The Smiling Data Scientist

This has happened to most Data Leaders. The CEO sends you an email, invites you for a meeting or (worse still) pops down to have a quick chat.

“Good news! I’ve been speaking to my good friend at The Big Consultancy. He has offered to do us some free work. His Data Science team has just finished a fantastic analytics project for our competitors and he has offered to do something similar for us. Even better, he said that it will not require any of your time and will only take four weeks. You’ll get an email later on.”

I’ll give you everything you ask for

After a couple of emails back and forth, you clear your diary and meet with Kirsty, the Smiling Data Scientist (SDS).

The SDS lays out the plan…

“First, you need to give me all of your data. Don’t care which format. We’ll profile it and send you back some insights. Then we’ll do some analysis and then we’ll draw some graphs which will impress your CEO. That’s it.”

“OK”, you say, “How about I give you the dataset that we use for reporting. That is well understand and documented, its nice and clean too.”

The SDS looks at you with a little pity. “No, I want all of your data. How can I expect to do magic with a reporting cube. Just give me all of the data.”

The meeting ends. “All of the data” you think. “I’ll give you all of the data”.

Montage scene followed by presentation

If this story were a movie, the next few scenes would be a montage of scrolling numbers, furrowed brows, late night pizza, post-it notes and eureka moments.

As the day for the final presentation looms, you offer an olive branch to the SDS. “Hi, just wondering if you want any input on the analytics and outcomes? We could go through the thoughts so far and I could help with the messaging.”

The SDS isn’t looking as chipper as before, but the confidence hasn’t wavered. “No, that’s fine. We are just reviewing the presentation internally and then you can see it when we meet with your CEO.”

Presentation day comes. You feel that you have done everything asked of you, but you’ve got a bad feeling on the way that this is going to turn out. You enter the room and look around. It’s you, the CEO, the Head of Sales and you three are outnumbered by the team from The Big Consultancy.

As the slides are presented, the Head of Sales gets visibly frustrated. “Yep, we knew that sales were down there. Yes, that’s true of that region, but it’s where we’ve strategically pulled out of. The plan is to go with a new partner, but that won’t be in the data yet.”

At the end of the meeting, the CEO sums up a disappointing outcome. “It looks like you weren’t able to find anything really insightful and actionable. Why is that?”

Hmm, looks like the there a bus heading your way and The Big Consultancy are getting ready for the big shove.

“Well, we spent a lot more time than we expected cleaning up the data. It didn’t seem to join up very well and the data dictionary wasn’t very good. Then we didn’t really get any guidance on where to look, what your business strategy was and what you have tried before. Finally, whilst we found some insights, we haven’t had chance to discuss which of them are practical in your organisation. Give us three more months and we should be able to start making you some money.”

The CEO thanks the team from The Big Consultancy and everyone goes home. You feel a little guilty and realise that you’ve damaged the CEO’s view of the usefulness of data. Next time you are going to play it differently.

What went wrong?

Whilst The Big Consultancy were guilty of hubris, they did have a few valid points on their diagnosis of the problems:

• The data was too big, too unknown and too dirty. How can you expect to produce amazing results from inputs that you don’t understand?
• They didn’t know anything about the business strategy and the context for the work. They assumed that they could take your competitor’s problems and that they were the same for your company.
• They didn’t understand the business operations well enough to be able to define actions. The intention to act was missing.

AI the Smiling Data Scientist

Why am I writing about this today? Well, AI/Machine Learning is the new Smiling Data Scientist. The CEOs are again saying “use some of this capability to find me some insights”. And we are all in danger of sleep walking into the same traps as the big consultancy.

As data leaders, we need to treat AI as an opportunity and get ready for it. This means:

• Get your data ready for AI - Get it cataloged (in some form), understand its limitations (breadth and quality) and get it loaded;
• Get your tech stack ready for AI - Do you need to move data around? Do you have model management tools? Are you sure to can manage data security and protect against leakage? Review your architecture with some experts (I know some who can help) and build a plan for implementation;
• Get your policies ready for AI - Who can build models? How will you authorise models and prove their safety? Build a data ethics capability and working group to help you with these;

If you would like to talk about getting ready for AI, contact me!