Why you’ll always need more Data Scientists … and that’s a bad thing

In short: resume appeal. Here’s how it unfolds in a totally figurative story:

Congratulations Tommy! You have landed a dream job at blazing hot tech company Gammazone. You start tomorrow as a Senior Manager in the Customer Success department. High five and have a drink.

And at Day 1 you met your boss, Director of Customer Service, Jeff. As the most important department of a fast growing company, Jeff wants you to drive the effort to provide insights for customer satisfaction in the upcoming year.

Now you have two choices:

Buy Tableau for $1000/month.

The Good: Your friend Andy is a marketing PM there. They have good customer service as well as great documentation. Probably you’ll even get a discount. You can start analyzing data and generate automatically updated metric portal from Day 2 (did I mention fancy graphs as well?).

The Bad: Tableau is sweet. But it’s not … sexy. It’s sounds better than Excel (even Microsoft itself is brandishing another thing called PowerBI). But the image is the same. You (Tommy) sitting in the back office churning out spreadsheets and answering calls from higher up. You feel insecure at night fearing that a scripting mistake would tell a totally different story and cost your career. And in 6 months you’ll be as exhausted as you were before going to Garbard Business School.

Here comes another choice:

Hire 5 Data Scientists for $100,000/month.

The bad: did you count the zeroes right? Yes Hesen did double check. There are two extra zeroes! And we are talking about Seattle rates.

The good: Let’s be honest. This route has much better resume appeal. This is so much more sexy. You now have a totally expanded vocabulary: Hadoop/Spark, real time analytics with Storm and Spark Streaming, deep learning, cloud computing, agile development.

And you don’t have to worry about making mistakes:

To start with, the data scientists you hired don’t have time for documentation. Your system is open enough to create a bazillion KPIs that you can brag about at different time. Even if some of them are self-contradictory, people will hardly notice.

Second, if you do get busted for mistakes, you can always add it to the scrum list and let the data scientists fix it.

Of course, you can always blame it on the data scientist. But you don’t always do that as a nice manager, since hiring them with good cultural fit is difficult.

The natural result: Hire 5 Data Scientists for $100,000/month.

As I mentioned in the beginning of this article, this story is totally figurative. I (Hesen Peng) conjured it up while chatting with friends at Tableau. We all observed that companies around the tech community need the features that Tableau provides. But hardly any one is buying the Tableau service suite. Instead, they are going about the more pricey route to build their own stack in-house.

Well, no analysis is useful unless you go to micro level and analyze the profit loss for each decision maker. And we quickly concluded that, in the current market, Tableau is in an awkward position that does not cater to the mid-level management for the lack of resume appeal.

We actually day-dreamed that our figurative manager Tommy got promoted and started to promote data science by sponsoring Kaggle competitions and going to KDD. Well … we’ll talk about that in a different story :)