Matt zamudio

Crowdsource by google
Making Google better for everyone.

Onboarding

Crowdsource (available on Android or desktop) isn't the easiest thing for people to wrap their heads around.

In essence, Google is asking users to complete "tasks" (which I promptly changed to "activities") . Their feedback trains AI and machine learning models to help make Google products better.

My job was to come up with a content strategy that positioned this request in a better light, and made users actively want to contribute. Then, I would use that content strategy to roll out a number of new features, activities, and games.

Starting with the strategy, I developed the idea of a Crowdsource community of "contributors". I also focused on the truly incredible impact contributors could make by using the app: they could make Google products, which are traditionally very focused on western culture, more inclusive of languages, religions, and cultures around the world.

Onboarding (seen below) was a huge first step. Feel free to watch the video mentioned in the text.

Crowdsource Companion

To further help users navigate Crowdsource, I worked with a designer to develop "Chibi," an alien robot and a contributor companion.

Beyond simply aiding users with helpful advice, Chibi's job was to personify AI/ML to make users more comfortable with the concept.

Notifications

I developed notifications to re-engage users and encourage them to reach the next level or achievement. These proved highly effective in bringing users back.

It helped, too, that with a better content strategy, users actually wanted to contribute, rather than feeling like they were completing a laborious task.

Image capture activity

Before I joined the team, the user flows at Crowdsource were very clunky and confusing. The user was slapped in the face with consent agreements and camera permissions before they had any idea what they would be doing.

I changed that by working with a designer to create smoother, clearer, and easier-to-use activities, like Image Capture.

In Image Capture, users simply snapped photos of objects around them, added a lable, and submitted them to the Open Images database, where a machine learning model would learn about them.

My job was to make all of this crystal clear. I wanted the user to know what they would be doing, as well as the impact their actions would have, before encountering a consent agreement. The groundwork (i.e. comprehension of the activity including how it works and its value) had to be laid before we made requests of the user.

Smart Camera

Smart Camera is another very cool activity I conceptualized, designed, and launched with the Crowdsource design team.

The user points their Android camera at an object, the camera automatically finds the position of the object, snaps a photo, and (attempts) to label it.

The user then tells the machine learning model whether Smart Camera's guess was correct or incorrect (a better title might have been Kinda Smart Camera).

This activity required the user to download a machine learning model to their device. This would be a challenge, to be sure.

The design I developed made this process simple, easy, and fun. The value was clear, how the activity worked was clear, and, given this context, the user was happy to download the model (which I specified would only occur over wi-fi unless their settings said otherwise).

Happily, we saw great adoption of this very fun activity when we launched, no doubt due to the very informative onboarding and in-activity experience.

Leaderboard

Everything a contributor does in Crowdsource is gamified. They earn badges, rise through levels, and track the amount of contributions they've made.

Top contributors were highlighted in our newsletters and social media pages, and we even issued certificates to the best of the best.

Still, I wanted to take it to the next level - to make contributing even more satisfying.

I worked with designers to create the Crowdsource Leaderboard, a global multiplayer experience where users can compete for the top spot in their country.

This not only added a competitive aspect, which increased contributions and usership. But it also emphasized the aspect of community - that the user was not alone, but, instead, part of a global community of contributors having fun while making Google products better for everyone.

External entry points

It made sense for us to have user flows from other apps to Crowdsource. This allowed us to increase awareness and adoption in a native way.

I worked to make these transitions as smooth and intuitive as possible to incite action and engagement.

The Results

The results of my work for Crowdsource were fantastic. After I joined the team, developed a content strategy, and put it in the product, the amount of users consistently using and contributing to Crowdsource skyrocketed. 

Now that users understood the product, why they should use it, and how they should use it, our KPIs steadily then quickly increased over the course of the next couple of months.

We saw increases in our HEART framework (Happiness, Ease-of-Use, Adoption, Retention, Task success), which I measured by the goals we set for each, their respective signals, and, finally, metrics.

And perhaps the coolest of all, we saw a major boost in the size and passion of the community of contributors. This sparked the idea for Crowdsource to host contributor meetups around the world, which continue to be organized on a monthly basis.

In the news

- Google Crowdsource adds leaderboards and 'Smart Camera' (9to5Google)

- Crowdsource app (Wikipedia)

Google Crowdsource app gets gamified leaderboards and new 'Smart Camera' feature (AndroidPolice)