/ Personalization

Recommendation engines: a must-have for website and mobile app UX

What do Spotify, App Store and Medium home pages have in common?


That’s right. Recommendations.

Recommendation engines are becoming a must-have user experience component for consumer apps. This is especially true for consumer apps where users interact with products. For instance, in Spotify, users listen to songs. In the App Store, users download apps. In Medium, users read articles.

In fact, recommendations are frequently combined with search, which is a UX pattern I'll refer to as Recommend + Search. You can see this UX pattern in all of the examples above, where recommendations take up majority of the screen real estate and there is a search button either in the bottom tab bar or header bar.

In this article, we’ll discuss about why this is the UX of choice for consumer apps and show you the different shapes and forms they take.

Why Recommend + Search?

To understand why Recommend + Search is the UX of choice, we have to understand the function of the home page.

The home page is the doorway to the contents of an app. The goal of the home page is to help users find what they want in the shortest amount of time.

Before recommendations became so mainstream, there was Search + Browse. The idea was to cater to two types of users:

  • Search: Caters to users who know what they are looking for
  • Browse: Caters to users who do not know what they are looking for

CraigsList is a great example of the Browse + Search UX pattern:

The problem with Browse + Search is that it passive and user-led. In CraigsList, users have to identify their broad category of interest, then drill down into the subcategory and then browse individual listings.

As a side effect, it also makes home pages static. After couple of hours of using a site, a user might run out of ideas as to where else to look.

Enter Recommend + Search.

With Recommend, apps are able to actively suggest things that a user is likely to enjoy based on his past actions on the site. When apps run out of things to actively suggest, they can fall back to the Browse behavior where they show categories to users. Hence, Recommend is a combination of browsing-based and personalized content.

Product recommendations come in many shapes and forms

Product recommendations may have been popularized by the “Things You May Like” block, but have moved on quickly to take on other intelligent and creative forms.

In many cases, apps tell their users why they are recommending these items. Doing so combines recommending with browsing, allowing users to decide whether they are interested to take a further peak into each recommendation block.

Let’s take a look at 11 different ways apps make use of product recommendations:

1. Plain vanilla user product recommendations

2. Plain vanilla browse

3. User product recommendations by category

4. New product recommendations

5. Trending product recommendations

6. On-sale product recommendations

7. Editors' product recommendations

8. Recently viewed product recommendations

9. Product recommendations because you did X

10. Time-of-day / weather-based recommendations

11. Location-based recommendations

What’s next?

As personalization and recommendation engines become increasingly popular, the Recommend + Search UX pattern will continue to be refined and optimized.

Over to you now. What might we see next as an evolution of this pattern? Interested to hear your thoughts below!

Justin Yek

Justin Yek

Partner & Cofounder @ Altitude Labs. Responsible for our in-house personalization technology Metisa.

Read More