verify that a purchase event was reaching GA4, or to check what data the event carried when it did.


The five most common GA4 tracking failures in agricultural eCommerce

Conversion events firing on page load rather than on order completion

The thank you page fires a purchase event the moment it loads, not when a transaction is confirmed. A buyer who refreshes the page generates two conversions. A bot that crawls the URL generates one with no purchase behind it. The conversion count inflates, the cost per conversion drops, and every channel looks more efficient than the order book says it should be.

Purchase events missing entirely

The eCommerce plugin was installed but the purchase event was never tested. GA4 records a session reaching the checkout and then records nothing. No transaction data, no revenue, no product information. The business cannot tell which products are selling through the site, which channel drove an order, or whether the checkout is completing or failing at a particular step.

Double counting from two data sources

The platform sends a purchase event to GA4. The payment gateway, whether that is PayPal, Stripe, or Sage Pay, also fires its own tracking on the post-payment redirect. Both events reach GA4 and both record as conversions. The conversion total is double the real order count. Every benchmark and budget allocation built on that number is wrong by a factor of two.

Attribution crediting the wrong channel

A buyer first found the store through a Google Shopping ad three weeks ago. They did not purchase that day. They returned through an organic search result a week later, browsed, and came back directly a few days after to place the order. Under GA4’s default attribution model, the direct session gets the credit. The Shopping campaign gets none. The business looks at its channel report, sees Shopping producing no attributed revenue, and reduces the budget. The channel was working. The attribution model could not see it.

Spam sessions contaminating all traffic data

Referral spam accounts for a material share of sessions in many smaller eCommerce stores. The traffic acquisition report shows domains in the top ten that have never sent a real visitor. Bounce rates are distorted. Conversion rates calculated against total sessions are understated. Engagement metrics are pulled down by traffic that carries no signal. Without a hostname filter, the data has no reliable baseline and every metric derived from it is off.


Three checks to run today before spending another pound on ads

These take less than 20 minutes. They will tell you whether the data your decisions rest on can be trusted.

Complete a test transaction and watch the GA4 Realtime report

Open GA4 and go to Reports, then Realtime. In a separate browser, navigate through your checkout and complete a purchase using a discount code that zeros the order value, or a low-value test product if your platform allows it. Watch whether a purchase event appears in the Realtime report with a revenue value that matches the order.

If nothing appears, purchase tracking is not configured. If an event fires but the revenue shows as zero, as a round number, or as obviously incorrect, the data layer is passing incomplete parameters. Either way, you now know the answer before allocating next month’s ad budget.

Compare your GA4 conversion count to your actual order count

Pull the last 30 days of purchase conversions from GA4. Pull the order count from your store’s back end or order management system for the same period.

If GA4 is higher than your real order count, events are firing incorrectly. If GA4 is lower, transactions are being missed. Both are problems and neither number is reliable until they reconcile within a reasonable margin.

Look at your top traffic sources for spam domains

In GA4, go to Reports, then Acquisition, then Traffic acquisition. Look at the Session source breakdown for the past 90 days. If the top ten includes domains you do not recognise, particularly .xyz, .info, or high-volume referrers with zero engagement, referral spam is in the data.

A hostname filter in GA4’s data stream settings removes the majority of it. This is a fix that takes under ten minutes and has a material effect on the reliability of every metric the account produces.


What your GA4 data should actually tell you

When GA4 is configured correctly, it stops being a liability and starts doing the job it exists to do.

Channel performance by actual revenue becomes visible. For a parts dealer running Google Shopping alongside Performance Max and relying on organic traffic, this is the difference between knowing which channel drives orders and knowing only which channel drives sessions. Those are not the same thing, and treating them as if they are wastes money steadily and quietly.

For supplies businesses, purchase data tells you which buyers have ordered once and have not returned. Elogic Commerce’s 2026 benchmark puts B2B eCommerce conversion rates between 1.5% and 3.5% depending on sector and product type. Supplies businesses with habitual repeat buyers should sit at the upper end of that range. If reliable data shows they are not, there is a commercial problem worth diagnosing.

Checkout drop-off becomes visible at each step once event tracking is correctly configured. If most buyers who reach the payment page do not complete the purchase, that is a checkout problem. But without accurate event data you cannot see it, so the problem stays invisible and keeps costing revenue. The full picture of where agricultural eCommerce businesses lose sales, before and after the checkout, is in the first piece in this series.

Google’s automated bidding in Shopping and Performance Max uses conversion signals to allocate budget across products, audiences, and times of day. If those signals are wrong, the algorithm optimises toward phantom purchases. Fixing the tracking is not an analytics project. It is an ad spend project, and that is why checking tracking accuracy is the first thing done in every Arlieo engagement, before any channel work and before any campaign changes.

Common questions

Frequently asked questions