April 15, 2026
Customers Don’t Shop the Way Analytics Says They Do
Think about the last considered purchase you made online. Chances are you did not arrive at a product page, read the description once, and buy. More likely, you searched, clicked around, closed the tab, came back a day later, added something to your cart, left it there, re-searched, read more, and eventually decided, on your own timeline, following your own logic.
That exact journey is the new standard, according to new research from Conviva. The 2026 State of Digital Experience Report finds that 67% of digital customers follow non-linear purchase paths, looping between search and product pages, revisiting carts across multiple sessions, and spreading decisions across days. The analytics tools most brands rely on cannot see any of it.
The Customer Your Analytics Labels a Problem Is Often Your Best Buyer
The cart revisits, the search loops, and the long consideration windows were all signals that a customer is about to buy. Customers who visit their cart five or more times, the group that conventional analytics classifies as its highest-abandonment problem, convert at 40%, nearly double the rate of first-time cart visitors and 54 times the site average. This is the logical outcome of understanding the cart as a research and comparison workspace that customers return to deliberately. The intervention designed to “recover” these customers, typically a time-pressured discount or a re-engagement prompt, risks disrupting the buying process of the people most likely to complete it.
The same logic applies to search behaviour. Nearly half of all converting e-commerce sessions contain at least one return-to-search loop — a backward navigation that standard analytics records as a drop-off. Conversion rates scale directly with the number of loops a customer runs: one loop produces a 13% conversion rate, two loops 16%, and three or more loops a 19% rate, representing 25 times the site average. In other words, the customers iterating most actively are the ones most likely to buy, and every system designed to read iteration as disengagement is misidentifying them.
Customers who engage extensively with specifications and detailed content convert at 3.5 times the session baseline. These users register as low engagement in bounce-rate terms. In revenue terms, they are among the most engaged buyers in the data.
Brands Are Watching the Wrong Customers
Non-linear purchase behaviour holds across two datasets with fundamentally different products, price points, and purchase cycles — 65% in e-commerce and 70% in travel booking. The report attributes the travel figure’s higher rate to the added complexity of that decision: flight dates, passenger counts, payment friction, and multi-day consideration windows.
Customers today move through digital spaces haphazardly: switching between tabs, apps, devices, often distracted by different stimuli, which causes them to return to decisions days later. The tools organisations use to track this behaviour no longer match it.
Those tools are now the foundation on which AI agents are being built, and trained to personalise, intervene, and optimise. Unfortunately, they are trained on data that cannot see the majority of customer behaviour, the loops, the returns, the cross-session research that precedes most purchase decisions.
Brands are already losing visibility into the purchase journey as customers increasingly use AI tools to research and compare products before they reach a brand’s own digital property. Even the behaviour that does happen on a brand’s platform is largely invisible to the analytics tools interpreting it.
