I spent past few days reading, thinking and talking to people about this whole initiative.

The questions in the spreadsheet are great place to start aggregating metrics that would give us a rough idea how we're doing in delivering value. These metrics are **quantitative**, **reporting metrics** that gives us a idea of **the past**. Correct me if I'm wrong, most of these metrics can be assembled by our BE team. Maybe except for second and third question.

What I need help with is finding the right problem that can be solved by machine learning. A problem that minimizes effort and maximizes impact. And for that I think we need different kind of data, user feedbacks (structured & unstructured).

We can start by making it _exceptionally easy_ for our users to complain (for both admins and learners). Users who complain often are much likely to churn. Complains are indication of _what's likely to happen_, rather than mere reporting of the past. We should lower the friction and allow users to leave feedbacks, without leaving the current page they're on. These **qualitative**, **exploratory metrics** will give us predictive understanding of **the future**.

Only then we'll be able get a overall picture of what we should be doing _now_ so that customers won't be abandon us in the _future_. We can even add ML to do some sentiment analysis on these feedbacks. This will also help us understand the second and third question in the spreedsheet.

However, most of these work sounds like something the product team or UX team should be looking into. Perhaps, we're already collecting some of these feedbacks in other forms that I'm not aware of, pardon my ignorance if we do. I know we already have [Fider](https://learndot-enterprise.fider.io/) but that's not enough, the users shouldn't have to leave their current page to give feedback. Google does this pretty much on every page.

I was big fan of [Get Satisfaction](https://www.youtube.com/watch?v=YYTtPjvA9xk) (I even recreated their exact widget with full fledged forum to embed in one of my sites long ago) but it seems like they're not in the business anymore after the acquisition. I came across [Appzi](https://www.appzi.com/) which looks pretty minimal _and_ promising.

Getting unhinged stream of user feedback is not a blocker for our ML intiatives, but it will guide us in the right direction.

Also, I came across few more of our competitors. Especially, after looking at what [Udemy for Business](https://business.udemy.com/) offers, I find it hard to be optimistic.

References
- [What Makes a Good Metric?](https://www.oreilly.com/library/view/lean-analytics/9781449335687/ch02.html)
- [Google: Ten things we know to be true](https://www.google.com/about/philosophy.html)
- [From design to development, user feedback shapes Google’s approach to accessibility](https://www.blog.google/outreach-initiatives/accessibility/design-development-user-feedback-shapes-googles-approach-accessibility/)
- [Various resrouces @ Udemy for Business](https://business.udemy.com/)
- [4 simple steps to designing a strong user experience](https://www.thinkwithgoogle.com/marketing-resources/experience-design/strong-audience-design/)
- [How consumer needs shape search behavior and drive intent](https://www.thinkwithgoogle.com/consumer-insights/consumer-needs-and-behavior/)