Sephora's #1 customer service (CS) query was online payments. Majority held negative sentiments about payment failures, refunds, and order cancellations. In fact, 75% of users give up after encountering payment failure.
Thankfully, we learnt that most users solve the problem through refunds before making a new order, and we didn't have that solution in our product. Well, the solution was straightforward wasn't it? Create a new user flow to automate refunds, and users can confidently make a new order.
However, in our discussion with tech, it was impossible to automate refunds. Every refund request would ultimately get routed to CS for a manual refund. It defeated the purpose and might even result in a larger load of CS tickets.
"Hmm, what we can do instead is to override the user's faulty payment if they make a second one", tech suggested. This meant that we would need to give users the opportunity AND encourage them to make that second payment, to override the potentially faulty payment.
We had this special payment state called "Payment Pending" (see below). This was where the system decided whether a payment was successful or not, and the decision time ranged from 15min to 24hrs. During this period, users could only wait with no other way out. We decided to take advantage of this limbo state, and allow users to retry payment. In fact, our PMs did up a numbers, and we found that the increase in retrying payment could have big value impact. So that was one opportunity.
However, we couldn't write away the possibility that users would want to cancel their order instead, since it came up in several CS tickets. So we made space for that option too.
Image: Payment Pending screen
By improving the retry rate (for otherwise cancelled orders) of all payment methods and markets, the potential for incremental revenue of €251.9K for a 1% increase; €504.7K for a 2% increase; and €2.527M for a 10% increase.
Users want alternative ways to troubleshoot payment, rather than just waiting at the payment pending page, either through retrying payment, or cancelling their order.
I breakdown the three main changes we made below: 1) Adding new CTAs and they're corresponding flows 2) Using visual hierarchy 3) Improving the copy.
Retry payment was higher in the hierarchy; we wanted users to opt for that rather than order cancellation. There were two main reasons:
This is a breakdown of the areas for improvement:
This teases out the rationale behind the new copy:
In order to answer the hypothesis, we set up the design survey with these questions in mind:
This group of users saw it as a way to take immediate control of their own payment outcomes.
These users were either afraid of double charges or misunderstood their payment status.
The users either wanted a sure outcome or misunderstood the copy that payment failed.
Majority users prefer to retry payment rather than to wait or cancel their order. They believed it'd lead to a higher chance of success.
However, we also realise that some users who chose to wait or cancel their orders, did so because they misunderstood the copy!
Users understanding of "payment pending" varied.
Indeed, the payment pending copy was confusing, and contributed to misinformed decisions (waiting or cancelling orders). Although it did not stop users from retrying payments, the unclear copy still poses a risk. It might bring users down the wrong path, which is detrimental especially at the end of the purchase funnel. It leads to a loss in credibility, trust, and our relationship with our users.
It was clear we should retain the "Retry Payment" CTA and remove "Cancel Order". The user flow for "Retry Payment" worked for users.
Improving the copy was trickier. It seemed that the most problematic phrase was "payment pending". Thankfully, there were insights that came out of the research to help us improve on the copy holistically:
The copy just needed to do the following:
Left: Initial proposed design
Right: New proposed design
To validate the new copy, we would recommend conducting another round of simple testing. It could be a quick and easy 5-second test. We could show a user the screen for 5 seconds, ask for their impression and what they understood from it. Alternatively, we could conduct a more thorough usability test, and ask questions that will help us find out if the user understands the troubleshooting option.
Because of the high value potential (revenue of €251.9K for a 1% in retry rate), the team has taken this initiative on and is currently in development.