The Candid Voice in Retail Technology: Objective Insights, Pragmatic Advice

The New Abnormal: How To Manage Continuous Uncertainty


There is a lot retailers and their partners can learn from the current COVID-19 crisis, if they are able to see the similarities between retail and other industries. Take this example: during a recent interview on NPR, Ed Yong, a writer for The Atlantic who predicted America’s chaotic response to a global pandemic in a piece in 2018, said this about the medical supplies supply chain:

“The medical system runs on a just-in-time economy, much like the rest of the world, and products are made to order and they depend on these very long international supply chains, many of which have fractured in this pandemic… you could envisage the same problems for all sorts of other areas. I think this is what happens when you rely on a medical system that depends on these large international chains and that really don't have a lot of capacity to flex and surge in the event of a crisis. And that's especially bad now, because the pandemic has spread so quickly that the entire world is facing down the same problem at the same time and is after the same supplies at the same time….”

Parallels Are Obvious

Take for example the supply chain for yeast. Now that we’re all at home in quarantine, baking bread is apparently all the rage (because “nothing says ‘home’ like the aroma of homemade bread!”). But with the pandemic, demand is surpassing what producers would expect even in the busy season. According to a recent article, a representative of the company that makes Fleischmann’s Yeast noted that there’s been as much as a 600 percent increase (YoY) in demand for the product.

After years of making supply chains as lean as possible (meaning, with as few days’ supply as possible), retailers and their supply chain partners are learning the difference between being lean and being agile. One isn’t the same as the other. But the pandemic is just the latest disruptive force in an already-risky global retail environment. Business challenges that before were merely “emerging” are here now, and retailers and their suppliers are feeling the full force of their impact.


The Good News Is That The Technology Is Available

We’ve been writing about AI technologies, primarily in the context of next-generation demand forecasting and digital marketing, for several years. Modeling alternative future-state demand scenarios is a critical success factor in retailers’ merchandise planning. And understanding paths-to-purchase is critical to be able to respond to consumers in real time as they shop in the digital domain. RSR’s research has shown for some time that retailers “get it”, even if they haven’t implemented it. And one of the big reasons for that is that these capabilities are predicated on a level of data science expertise that most retailers don’t have on staff and aren’t likely to have soon – or ever.

One company, Shortest Track, sees the challenges and the opportunity. The company’s CEO, Eric Hillerbrand,recently released a thought leadership piece that pointed out that long term modeling isn’t possible in the face of suddenly shifting unexpected short-term demands - the current case being how people have responded to the coronavirus. He suggested that new short-term modelling approaches are needed now to enable “fast, agile, demand forecasting and digital hyper location planning and execution”. Shortest Track is an “AI as a Service” company, that offers “deep learning” (think Google-like) solution capabilities with a different look, feel, and touch. Its goal is to make advanced analytics accessible to companies that don’t have data science staffs and expertise in-house.

Adapting To The Abnormal Will Be Normal

I wanted to explore what this could mean for businesses. So last week, I reached out to Mike Blyth, president at Shortest Track For Health, to understand how the concept of hyper-local responsiveness, based on short term modeling can help retailers and to learn how hard it would be to get there. I’ve known Mike for years through both of our several career iterations. He was the person who clued me into a book by Stephen Haeckel entitled, Adaptive Enterprise: Creating and Leading Sense-And-Respond Organizations, in 1999. I’ve used ideas from that book for years in speeches and blogs, about how businesses must be organized not as “Plan-and-Sell” companies, which assume steady demand of stable product offerings and for which forecasting is based on probabilities, into “Adaptive” companies that are organized to respond quickly to new insights from the markets they operate in.


An Example: AI And The COVID-19 Disruption Index

AI-enabled planning is central to this kind of a transformation. To illustrate the point, Mike shared with me something his company has developed called the “COVID-19 Disruption Index”, which is updated daily to help companies adjust to the spread of coronavirus and make decisions accordingly. The Disruption Index analyses 20 factors and scores them by county to deliver a predictive indicator, which is used for such things as directing products to locations that will soon experience sudden spikes in demand.

Mike expressed confidence that the Shortest Track Disruption Index will help companies make a faster transition towards being a more agile enterprise. “The idea is to take customers on a fast, agile, impactive AI journey. We start with a small project, gain success, then grow the AI footprint based on credibility. If you can ‘bend the cost curve’, if you can deploy rapidly, if you can ensure you’re lowering risk, you’ve got a high value proposition”, said Mike. “Many companies want to do AI themselves. But, big upfront investments, scarce data scientist skills, and lengthy development add up to a high-risk proposition. The choice may well be ‘value in two months or two years. Shortest Track has the proven AI solutions, the AI platform, we’re ready to go.”

In our conversation Mike made the observation that many companies tend to look at the new analytic challenges based on a “data first”, rather than an “analytics first” approach. “I know a number of Chief Data Officers that had to look for other opportunities. A lot of time was spent getting the data nice and neat, but some of the CDOs didn’t focus on the business issues”, Mike explained. “They didn’t take the time to understand the business’s problems, what the outcomes must be, what analytics are required, and what data is needed. In today’s day and age, you don’t want to have to wait for these huge elephantine <data> structures to be developed – you’ve got to be fast, agile and agile. You’ve got to be able to get to the heart of the problem quickly. If you come at the problem from a ‘data’ perspective, you won’t win – you need to be harness solutions that show a fast route to ROI success, verified by experiences, to drive success.”

Seeing The Opportunity To Help In COVID-19 Times

How did Shortest Track get into the business of analyzing the spread of COVID-19? Mike explained: “Ironically we’ve had AI solutions and data assets we wanted to bring to market as the ‘Consumer Health Exchange’. We had category management, demand forecasting, supply chains, store operations, and digital marketing solutions, that we were planning to be part of a mid-April launch. Oops! So, we asked ourselves, ‘what can we be do to help? Because this <the pandemic> is going to cause incredible disruption.’ We asked ourselves, ‘What can we develop to help our clients? Well. We needed an indicator of business disruption. It needed to be hyper-local. It needed to reflect changing conditions at location <granularity> Every single location is different.’ We saw the need to update the Disruption Index daily, and we needed to forecast the Index 7 and 14 days ahead on reflecting how risk factors are likely to change.”

Based on those internal conversations, Shortest Track started to develop the models (“they were good, and every day they’re getting better and better predictors of risk”, said Mike).

Mike said Shortest Track clients are starting to think about a resumption of business. For example, companies are using 7 and 14 day Disruption Index projections to improve their regional DC’s forecasting, to help them prepare for sudden changes in the demand profile and then adjust allocations accordingly.

The idea is to get clients the advantage of the hyper-local risk factors without forcing those companies to replace their existing solutions. “We deliver a single Disruption Index”, said Mike, “it’s another input into their process.” Essentially Shortest Track is shielding the forecasting engine from the complexities of hyper-localized demand attributes, and instead analyzing the complex data to deliver a result – an indicator of risk, that can be consumed by the forecast.

Winning In The New Abnormal

Our benchmark data shows that there isn’t a whole lot of dissent when it comes to the value of AI-enabled analytics in retail. For example, in a new (as of yet unpublished) benchmark on the state of AI enablement in the retail supply chain, the #1 opportunity identified by retailers to improve supply chain performance is to “use predictive models to anticipate supply chain disruptions at the individual SKU level and recommend cost-optimized corrective actions”. That sounds great, but the question has been “how to get there from here?” Retailers’ analytical and planning systems are largely based on examination of transactional data that have been gathered from operational systems. Trend analysis is the basic tool used to inform decisions, and at best “localization” means store clustering.

To get from conducting a “plan and sell” business to becoming a “sense and respond” one, retailers can’t be forced into a forklift replacement of all their planning and analytical systems. For one thing, most retailers have neither the time nor the money to do that. For another, even if they did, they would still have a very difficult time finding the data scientists needed to get the job done.

That’s why new companies that seek to help retailers “flatten the cost curve” for next generation analytics are getting a closer look. Understanding the variables that affect hyper-local demand is the next step. “But you’d better get there sooner rather than later, because that’s where the battle is going to be”, said Mike. “How you are going to service people in a particular market in particular conditions with a set of parameters in a way that differentiates, will be the key to success.”


You must be a member and logged in to view comments

Articles & Opinions April 28, 2020
Related Research
Privacy Policy

This website uses tracking technologies to learn how our visitors interact with our site so that we can improve our services and provide visitors with valuable content. By continuing to use this site, you are agreeing to our privacy policy.