At the intersection of social sciences and design

Preview Web

Prismatic Preview Web

The Prismatic launched on web in mid-2012, with the iPhone app joining a couple months later in September. These products were a proof of concept meant to feature the powerful machine learning technology that the team had worked on for over a year. We learned a lot from people's use and also knew were we wanted the product to go. Prismatic is a product that ultimately wants to be an interest network, and the product needed to start reflecting that.

For example, one of the goals is to connect with others. This wasn't coming through. We wanted it to feel more fresh and fun to use.  If we were going to connect people with their interests, needed to better understand that. The new web app, which came to be known as preview, was going to be an experimental testing grounds for a lot of this change.

During this time we worked with Teehan + Lax. They have a great post that gives a summary of the process.

Following and sharing your Interests

We embarked on research to question everything: how did people currently connect with who they are interested in? 

Follow Model

Prismatic had a simple follow model. But we wanted to take a step back and ask ourselves: is this the best way for Prismatic to be structured? Is there another way? This was a big study for us. We came up with  different models for how Prismatic could work and created paper prototypes to test with people. This was everything from a Pinterest type model where you can follow a person or a person's activity in their interests, or even just thumbing up things and never following.

Onboarding Usability + Relevance Perception

To test some of the new features of this new web product, we created a simple onboarding. We experimented with aspects such as search in onboarding, but found it was paralyzing. Most people preferred selecting who they were from a set of options rather than explicitly having to come up with points of interest. Did a number of tests on onboarding for usability as well.

The more interesting aspect of this is perception of relevance. We have a lot of data about how our relevance performs, but we did more exploratory research on how people decided whether a story was relevant to them. Were stories with photos likely to be more relevant? Titles were an interesting aspect of it. And the new explain system we had just created.

Story actions
Actions were confusing. We knew this from data but also user interviews. Many had interesting guesses as to what they were doing. Generate actions and test them on people to get a sense of direction.

Social Prismatic

I focused a lot on the social aspects of the product. More social aspect, people were seeing things because of social. wanted social communities. better understand how they could influence change. in addition to actions, 

  • added explains
  • added discussion. threads/autocomplete. 
  • profile
  • activity
Activity grid exploration showing some of the different activity types

Activity grid exploration showing some of the different activity types

Activity deserves a deep dive. How can we build Prismatic to be a community - and further yet, what is a community? We came to use a working definition. Emotional sense of community. And it's different components -> that informed a lot of the views of the new app. Why even make Prismatic a community? Good question. Behind the scenes, Prismatic works like a community. Sense of community -> activity types, want to see your effect on community.