Mobile drop-off charts often show longer dwell between address entry and delivery selection. Teams rush to redesign. Sometimes the pause is a shopper comparing shipping prices on another tab. Sometimes it is a sticky postcode lookup.
Pair time with exit reason proxies
Where support tags or on-page survey scraps exist, we align them loosely with the timing band — never as proof, always as context. A cluster of “delivery too expensive” notes beside a long dwell argues for price communication; a cluster of “page froze” notes argues for technical friction.
Show distributions, not only averages
Average dwell hides the shoppers who sail through in twelve seconds and the ones who stall for three minutes. Our visualisation packs include a simple distribution tile when the export supports it. That tile has ended more than one argument about “everyone struggling.”
Know when to stop
If timing data is sparse, we say so and lean on step-to-step completion instead. Inventing a behavioural story from thin timestamps helps no one — least of all the engineer who would inherit a phantom ticket.