Logging in the Fast Lane

TLDR? This project was also covered in Noom's blog.

The problem:

When we asked our most dedicated Pro users what frustrated them about the product, they said “it takes too long to log my food”. We were already providing shortcuts to find a user’s frequently logged foods, and had polished the food logging UX experience, but it still wasn’t satisfying the user.

We found that 78% of users were using our “Frequent foods” feature, and that if a user logged a food 5 times, ⅔ of those times he would log with the same portion. We decided to tackle reducing the speed to log a familiar portion size.

Prototyping and first usability tests:

Our designer Youngin and I brainstormed different ways to address this problem. Here were some of her many wireframes:

Wireframing brought up some questions:

• What’s the best way to show that “one-click” was available?

• Is a swiping interaction discoverable enough? Do our users feel comfortable with swiping?

• How do we show feedback when the user logs an item? We tried marking the item in the list and showing each logged item at the top of the screen.

• Should we support deleting an item or logging the 2nd most popular portion size?

We decided to pursue a “fast lane” approach: expert loggers should become faster at logging, without modifying the experience of normal logging. We decided to de-scope the edge cases of deleting and logging the 2nd most popular portion size (which only accounted for 30% of logs on average, vs. 66% for the most popular portion). To answer the questions about discoverability, comfort level with swiping, and logging feedback, we made videos of the prototype and usability tested them with users.

We found that:

• No one noticed the grab bars we put on the left of food items to show that the user could pull them to the right.

• Once a user saw swiping demonstrated, she got excited — it was clearly a delightful action.

• We needed to link the swipe to an increment in the number of items logged. When this happened, users understood how and why to use one-swipe. Unfortunately, only 1 out of 4 users made this connection.

We added emphasis to the cart increment animation and saw good results in usability testing right away. Since we had validated the logging method and could separate the discoverability problem, our engineers started building the swiping functionality.

On to V2...

While our engineering team worked on building the solution, we worked on discoverability. We built an interactive tutorial for users on their 3rd meal with Noom that would show them how to log an apple:

The usability tests came back… ok. Users swiped to log, and could be prompted to swipe again after the tutorial. But the tone of the responses was worrying. “I didn’t want to log an apple!” they’d say, or, “Phew! we’re back,” after exiting the tutorial.

We had iterated for weeks and were out of ideas. So we shipped anyway and waited to get more data.

After the launch...

One Swipe and the Food Oracle came out at the same time, and our expert users responded immediately by boosting our food logging happiness scores. A success! But we still wondered about that tutorial. Was there a way to introduce experts to food logging without confusing people who wanted to log the way they always had?

After a couple of weeks, Youngin came up with an idea: skip the text and forced interaction, and use a “jumping” animation instead. It looked like this:

In usability tests with this animation, we saw much less confusion, and much more delight. Once we launched, we saw usage jump to 30% for long-term users.