The 10 TLDR Commandments for Data Scientists

Your value is measured not by your technical skillset, but by the trust of the people who work around you. Guard it fiercely. This is how.

Thou shalt heed who is paying ye and why

The only thing that matters is the value you bring your client. Let others focus on the ‘how’. Commercial awareness sets you head and shoulders above your peers. Allow ‘why’ to influence how you tackle your day. Conversely, if the business makes the same revenue with your efforts than it could do without, you are wasting your life.

Thou shalt simplify without misrepresenting

A simple analysis that creates a commercial narrative is preferred to a highly predictive model that does not. A beautiful SEM will be chosen ahead of a set of correlations that make the same point. Explain everything in lay terms. Use analogies and real-world examples to get your point across. It’s your job to document the litany of assumptions and technicalities – not your client’s. No one cares where you buried the bodies. Build defensible funnel-like arguments from objective to conclusion or go home.

Thou shalt script everything from scratch

Reliability is a precursor to validity. If your analysis won’t hold water when fresh data arrives, then sharing it will do irreparable harm. Don’t associate yourself with black box methodologies that insult your common sense. Everything you do eventually returns to help or haunt you. Use Git. Script everything from scratch, no matter how trivial.

Thou shalt work as if everyone is watching

Never share unfinished work - if you make a breakthrough, spot an error, or develop a new angle… you cannot recant without harming your reputation. Accept that your client will share your deliverables with the competition, so treat it like your business card.

Thou shalt respect thine clients and colleagues, even the ones ye don’t like

Make more opportunities for juniors than you had; share without hesitation. Lead with anything but authority. Your clients shine when you anticipate their next move, not when you wait for them to catch up. Pay more attention to those who are not technically minded – they have the most to teach you.

Thou shalt believe that ye know less than others

Never feel intimidated or suffer those trying to assume the higher ground - this field is too dynamic and broad for that. Enjoy cutting them down like it’s a sport - you owe yourself that much. You can easily spot the true heavyweights - they are without exception, the humblest and quietest in the room.

Thou shalt be curious

…and when you can’t, fake it. Revel in nuance. Both help you tackle bigger problems and tell better stories. Do it by force of habit. Begin immediately.

Thou shalt never promise anything specific

Ignore pressure to assume responsibility for data unless you designed its collection. Make plain your assumptions. Specify your deliverable. If the data quality is poor or the conclusions disappointing – that is no reflection on you, unless you claimed otherwise. Don’t tolerate others who overclaim on your behalf.

Thou shalt know thine niche then study it relentlessly

Follow what interests you, develop a broad appreciation, but don’t bother trying to learn everything. That’s just stressful. To escape tutorial hell, remember why you want to learn new things. The most successful data scientists do one or two things very well and muddle their way through whatever else is thrown at them. Your continual learning inspires not only your own confidence but that of everyone around you. Don’t cite online boot camps on your CV. It only opens you up for a line of questioning in job interviews, and role responsibilities, that the videos can’t prepare you for. Steer clear of companies who don’t know this.

Thou shalt outgrow whatever ye are doing today

As you read these words, software and automation is robbing you of your technical advantage. While building your skillset, study business, people, and management with equal gusto, because no one can take these from you. Drown out the so-called experts. Don’t get trapped by their stereotypes, labels, levels, specialisms, job titles, infographics, ideas of progression or whatever they’re peddling. Rest assured that by the time you reach wherever they claim to be, nothing will be the same, except for these ten commandments. Rinse and repeat to wipe away all who stand in your way.

And if you do only one of these… seriously, trust me on the scripting bit.

Compiled in memory of the late Geoff Inglis who taught me how to model data with purpose.