The data science field is so young that this will be its first recession. And make no mistake, this will be a test for all of us. It reminds me of that one quote from Warren Buffett:
It’s only when the tide goes out that you learn who has been swimming naked.
But this is also an opportunity – a chance to look at each aspect of our field and keep the good while purging the bad. That’s what a recession does. The hottest field, prior to the Great Financial Crisis (2008) was quantitative finance (the spiritual predecessor to data science). And finance definitely took its lumps during the crisis, not least because its flawed risk models were part of the reason the crisis occurred in the first place. But the recession proved to be good for the field. The profession emerged leaner, more battle tested, and theories that made no sense were discredited and deemphasized (though many doubtful ones still remain).
So let’s try to imagine some ways that the data science field might be impacted and changed by the current coronavirus recession.
Consolidation And Job Losses
This is the painful part. While we data scientists are fortunate in that we can work remotely, furloughs and job losses are inevitable in an economic crisis of this magnitude. The hits to overall employment, consumer spending, and economic growth are too much for any profession to escape unscathed, let alone one that was arguably running too hot heading in to the recession.
Budgets and headcount will be cut and peripheral projects that have little potential of generating cashflow will be put on hold or canceled. None of this is any fun, but what it will do is force the entire profession to honestly assess its own value add. Companies will ask and attempt to answer the following questions:
- How big does my data team really need to be?
- Do I need data scientists at all?
- What is the expected ROI (return on investment) of data science projects?
The answers to these questions will vary by company. Companies with a financial buffer, proprietary data, and management willing to think long term will answer these questions much more positively than companies that are strapped for cash. When a company has its back against the wall (due to debt and/or a collapse in sales), it’s easy and sometimes even necessary to be shortsighted.
So what should we, employees, do to prepare? Sadly, the decision of who gets let go is often a very subjective one and not in our control. It’s often mainly a function of how much someone makes (the objective is to cut cost after all). And I won’t tell you to make yourself indispensable – that’s not realistic and it’s not something that can be done in just a few weeks (you either already are or aren’t indispensable at this point).
Rather, it’s time to be proactive and realistic rather than passive and foolishly hopeful. Study your company’s financials and unit economics. Does your company, even during the best of times, barely breakeven. Did it require round after round of capital injections from investors to stay afloat? If so, it’s time to look for other opportunities. The good news is that the companies hiring now are definitely both competitively and financially strong; the bad news is that there are much fewer such openings at the same time that there is a lot more competition for those openings.
Model Breaking Paradigm Shift
The world we come back to when we finally finish lockdown will feel very different from the one we left when we all went into hiding. Business models that made great sense a mere 8 weeks ago may no longer be feasible in the near to medium term (e.g. StubHub, Airbnb).
That means consumer behavior in the post coronavirus world will likely also be very different. This, unfortunately, reduces the value of a lot of the datasets out there. An easy bet to make right now is that many models (whether it be a recommendation engine, a user behavior model, a time series forecast, etc.) will break in the coming months.
The phrase paradigm shift gets thrown around a lot, and usually, it’s unwarranted. But coronavirus and the recession it caused are true paradigm shifts. Until there is a vaccine, the way we interact, do business, and work will remain dramatically different. Models trained on yesterday’s data will no longer be as predictive, sometimes dramatically less so.
I have 3 random thoughts about this and how it makes the future of data science both more challenging and more exciting:
- The race is on to build the models that explain user behavior in the new world (of social distancing and economic downturn). The samples will be small and smart inferences and intuition will be key. Some basic knowledge of psychology and an understanding of how stress and anxiety impact behavior will probably come in handy at some point.
- This also doesn’t mean yesterday’s models are done for. Even if a vaccine takes a long time, people’s behavior will begin shifting back towards the norm over time (hopefully not at the expense of a higher infection rate). Fear tends to recede as time passes and we get used to the new state of the world; this pushes us back towards our normal state (but probably not all the way; severe enough events like an economic depression can permanently shift people’s behavior).
- In a world of heightened uncertainty, there will be more opportunities to nudge people for better or for worse. Data science models are as much about suggesting as they are about explaining or predicting. The whole point of a recommendation engine is to shift peoples’ behavior (by nudging them towards that product we want them to buy or service we want them to sign up for). As the world and people’s mindsets start the long journey back towards some new, somewhat different state of normalcy, there will be unique and perhaps once in a lifetime opportunities for companies to earn a place in that new normal – demonstrate real value to a customer now in these dark times, and the ROI could be huge.
A Suggestion To Companies (And CEOs) – Build Your Data Expertise Now!
A crisis is an opportunity to be your best self. It’s ironic how companies that spend billions of dollars on PR, carbon credits, company jets, and stock buybacks during flush times can so coldly turn around and lay off a quarter of their workforce during a recession.
Yes, I understand the fear of bankruptcy. But the whole point is to build up a buffer during good times so that you can invest when everyone else is running for the hills. So CEOs, if you’ve been fiscally responsible and are now flush with cash, consider investing in the goodwill of your employees. Capital goods like airplanes, factories, and data centers are not the only investments out there – your employees and their long term development are the most important investments your company can make.
This pandemic and recession won’t last forever – if you cultivate and develop your Analytics and data science teams while everyone else forgoes and fires theirs, imagine the huge competitive advantage you can have a few years down the road when business begins to boom again.
In the past few years, everyone hired data scientists and data engineers in a somewhat futile attempt to gain a competitive edge via "big data". But doing the same thing as everyone else is not how you get an edge. You need to be different (in a good way) to outperform. A crisis is a rare and precious opportunity to zig when everyone else is zagging – to invest, build, and act with the long term in mind when everyone else is is focused to a fault on the short term. To that end, I have 2 suggestions for CEOs and Chief Data Officers:
- Offer paid internships to both recent graduates and experienced folks fallen on hard times. In my opinion, the Data Science interview process has always been pretty hit or miss with a high false negative rate. Rather, quickly screen candidates for curiosity and hunger, and then give them a chance to shine (and earn dollars and a full-time job) via an internship.
- I’m not advocating that companies pay data people to do nothing. Rather, embrace a hackathon mentality. During normal times, everyone is busy putting out fires from existing customers and on-boarding new ones. A recession (when there’s sadly fewer customers to service) is an opportunity to focus on developing new ideas and products (and strengthening existing ones). So my advice is to build and test out a bunch of new stuff. That way when the economy starts to recover, your company can hit the ground running.
The Sun Will Shine Again
We’re in for some hard times in the months to come, but make no mistake, we will come out the other the side OK, albeit with a few scars. As will the data science field and profession. Recessions have a habit of shaking out the loose hands. In 2008, many financiers who were doing it purely for the paycheck, left the industry altogether. The same thing will probably happen now for data scientists (my guess is that the intense competition for a fewer number of jobs will result in each of those jobs becoming less well-paid; basic supply and demand). But if you are truly interested in the field, stick with it, be prepared to Work really hard, and eventually, you will find your chance.
Cheers and stay safe everyone!
