KFC Mobile App

At KFC, at one point I was responsible for three teams: the mobile app, the website, and the restaurant and menu management system. I set the tasks, looked for where the problems were and made the decisions. There was also the kiosk team. It didn't report to me, but I helped it: I was in restaurants a lot, and the kiosks had the same problems as the app and the website. When we shipped something new, it sometimes made sense to support it everywhere.

This is one story from that work: how we rebuilt the app's checkout and how conversion from app login to successful payment then grew from about 10% to 19%.

How we saw where the problem was

The problem showed from several sides at once. The funnel showed how many people dropped off at checkout. The app reviews had complaints. I kept ordering through the app myself, went to restaurants and talked to staff and guests.

We watched conversion from app login to successful payment. We counted up to a successful payment, because a payment can still fail after you tap "Pay". From January to June 2021 that conversion stayed around 10%.

What was wrong

A customer doesn't want to spend time on checkout: choosing how to pay, the pickup or delivery time. In the old version the payment method and the time were chosen in separate sheets. Each opened over the cart and closed with its own "Done" button.

Yet the app already knew almost everything it asked: the payment sheet held the customer's saved cards and Apple Pay. The promo code link sat right above the pay button. And the "Try this" block could suggest fries for an order that already had fries.

What we changed

We showed the discounted price on its own line and repeated it on the button. The customer sees how much they'll pay right where they tap.

We moved the promo code to the top so it wouldn't distract from paying.

Recommendations started to depend on what's already in the cart. The dishes for them are set up in the restaurant management panel, which I was also responsible for.

We showed which restaurant the order can be picked up from and how far away the customer is, so that people would less often order at the wrong restaurant. In the funnel such an order looks like a successful payment, so the number doesn't show this problem.

The time is picked right on the screen, without a separate sheet, and the phone vibrates as you scroll.

The payment method is visible at once, so we don't have to ask for it separately. If Apple Pay or a saved card works for them, the customer just taps "Pay".

In the end everything a customer checks before paying, the amount, the restaurant, the time and the payment method, ended up on one screen.

How we made decisions

We discussed the changes within the team and with other product managers. As far as I remember, nothing had to be rolled back. I think that's because the team already knew how to do things well, and every decision went through a discussion.

What came of it

We started rolling out the changes in June 2021. From July conversion grew every month and by October reached about 19%.

There were no big changes outside the product in those months, so I attribute the growth to these changes. A single number can't show how much each of the six changes contributed.

What I took from it

• Look for the problem in several places at once: the funnel, the reviews, your own orders and conversations with people on site.
• Don't ask the customer what the app already knows.
• Put everything a person checks before paying on one screen, and move whatever leads away from the pay button.
• Count conversion up to a successful payment, not up to the tap on the button.