I was in a meeting with a senior business leader in the health space who had just lost her patience for hearing the word “algorithm” as a catch-all for the value our company could deliver through different technologies.
It’s not that the senior business leader was technologically unsophisticated. Not even close. In fact, she’s a genius. However, she was frustrated, wanting specific solutions to inspire consumers take on healthier behaviors, to help doctors detect disease more quickly, etc..For their part, the domain experts and product engineers understood the challenges she’d outlined and were brilliantly plying their trade to solve them.It was a matter of translation. The entire team needed to understand and articulate how the math achieved the mission.Forbidding the use of “algorithm” was actually a great thing.
Without that shorthand, the working teams articulated the problem and solution in customer terms, then in business terms, and then put innovations in market with the clear articulation of value to the customer.Currently my time is spent heavily with people steeped in advanced analytics. I’ve spent the last year trying to understand the psychology of data scientists and the dynamic of the marketplace.After a highly successful beta test and recent commercial launch, I’ve found myself wondering whether we’ve limited too much who we define as our customer.Communicating value to a data scientist has become straightforward. It feels easier to talk about process and technologies and how they contribute to speed and accuracy.
We can speak in shorthand and I’m reminded of the experience of striking the word “algorithm” from discussions.Words I’d like to eliminate here include: ensemble, training, label (outcome), and my favorite term of all “confusion matrix”Though my reasonably developed stats skills barely earn me a pat on the head in data science circles, I am fluent in the language of business: cost, price, sales volume etc.I worked with our data team to make the tool more business person friendly by incorporating business logic and reporting into the predictive analytics platform we’ve launched.
As predictive accuracy and risk tolerances change, the value of a particular model is articulated in revenue and cost terms.This a great first step. Unfortunately, business has a lot of shorthand too. We have more translating work ahead of us. I’m already working with the data science team to create documentation and reporting that is specific to the types of challenges businesses face.It’s like a micro-UN learning to understand each other’s language and culture so that we can move our innovation forward. If you are interested in checking out the platform (including business logic), log into www.purepredictive.com/registration. It’s free for 30 days and no billing info required.