Adding a product to a cart signals interest. It does not mean the shopper has decided to pay now. They might be checking shipping costs, comparing products, or saving an item for later.
When I look at cart abandonment, I first check whether there is evidence to distinguish people who hesitated to pay from people who tried to pay but could not. The first situation may involve price, purchase conditions, or timing; the second may involve input, authentication, or payment errors. A single abandonment rate cannot tell the two apart.
Add-to-cart events and people who leave are different counts
In GA4, add_to_cart is an event for measuring the addition of an item to a cart, begin_checkout measures the start of checkout, and purchase measures a completed purchase. Collecting add-to-cart events does not mean the other actions are automatically measured correctly. First check whether the site sends events that reflect the actual customer journey. GA4 ecommerce measurement guide
If one person adds products several times, the add-to-cart count increases. That is why dividing purchases in the same period by that count cannot, on its own, tell you what percentage of people who added an item left without buying. If the numerator is orders and the denominator is button clicks, you are comparing different units.
Consider a small illustrative example. If one person adds items three times and places one order, there may be three add-to-cart events and one order. You should not interpret this as two out of three people abandoning their carts. If the site also offers a Buy Now or express checkout route that bypasses the cart, valid orders can occur outside the cart funnel.
Start with a narrower question, such as how many users who added something to their cart went on to start checkout and complete a purchase within a defined period. Record whether the analysis is user-based or session-based, and how many days you allow for the purchase.
First, use your phone to go as far as the final purchase step
Some problems become obvious on the actual mobile screen before you examine an abandonment chart in more detail. An administrator's account may already be signed in and have an address saved, so review the screens a new customer sees as well. Keep the test within steps that do not create a real payment, or use a prepared test environment.
I would work through the following tasks in order, rather than simply judge how attractive the screens look.
- Change product options and quantities in the cart, and check how the final amount due changes.
- Check whether you can continue as a guest. If registration is required, note when you find that out.
- Enter an invalid address or contact detail and check whether you can tell which field needs correcting.
- Select a payment method, then return to the previous step and check whether the cart and entered information are retained.
- Check whether the mobile keyboard, a bottom banner, or a chat window covers an error message or the next-step button.
Record observations precisely: instead of “checkout is difficult,” write “the address error message appears above the current input position and is not visible on screen.” Note the device, browser, and sequence of actions so someone else can reproduce it. Failing to find a problem in one test does not mean every customer's environment is free of issues.
Check when shipping costs and registration requirements become clear
If shoppers think they understand the price and then encounter shipping charges or other costs just before payment, they have to reassess the offer. For someone who needs an item quickly, the arrival date can matter as much as the total price. Also check whether an offer that looks appealing on the product page changes for the selected option or delivery area.
Baymard's research with US online shoppers identifies extra costs, slow delivery, account creation requirements, and complicated checkout processes among reasons for abandonment. Some people, however, were simply browsing and were not ready to buy. Use this research to identify possible causes. Its figures do not establish the share of each cause at your store or the effect an improvement would have. Baymard research on cart and checkout abandonment
Checking whether shipping charges are a problem does not mean you should immediately offer free shipping. Start by asking whether shoppers can see the relevant conditions before deciding to buy. Making a free-shipping threshold, surcharges for particular areas, and an estimated delivery date understandable is a separate question from changing the pricing policy itself.
For guest checkout, checking that the option exists is not enough. If it is hidden in small text beneath the login form, customers may assume they have to register. Baymard's usability research also examines cases in which shoppers fail to recognize the guest checkout option. Check for yourself whether a new customer can easily find it. Baymard research on making guest checkout visible
Use payment records to investigate failures after a payment attempt
Once a customer reaches checkout, you can examine more specific evidence of what happened. Separate whether they clicked the payment button, moved to card authentication, received a declined authorization, or had an approved payment correctly reflected in the order status. The absence of a purchase-completion event alone does not show where the process broke down.
Check the time of failure, payment method, and error message in your store administration system or payment provider's records. Shopify, for example, explains how to review payment-attempt events and messages in an abandoned checkout's timeline. Card declines, authentication failures, and stock or discount-code problems may each need a different person to resolve them. Shopify guidance on abandoned checkouts and payment events
Do not assume that GA4 and your store administration system will report the same number of abandonments. Shopify's abandoned checkouts follow their own collection criteria, such as a checkout remaining incomplete for a period after an email address has been provided. The list does not represent everyone who added something to a cart and left.
If the payment provider has an authorization record but the store's order status or GA4 purchase record is missing, investigate communication between systems before attributing the issue to weak purchase intent. Check the return URL from the external payment window, payment-result notifications, and the timing of the purchase event separately. Do not paste payment details or personally identifying information directly into analysis reports or screenshots; document the error type and the conditions under which it occurred.
Use the funnel to narrow down where the drop happens
Once you have confirmed that events are being collected correctly, build a funnel from add to cart → begin checkout → complete purchase. GA4 funnel exploration distinguishes a closed funnel, which includes only users who enter at the first step, from an open funnel, which also allows entry at later steps. Two funnels with the same name can count different numbers of people if their settings differ. GA4 funnel exploration guide
For the initial analysis, also decide whether to allow other actions, such as page views, between steps and how much time to allow before the next step. Different products involve different consideration periods. Requiring every purchase to finish within one short visit can miss valid purchases made on a return visit.
Next, break the results down by device and traffic source, but avoid making the segments too small at the outset. If progression after checkout begins is especially low on mobile, inspect checkout screens and error records in that environment first. If the larger drop is between adding to cart and starting checkout, examine what appears between those steps, such as the total cost, shipping conditions, and registration screen. This interpretation tells you where to investigate; it does not establish the cause from the funnel alone.
A device breakdown in a GA4 user funnel does not always mean the device actually used at each step. Users may be classified by their first applicable device. To attribute a problem to a particular mobile browser, you also need evidence such as a reproduction in that environment or payment logs.
Define one hypothesis and the result you will check for each change
A problem is easier to act on when you document it as “observation → possible reason → proposed change → result to check.” If you observe that shipping conditions are hard to find, for example, consider showing them earlier rather than waiting until just before payment. Look beyond button clicks when checking the result: examine checkout starts and completions, profit per order, and changes in shipping-related inquiries.
When an error is actually preventing payment and its cause is confirmed, fix it as a priority. Changes that remain hypotheses, such as the effect of a layout or wording, should be examined through a comparative experiment where possible. If an experiment is impractical, record the change date alongside other changes in advertising traffic, discounts, or stock availability. An increase in orders after an edit does not prove that the edit caused the entire increase.
The purpose of improving checkout is not to drive down abandonment at any cost. It is to help people who want to buy understand the conditions, enter the necessary information, and complete payment. Start by connecting one obstacle you can observe on today's mobile screens with one source of evidence showing how often it occurs.