Where Have All the Retail Data Scientists Gone? SAS Analyst Day Report Out
Username: Name: Membership: Unknown Status: Unknown Private: FALSE
Last week SAS hosted its largest-ever analyst meeting in Steamboat Springs, Colorado. The company had lots of good news to share: topping $3B in revenue, strong growth regionally, and continued growth and success across its solution areas.
However, despite its success, I find SAS to be one of those companies that has struggled a bit in marrying its horizontal strengths to industry depth. And the company is not unique in this struggle – “speaking retail ” to retailers isn’t easy, especially if your solution is not exclusively focused on the retail industry. If it was easy, I suppose RSR wouldn’t exist. And to be fair, retailers in general tend to be a tad too righteous in asserting their unique qualities. Innovation doesn’t always come from within an industry – sometimes the best ideas are transplants from other industries. But I’ve found retailers to be particularly hard-headed in insisting on looking only to their own for new ideas.
But back to SAS. I’ve often scratched my head over SAS’s road to the retail market – not that the company doesn’t have the right solutions or the right message. But somehow that message never seems to get heard. I’ve long thought that SAS’s customer insights capabilities are some of the best-kept secrets in all of retail. And their understanding of the finer points of size optimization is difficult to match.
So why isn’t SAS just absolutely killing it in retail? Well, I think I finally figured it out. During the presentations in Steamboat, the company referred multiple times to four key user groups. I’ll paraphrase a bit – the four are business management (basically, deciders), business analysts (the people who prepare the analysis that helps deciders decide), IT staff (which keep the infrastructure running), and data scientists (who basically make the data useful and easily consumable by the business analysts). Data analysts also do a lot of the heavy lifting around predictive modeling – these are the white lab coat people that SAS has historically infamously courted.
Data scientists? I can honestly say that I have never met a data scientist in retail. They are a class of user that appear to have all the frequency of unicorns and dragons in the retail enterprise. This must be why business intelligence and analytics is such a relative mess in retail. I mean, aside from the startling finding from our latest BI benchmark that there is a large contingent of retailers out there who persist in believing that intuition is more important than information for a lot of functions, if business analysts are driving BI investment decisions then I can easily see how the advantages of an enterprise-wide BI platform go down in flames as each group fights for the tools they like best. Throughout my career I have met retailers that easily had three or four different BI solutions – all of them world-class solutions that should have individually been able to meet all needs.
The strength of the SAS story is platform. It relies on the existence of data scientists that help business analysts model and understand their data so that the analysis they do becomes more effective. When data scientists exist, data can actually be democratized (as in pushed closer to the front line), because it is the data scientists that ensure that the data is described, staged, and modeled in a way that makes it more accessible to casual users. And it’s the casual users that best understand the problems they are trying to solve, just as data scientists best understand the opportunities and constraints of the data that is available to provide insight into how to best solve problems. Only when the best of both worlds is combined can real insights be gained – and acted upon.
So why are there no data scientists in retail? Well, first, they tend to be expensive resources. Retailers aren’t competing with other retailers for data scientists, they’re competing with investment banks and pharma and insurance and telcos – industries that have known the importance of data actually for a lot longer than retail has. The demand for these resources is high enough that retailers can’t really compete well in attracting or retaining them – a point our benchmark participants made when they said that analytics resources were scarce.
But for a long time, the only data that was important to retailers was product data. Specifically, sales data (demand) and inventory levels (supply). That’s it. There are still retailers out there (though they are fewer in number than they used to be) who believe that straight demand – which products were sold through which locations – is all they need to know about customers. All that behavioral, demographic, or attitudinal stuff is just noise. And so why bother with data scientists, in that world view? The data types are well-known and well-understood, so the idea of investing in resources that just basically shepherd data seems a little silly.
Of course, retailers have added a lot of new kinds of data in the last decade or so, most notably a realization that customer data is actually pretty important, and may have a lot of bearing not just on our understanding of product sales, but in its own right. And, of course, there’s the whole channel thing. Sales don’t just happen in stores, and the relationship between influence and sales has become very messy as new channels and touchpoints are added.
Then top it all off with the extremely messy realm of sentiment and social media, and all these issues become compounded. SAS put forth a panel of four customers, two in media & entertainment, one in insurance, and one retailer. Across the board, all four of them said that learning how to marry their internal data sources with that wider, messier world of social media and sentiment was a top priority – not for customer service use-cases, but as an early-warning system for strategic business decisions. All of these issues point to a growing need for data scientists.
I think the big question for SAS in retail in 2014 and beyond will be focused on these data scientist resources. Will retailers recognize and respond to the need for this layer of resource within the enterprise? If they do, then SAS’s job in retail will become much easier. But if retailers resist, then the road will be much more difficult, at least in this vertical.
Either way, SAS continues to deliver solutions that retailers need. Between merchandising, customer intelligence, and even fraud & security, the solutions that sit on top of SAS’s platform have enormous potential in the retail industry. The challenge will be convincing retailers – who even in a data-overloaded omni-channel environment remain somewhat mistrustful of data-driven insights – that the platform, and the right people using that platform, is far more important than the stated needs of any individual group within the enterprise.
Unfortunately, in retail, that’s a tough sell.