I wrote before about how data science is a vague concept that people don’t quite understand. To illustrate this point, I’d like to tell you a conversation I once had with a VP that went something like this:
Me: "I can build a Machine Learning model to select the top users most likely to make a purchase."
VP: "You mean like a regression model?"
Me: "Yeah… something like that." ( resisting urge to make a facepalm )
If all of your modeling efforts is viewed like a regression that can be done in Excel how do you get across the value you’re providing as a data scientist?
Today I’m going to share a few ways I’ve translated machine learning results into business impact that **** executives and your stakeholders can understand.
1. Show improvement relative to company KPIs
Executives need to relate how the model can improve on the company KPIs that they care about – i.e. conversion rate, user engagement, renewal rates.
For example, you built a model that found the top features increasing user engagement. By having the product team highlight those features during the user onboarding experience, you’ve helped improve the user engagement KPI for the team.
This shows the VP of product the value of the model you built that otherwise would’ve taken the product team a lot of time to find.
2. Show incremental revenue impact
If there’s no KPI that’s a good fit to the model you built or you want to show the size of the relative KPI improvement, calculate the the incremental revenue impact to the business.
For example, you’ve built a model that has increased conversion rate from 1% to 2%. A 1% difference doesn’t sound like a lot but what if you also said the 1% percent increase in conversion rate translates to $1 million dollars in additional revenue.
Wow, now you’ve really shown your value because that’s more than your annual salary ( and if you make more than a million I want to know your secret ).
3. Show cost reduction or time spent
If your model doesn’t relate to a company KPI or you can’t associate a revenue amount then try to show the value by reduction in costs or employee time.
For example, I developed a model to select the top single users that were likely to convert to team licenses. Instead of two sales reps spending their day looking through millions of users in the database to find the best leads, they could now select users with the highest probability to purchase and increase the chance of conversion in less time.
I didn’t know the exact conversion rate increase the model could have but it was enough to convince the sales team to use it because there was clear value that this would reduce the time spent by the reps manually looking through the database.
As data scientists, the focus is on model accuracy but occasionally you have to step back and look at the big picture to demonstrate the value machine learning models can have on the business.
Thanks for reading! Please leave a comment if you’re interested in hearing more about a specific data analytics topic in the future.
You might also like…
How I Used a Machine Learning Model to Generate Actionable Insights