Bring Location and Company data into Contributor Insights

[isabensusan]

Context

Our original designs for Contributor Insights included charts for contributor's Company and Location, which would help users understand their contributor base more providing interesting insights for communication, sales and project management purposes.

The concepts were dropped because the data wasn't in the state we needed it to be, and they got deprioritized, but from recent conversations with users it has been made clear that this could be valuable information to add to the platform.

As an example, the Serverless.com use case is to be able to drop a list of people into a contributor insight to understand who they are, where they work, and what projects they are working on. Adding company and location data (but especially company data) would be extremely valuable to them. And in the case of Prisma, since they are looking into contributor insights to help their sales funnel, these insights could help them shape their sales strategy.

Main Issue: free form text data

Since both company and location are free form text fields on GitHub, this results in fairly inconsistent data on our database. One person might have "OpenSauced" as their company, while others might have "open-sauced", "opensauced", "@OpenSauced". It is the same case for location.

In addition to this, people don't always fill in the field with serious information, ie. when users put "earth" or "the universe" as their location.

Charts would not look like the mocks

The original concept for the Company chart was a bar chart, but given that we have incomplete and inconsistent data, it wouldn't look like the mock:

In reality it would most likely display duplicate companies with slight grammar differences and typos, which would still provide good insights, but could become useless in a lot of cases.

Same happens with the Map, users whose locations don't match an exact location in the map simply won't show up, resulting in incomplete data.

Proposal - Normalize the data

From chatting with @jpmcb, we came to the conclussion that:

1. We can still ship the charts with what we have right now as an MVP, adding a disclaimer that the data isn't 100% accurate. We can even do a quick calculation on how many users are missing a location or company and add that to the disclaimer like "Only 68% of users have added their location information"
2. We can run the data through a similarity algorithm to clean up duplicates as much as we can.
3. In the future we could even think of a paid "Enhance Feature" that enriches and improved user data for more accurate insights.