What was the scope of their involvement?
We’ve worked with them on a number of projects to help us evaluate our telecom data. In one case, we ran a project for internal stakeholders who’re responsible for our B2C points of sale and branch locations. The other project was for large fast food chains.
TEONITE helped us implement different machine learning models in order to locate relevant patterns in the data that could predict which customers would buy our services, as well as their motivations for doing so. We’d then use those calculations to make business decisions, such as where to open a new store for internal customers.
For each project, we started off with discovery workshops involving internal stakeholders and external vendors. We identified what kind of data we were seeking and tried to identify significant variables. Next, we provided aggregate data for TEONITE to cleanse and process. There was a lot of back and forth between their team and ours to ensure the data was complete and consistent. Once we had results, we experimented to see if the patterns made sense with consumer behavior.
After we had a working model, TEONITE helped us build an application that could estimate the accuracy of our different machine learning algorithms. It’s able to handle hundreds of variables and analyze years’ worth of data. Once they got us a working forecast model, we used to provide additional insights to our internal staff and partner organizations.
What is the team composition?
I worked with their CTO, a project coordinator, and a data scientist, who was responsible for implementing the algorithms. Several other people also worked on the project, but I didn’t interact with them directly.
How did you come to work with TEONITE?
I’m not really sure how we came across them. We’re involved in a networking organization for startups, so it’s possible that’s how we got in touch. Once we met them, it was clear that they could offer us a lot of value.
What is the status of this engagement?
The project officially ran from January–June 2018, but we also handled presales and business development prior to the official start date.