How a single conversion rate can hide the very problem a redesign was meant to fix
Imagine your team has just redesigned its checkout. A week later, conversion is up. Someone posts a screenshot in Slack. Someone else calls it a historic win (перемога історичного масштабу).
Then you slice the data by device type (розбиваєте дані за типом пристрою).
Desktop conversion has gone up. Mobile conversion has gone down. And mobile is where most of your customers shop.
Find the step where people get stuck
An average has a sneaky way of making two very different experiences look like one (середнє значення хитро згладжує різницю між двома зовсім різними досвідами). In this case, a jump in desktop purchases masks an underlying problem (приховує проблему, якої не видно за загальним показником) on mobile.
To pinpoint where users drop off (точно визначити, на якому кроці користувачі переривають оформлення замовлення), look at the checkout funnel one step at a time. Say the biggest drop happens right after shoppers enter their delivery address. Session recordings might surface a point of friction (виявити, що саме заважає користувачам) the funnel can’t show you: on smaller screens, the keyboard covers the Continue button. In a usability test, you might hear someone say, “I thought I’d already filled everything in. What am I supposed to do now?”
Get Olha Nonko | English Tips Lviv | ETL’s stories in your inbox
Join Medium for free to get updates from this writer.
Stumbling upon that moment in a recording (випадково натрапивши на цей момент у записі) gives you a hypothesis. You can triangulate your findings (звірити висновки за кількома джерелами), checking whether the same pattern shows up in the funnel, in other recordings and in usability tests.
See whether the uplift holds up under scrutiny (витримує ретельну перевірку)
Before you attribute the uplift to the redesign (поясните приріст саме редизайном), check what else changed that week. Did a free-shipping promotion go live? Did a campaign bring in a different mix of shoppers?
Historical data (дані за минулі періоди) can show whether conversion typically rises during promotions. A longitudinal study (лонгітюдне дослідження, у якому за тими самими учасниками спостерігають протягом певного часу) that follows the same shoppers over several weeks can help answer a different question: does the new checkout get easier with repeated use, or do people keep getting stuck at the same step?
Make the analysis comprehensible (зрозумілим)
A comprehensive analysis (всебічний аналіз) might cover device type, traffic source and every step in the funnel. A comprehensible summary (зрозумілий підсумок) tells the team where to look first.
For this checkout, that could be one sentence: “Overall conversion rose, but more mobile shoppers dropped off at the delivery step, where the keyboard can hide the Continue button.”
That’s specific enough for a designer to reproduce the issue and for an analyst to measure what happens after the fix.
