OĞUZ EROLADS & AI

E-Commerce Conversion Tracking | Measure Sales by Value

2 min read18 August 2026

Measuring conversions in e-commerce is not just counting sales. The value of each sale has to be correct too, because the bidding strategy and the ROAS calculation rest on it. If the value is wrong, all optimisation pulls in the wrong direction.

Is the purchase event correct

The purchase event sits at the centre of measurement and must carry order ID, value, currency and product information. When one of these is missing, measurement appears to work while producing incomplete data.

The most common error I see is sending value including tax and shipping. That makes ROAS look better than it is and breaks the profit calculation. Which value you send is a choice, but it has to be consistent and you have to know what it includes.

Double counting in e-commerce

In setups tied to an order confirmation page, a user refreshing produces a second sale. Similarly, a user returning from a payment provider gets recorded twice in some builds.

Sending the order ID largely solves this: a second record arriving with the same ID is not counted again. In setups that do not send an order ID, double counting is nearly inevitable.

The payment provider trap

If the user goes to another site to pay and comes back, GA4 can treat that as a new session and record the sale with its source lost. The sale that advertising produced then shows up as direct traffic.

The fix is adding the payment provider’s domain to the referral exclusion list. That single setting corrects source attribution across many e-commerce accounts. I always check it during GA4 setup.

Returns and cancellations

This is the most skipped part of measurement. If returned orders stay recorded as conversions, high-return categories look more profitable than they are and the algorithm drifts toward them.

Feeding returns back brings measurement closer to reality. If automatic feedback cannot be built, applying the return rate to conversion value as an approximate correction also helps.

How do we start? First I test your purchase event: is the value right, is anything double counted, is the source being lost. I send the findings in priority order in writing. You can reach me from the contact page.

Frequently Asked Questions

Should conversion value reflect margin?

Ideally yes. Making the higher-margin of two equal-revenue products look more valuable to the algorithm pushes budget toward the profitable side.

Why does the GA4 sale count differ from my admin panel?

Some difference is normal — ad blockers, consent refusals and attribution differences all contribute. If the gap keeps widening there is a structural problem; comparing by order ID reveals the cause quickly.

Is the Shopify or WooCommerce plugin enough?

In most cases it sets up basic measurement, but default settings can produce double counting and incorrect values. Do not treat a setup as sufficient until it has been tested.