Retail is currently in its most dynamic and transformative phase ever. That is not an exaggeratory statement — not by a long shot. Just a sample of what was on display at NRF:
Omnichannel fulfillment powered by robots; a diversity of click-and-collect methods and technologies; RFID, video, and mobile embedded into inventory placing, picking, packing;
- Supply chain and shopper applications powered by AI platforms;
- Customer-experience technologies such as people counting/traffic management using mobile, RF, sensors, and facial recognition that link to social and financial data;
- Inventory locating and picking with cellular, video, sensors, and mobile;
- Loss prevention using RFID, video, and facial recognition; and
- Checkout-less stores that integrate all of the above.
Leaving aside the privacy and societal implications of all this for now (read, in this issue, Going Too Far With Technology), lots of innovative functionality in categories that were stove-piped in the past are now being integrated and powered by intelligent AI-driven data platforms and algorithms.
So, will the tail wag the dog? Will the power of technology and the ability to share and leverage information override the kings of merchandising? The fact is, the use cases, improvements, and benefits are so compelling, that to many, the tail must rule. And that will take some getting used to by the various fiefdoms within organizations.
AI for planning and merchandising was one of the more compelling and positive innovations to come along in retail in a long time. Not getting it right — from the design board to inaccurate over or under purchasing, pricing, and allocating — dissatisfies customers who are looking for just the right thing and it costs the whole chain lots of money and missed opportunities.
This is why consumers participate, in so many ways, in giving feedback to retailers and brands. Better fit, colour, fabric, sound quality, form factors, ease of assembly, ease of use and so much more are the customer’s real experience. It’s not all about the checkout system! And applying machine learning to all that data we have been amassing, scanning customer comments on web pages and social networks and bringing some clarity to products and service methods, can ultimately be a game changer in the retail world.1
AI was BIG across multiple platforms, as well as with what we used to call device/data-collection companies. They have now evolved into real retail-solution players. For years we had been advising some of these players — both device companies and traditional software firms — to partner. But that did not happen, certainly not the way it had in the past. Looking at a company like Sensormatic2 as an example, they have leveraged their ubiquitous loss-prevention3 infrastructure presence, all the data they have collected, coupled with an AI-driven information platform to develop some powerful aids to retailers — click and collect, checkout-less payment and exit, and most notably, merchandise planning and replenishment.
While I was talking with Bjoern Petersen, President, Sensormatic Solutions, about integration with ERP, he said we “are overriding the ERP data.” Override? Think about that. That is bigger than integration. He went on to add, “Our inventory data is more accurate and actionable.” Or as Karin Bursa of Logility, said, “when retailers need it.” Think about that, too. If you need more inventory in the next hour, knocking on the door of your ERP back in corporate is not going to deliver those blue jeans to the shelf now.
Again, the tail is wagging the dog. The in-store, hourly, accurate, rich data is actionable in profound ways. And with the information platform, these systems become not only rich details to apply to new use cases, but the definitive source data. (We have been debunking for a while, in any case, the concept of the ERP system being the sole system of record.) Today, our planning and event-driven systems are just more intelligent, because they have richer data, smarter algorithms and are actionable. ERP’s data models just don’t house all this interesting stuff and probably can’t handle the billions and, yes trillions of data points that Harve Light, of Churchill Systems, told me of data that ML needs and can handle.
However, you have to have the data model and be data savvy to be able to deal with the data. That means going to those domain-specific players who have had the rich model, the algorithms, and the people who can make it all work.
In the short run, we are still learning about technologies like sentiment and facial recognition,4
and it is true that many retailers are using these technologies in trial phases or more. For loss prevention we have moved from capturing “old school” video to digital-capturable imaging. And integrating this to other source data, provides a huge treasure trough of information about shoppers.
More to come on retailer tech — people systems such as workforce management, people counters, and making people’s jobs easier with robotics in fulfillment.
_______________________________________________________
1 Note here: some key supply chain/retail planning providers are Churchill Systems, First Insight, JDA, Logility, and Symphony, to name a few, with established and successful customers using their AI/ML solutions. — Return to article text above
2 Previously Tyco Retail Solutions — Return to article text above
3 RFID, video camera, edge boxes, etc. — Return to article text above
4 Note: U.S. government facial recognition systems actually identified twenty-eight congressmen as belonging in their criminal and terrorist watch list. No, that is not a joke. But the point is that unleashed, these technologies can go beyond their intended use cases such as protecting retailers from repeat offenders to racial profiling or other unpleasant outcomes. — Return to article text above
To view other articles from this issue of the brief, click here.