How to ensure that your data science team isn’t the bloated cost center ?

Priyanka
4 min readApr 15, 2024

--

Photo by Steve Johnson on Unsplash

Who are data scientists ?

For the purpose of this article, I have used the term Data Scientist loosely to describe people who understand AI, Machine learning and data analytics.

Who are “Business” Data Scientists ?

“Business” Data Scientists who have been able to get out of their algorithmic pigeon hole and are able to understand the business and its various aspects as well.

What are the strengths of the data science team and how to best leverage it ?

Data Science team strength is data. They can slice and dice data and build predictive models.

Marketing

Data Scientists have a memory muscle developed that helps them analyze the data in all forms and sorts. Marketing at the moment is largely intuition and experience driven. Data scientists in an organization can analyze the campaigns data analytically to figure out rules that can improve the campaigns performance. In my experience as a data scientist and a founder, we were able to reverse engineer the Instagram algorithms rules governing our target audience by consistently studying time during which the post was being advertised, time of post, length of post, colors, palette as well as language. I was able to achieve the CPM cost INR 9. For the Indian audience the cost hovers between INR 8 to INR 12. This was when the target audience was hard to target, that is, women interested in finance. Moreover, I believed it was possible to bring the cost down further as more data was collected and consequently analyzed.

Customer Success and Service

We all are aware how ChatGPT is the new customer service agent. But data scientists working on customer service data and tickets are in a unique position to understand and analyze the data. When given unfettered access to customer data, data scientists can champion your customers on the table along with the customer success team. Most Data science teams work in an environment that impeds data scientists in voicing their analysis of the customer behavior. By my own experience, a quick analysis of customer behavior based on the device usage and analysis of tickets by multiple techniques can accelerate the customer understanding, recurrent behaviors etc that customer success team needs. Customer success team can read data, but when data runs into GigaBytes or more, this becomes hard. Along with the customer success team, the data science team can meaningfully contribute to decision making at strategy level.

Product

In product, there has always been a hype on how product managers need to be data oriented. At the cost of repeating myself, I would say data scientists are really exposed to customer data through the funnel. While product managers will often see mixpanel as a means to understand the funnel and customer behavior. I have first hand experienced the horrors of how easy it is to unknowingly mess up the mixpanel implementation in the background. Consequently, a data scientist, with his SQL skills can query the database and identify flaws in the funnel, actively participate with product managers to analyze and form a hypothesis and validate it further on.

How can you modify your selection process to hire a “Business” data scientist ?

Currently, data scientists/ML professionals are completely siloed into a box where they are only evaluated on the basis of algorithms they have mastered. It should be mandated top down that business understanding is evaluated in the prospective candidates. While interviewing, candidates can be tested extensively on feature engineering where they can be given the opportunity to show the domain knowledge they have acquired or their ability to think out of the algorithmic pigeon-hole.

At the end I would summarize it by saying the following:

  1. Give Data science team a seat at the table, they are the ones closest to data and consequently your customers.
  2. Data science can have an impact on multiple aspects of the business, leverage that skill. They aren’t your bloated cost centers; don’t treat them as one.
  3. Business Data science is what you need more. Data Scientists who pigeonhole themselves into the world of algorithms and don’t learn anything about the business is not what you want. Atleast, not the full team needs to be super specialized; a mix is good.
  4. For the data science community too, it’s necessary to come out of the algorithmic pigeon hole and learn business so that they can have greater impact on the organization and more importantly be seen as “value” centers; not expensive cost centers.

Eager to hear your comments and thoughts on the post. Feel free to reach out to the author on https://www.linkedin.com/in/nath-priyanka/

--

--