Are you a sports fan?

A scientific approach to proving that words matter

The Problem

If it’s not broken, it can still be fixed. As a UX writer for Google, I worked with a team of language specialists and engineers to run text experiments aimed at creating and identifying strings that could increase user engagement and overall copy quality. I wrote the copy variants to be tested in comparison to the “control” string, which was the one in production for every Google user to see.

For this experiment, the subject was a promo asking the user if they want to add an icon to their phone’s home screen. When a user searches a sports team on their phone, a promo may appear asking if they want to add the icon as a shortcut to the search. I wrote two sets of strings for two separate tests – one for the Google App and one for Google on the mobile web – measuring user interactions.

Constraints

Adhere to UX best practices
Align with brand voice and tone
Limited space
5–8 variants needed

Area

Google Search

Users

General public

String Variations

For the Google App

Control: Add a shortcut to your home screen for “<sports entity>”

A: Want to get <sports entity> info quickly? Add a shortcut to your home screen.

B: Are you a <sports entity> fan? Add a shortcut to your home screen to get here quickly.

C: See <sports entity> info faster by adding a shortcut to your home screen

D: Add a <sports entity> shortcut to your home screen to get here quickly

E: Jump to <sports entity> info faster with a shortcut on your home screen

F: Search “<sports entity>” quickly with a shortcut on your home screen

G: Add a shortcut to your home screen to jump to <sports entity> info with one tap

For mobile web

Control: Add a Google app shortcut for “<sports entity>” to your Home screen

A: Want to get <sports entity> info quickly? Add a Google app shortcut to your home screen.

B: Are you a <sports entity> fan? Add a Google app shortcut to your home screen to get here quickly.

C: See <sports entity> info faster by adding a Google app shortcut to your home screen

D: Add a <sports entity> Google app shortcut to your home screen to get here quickly

E: Jump to <sports entity> info faster with a Google app shortcut on your home screen

F: Search “<sports entity>” faster with a Google app shortcut on your home screen

G: Add a Google app shortcut to your home screen to jump to <sports entity> info with one tap

Decision-making

I hypothesized that a more compelling message that stressed convenience and ease would lead to more “add” clicks and subsequently more entry points into Search. I focused on value-forward calls-to-action, telling the user what they’ll get before telling them what to do. I decided the core value proposition was speed – the ability to land on these regularly updated results pages with one tap instead of needing to type in their search again and again. For something like scores from a favorite sports team, I decided the user would be most open to a conversational tone that feels like it connects with them personally.

Metrics

For the experiment for the Google App, all but one of the seven string variations performed better than the control string. Four led to increases in “good interactions,” which are clicks with a duration of at least 20–30 seconds in which the user doesn’t perform any interaction outside of that content. The best performer – “Are you a Lakers fan? Add a shortcut to your home screen to get here quickly” – experienced about a 34% increase in both interactions and good interactions.

Three strings for the Mobile Web experiment saw increases in good interactions. “Are you a Lakers fan? Add a Google App shortcut to your home screen to get here quickly” (an adaptation of the app’s top string) remained in the lead, with a 26% increase in interactions and a 44% increase in good interactions. An even more telling measure of performance is when the ratio of good interactions to interactions is positive — in other words, good interactions increased by a larger amount than the interactions overall. These Mobile Web variants performed remarkably well by this success metric. The strings were also translated – using detailed message descriptions that I wrote for them – into about 50 different languages and they saw increased engagement in dozens of those languages.