Your abandonment rate describes your tracking before it describes your shoppers
A checkout abandonment figure looks like a fact about shoppers. Before it is that, it is the outcome of a handful of settings: which step counts as the start, which visitors are measured at all, how long a shopper is given to finish, and whether the moment of payment is recorded when it happens. Change one of them and the same week of orders produces a different rate. Google’s help pages for Analytics set out these settings plainly, and the questions they raise apply to any analytics tool a webshop uses. Read the rate with those choices in view, or the team ends up repairing a checkout step that is in fact a gap in the measurement.
The funnel definition sets the size of the loss
Google’s documentation on funnel reports distinguishes two types. In a closed funnel, a shopper must enter at the first step to be counted. In an open funnel, a shopper can enter at any step. The difference matters most around the payment step. A shopper who returns through a reminder email straight to the payment page, without passing the first step in the period you report on, is counted in an open funnel and missing from a closed one. Neither view is wrong, but they answer different questions, and a rate quoted without saying which one was used cannot be compared with last quarter’s.
The same documentation says the report counts users who complete the steps in the order defined, and that a step can be given a time limit within which it must follow the previous one. A shopper who skips a step drops out of every step after it. The time limit and the date range then decide how patient the report is. Set a short window and a shopper who leaves on Monday and pays on Thursday is booked as an abandonment. Set a week without a time limit and the same shopper is a completed checkout. A shopper who compares, asks a partner or waits for payday fits that pattern, so the setting changes the story about them.
Breakdowns have a rule of their own. A user is attributed to the first value of the breakdown that applies to them, so a shopper who starts on a phone and finishes on a laptop counts under mobile only. A weak mobile figure can therefore include orders that were completed, just on another device.
The payment step is where measurement breaks first
Bank payments move the shopper away from your site at the decisive moment. With iDEAL, the shopper is sent to their own bank’s environment, confirms the payment there, for example in the banking app, and the merchant is notified as soon as the payment is approved. From that moment the order is paid, whether or not the shopper’s browser ever returns to your confirmation page. If your analytics records the purchase only when that page loads, every shopper who pays and then closes the banking app shows up as someone who left on the payment step.
The trip to another domain causes a second distortion. Google’s help page on unwanted referrals names third-party payment processors as a typical case for the setting. Without that setting, a shopper coming back from a processor’s domain can be reported as arriving from that domain, so the processor turns into a traffic source and the campaign that brought the shopper can lose the credit for the order. The page’s own example shows that adding the domain later does not change sessions already credited to it.
The combined effect is a funnel that loses shoppers at exactly the step where your team is most tempted to act: the list of payment methods. Before you reorder methods or add one, check that the loss is real.
Some shoppers are not in the funnel at all
Where visitors are asked for consent before analytics runs, part of your traffic is measured differently or not at all. Google’s documentation on consent mode describes two set-ups. In the basic set-up, the tags do not fire for visitors who decline, and no data about them is sent. In the advanced set-up, the tags still send measurements without cookies, and Google can use them for modelling once the site passes a data threshold. In a funnel report, though, Google’s documentation names modelling only for the first visit and the start of a session, and says nothing of modelling for later steps. Do not assume the checkout steps and the purchase include the shoppers who declined. Whether those who declined behave like the rest is something the report cannot tell you.
None of this makes the data useless. It makes the data a description of where a measured shopper stopped. It does not say why they stopped. For that, you need a different source: your own test purchases, session recordings where shoppers consented, and the questions that reach customer service.
Count your orders, then read the funnel
Start with a reconciliation. Take last month and set the number of purchases your analytics recorded next to the number of paid orders in your shop system, per day and per payment method. A gap that appears for bank payments and not for cards points to the return trip from the bank, not to shoppers who changed their minds. Then open the list of unwanted referrals and make sure every domain your payment step sends shoppers to is on it. Write down whether your abandonment report uses an open or a closed funnel, which step counts as the start, and which date range and time limit apply, and keep those settings fixed from one report to the next. Only after that, look at which step loses shoppers. The rate you then read is about your checkout.
Sources: support.google.com, support.google.com, developers.google.com, support.google.com, ideal.nl
