Build a community

Onboarding process already have several steps. How’d we effectively recommend users who to follow within few seconds? We know educators read and validate carefully, how’d we include the information in the way that our users find our recommendations reliable and relevant to their teaching experience? How could we minimize the interaction efforts and provide a good start for educator to build their own community?

We start with two assumptions of how our users decide on whom to follow: 1. I want follow and learn more about the materials they uploaded, 2. I want to follow people who who teach similar topics/courses in my field. I started with two design options:

Through the user testing, we found out that educators tend to use “educator” as the anchor to make their follow decision. We also noticed that eduactors convey their interest in looking inside an educator’s classroom, their teaching pedagogy and styles, rather than in searching for specific types of teaching resources. In conclusion, the design should iterate and follow the mental model of “connecting with people with similar teaching topics, courses, and level”. We also find the user photos helps with the reliability of the suggested follow subjects.

We continue to test on 3 designs to explore:

• A longer list versus a shorter list of recommendation;

• list versus grid layout;

• detailed versus less detailed information attached to each educator.

• UX copy strategy on the CTA.

We want to know how’d they feel about the process (efforts), are they satisfied with the follow recommendations (validity of the information), and whether they understand the value to of the educator product in Course Hero.

Educator tend to read and validate carefully, thus providing just enough information (ex. course code Bio101) help them to select with more confidence. Furthermore, they expressed a longer list of recommendation is preferable given they have more choices to select from. Lastly, a bigger selecting area on each follow recommendation and a very clear CTA help them to interact with less efforts.

After introducing the feature, educators follow 173 educators in total per week throughout the year (it was 12 follow per week previously).

Aside from this follow feature, we also dedicated times in discussing how’d we could leverage machine learning to help with a more precise recommendation. Different factors are considered including the type of educators (adjunct vs full-time), the focus of the educator (research vs lecturing), timing of a semester when they join Course Hero (exam season, lecturing season). We still provide mild list of educator recommendation and concised information, but most importantly, how’s we let the machine to do the work.

It is through the design iterations and the research that we start to map out machine learning strategy. And it is awared how UX process could be impactful to a machine learning design process.