ERP and Supply Chain Innovations for Retail at NRF 2016

This was the biggest NRF show yet, with about 34,000 attendees and 580 exhibitors, covering 228,000 square feet of the Javits convention center in the Big Apple. There was plenty of innovation on hand from ERP and supply chain vendors. Here we look at important developments from four of them: JDA, Logility, NetSuite, and Infor.

JDA

JDA Labs — Foundation for the Reimagination of JDA

Last year we wrote about JDA Labs and how they are co-innovating with their customers. Since then, Suresh Acharya has taken over the labs and I had the pleasure of speaking with him at NRF. Coming from the product side, Suresh has oriented the labs to practical innovation and incubation. The core team of about 40 permanent scientists and engineers is supplemented by another 10 or so from JDA’s product groups who take one year sabbaticals, helping the labs to stay closely connected to the rest of the organization. This interweaving of the product managers helps ensure that innovations address real-world needs in a way that works well with JDA’s current portfolio. Suresh characterized the labs work as falling into three categories:

• Greenfield Research — New areas not currently being addressed, such as Returns Forecasting. In the past, retailers used a rule-of-thumb percentage to forecast how many returns they expected. The rise of e-commerce and other factors has made returns skyrocket. It has become important to manage goods being returned as an integral part of the overall inbound supply of goods. Unlike the deterministic rate of supply from the traditional supply chain, returns provide a more stochastic rate of supply.¹ Another greenfield area JDA Labs is looking at is 3D visualization for solving supply chain challenges.
• Collaborative Innovations — JDA Labs collaborates with their customers on solving specific issues. For example, a large tire manufacturer wanted a way to improve space utilization in the truck when shipping tires. JDA’s product team developed a good rule of thumb heuristic, then the labs created further improvements with runtime load-building algorithms, using combinatorial optimization to squeeze out even more efficiency, while still respecting the required rules and constraints for transportation loading.²
• Transformative projects – n example of a big transformative project is JDA’s partnership with Google and Retail.me, building out of the next generation of retail planning, incorporating external data and events and machine learning. It includes segmentation of customers by their behavior (what and how they shop and buy) as opposed to traditional demographic segmentation. Retail.me is more than an update to allocation and assortment planning; it is a whole new way of thinking about the problem. We talk more about Retail.me below.

Other examples of ideas the lab is working on include risk-aware planning (probability forecasts)³ and augmented reality. They have a multi-stage process where many ideas come into the funnel as new concepts are being discussed. The ideas that look promising are assigned to an intern or associate to do a bit of due diligence on feasibility and whether it addresses a real need. For ideas that pass that filter, the labs will do some sort of proof of concept (PoC), when possible partnering with one of their customers. If the PoC proves out the merits and value, then it goes into the product roadmap. From these innovations, JDA Labs is generating about a patent per month.

The labs have partnerships with various universities, such as the University of Montreal, Georgia Tech, MIT, and the University of Maryland, as well as internship programs. Being a central group — a sort of hub that talks with customers, JDA product groups, implementation services, market research, and academia — they can spot challenges or issues that arise in one area that are very similar to another challenge in a different area. For example, optimizing how many of what items a retailer should put on the shelf has many similarities to figuring out how many of what items to put on a promotional flyer to extract the best value out of that limited real estate.

Retail.me

One of the biggest tangible outcomes of JDA Labs so far is Retail.me, JDA’s next generation SaaS planning and execution suite. This will be a full suite eventually; initially they have launched the first application, assortment planning and execution, which includes building an assortment, selecting items, and creating a buy plan. A major part of it is a new approach to segmentation based on customers’ behaviors, rather than just traditional demographic segmentation.

Behavioral Segmentation

Retail.me’s segmentation algorithm ingests behavioral data and mines it to identify similar groups of customers based on their behavior: what they buy, how they buy, do they buy online, how they shop in the store, what motivates them (are they price-driven, promotion-driven, impulse buyer), their ‘returns behavior’ (% items returned by channel, where returned), and where they live and travel to. This drives segmentation algorithms that can be applied to make decisions such as how many of which products to buy and, critically, where in the supply chain/store network to place the inventory. Historically assortment only applied to product choice and price, but the rise of omni-channel forces retailers to think about fulfillment as well. Further, knowing returns preferences by behavioral segment, the system can do a much better job of forecasting returns by location and adjust the actual allocation to each location accordingly.

One goal for Retail.me is to manage assortments by exceptions, rather than rules. This helps in forecasting sales of new items; the algorithms can do analysis based on attributes, such as the color, style, sleeve-length, etc. of the new item and make its forecast based on history from a combination of ‘like items’ with similar attributes. Part of the appeal is the need to speed up the pace of planning and execution and do it iteratively within today’s very rapid (e.g. 8-12 week) fast fashion cycles.

The Next Target for Retail.me: Demand Management

Retail.me is a pure multi-tenant SaaS offering, with continuous releases of new functionality. After assortment planning, the next big need is to transform demand planning, including leveraging real-time inventory visibility from RFID, beacon technology, and other information that can provide much more granular shopper behavior (such as when a shopper took an item off the shelf and then put it back — or tried it on then didn’t buy it). JDA said the paradigm of demand management has to transform from ‘forecast and measure’ to ‘sense and respond.’ Retail.me represents really exciting and important new developments, another indicator that there really is a ‘New JDA.’

Logility

Helping Retailers Master S&OP

Logility has been focusing on S&OP for retailers, applying their considerable expertise and key learnings gained from developing and implementing world class S&OP with many manufacturers. Most retailers have challenges simply integrating their merchandising with their assortment and allocation. S&OP is a good way to break down walls, bringing together the creative side with the operational side. There is a considerable change management component, since behaviors are so entrenched. Some of Logility’s customers are leading edge and further along in doing S&OP, while others are just getting started. Retailers can gain a lot of value by mastering S&OP, especially in today’s fast moving omni-channel world.

Virtual DCs

Many retailers have separate pools of inventory for store vs. online vs. wholesale. Another of Logility’s ‘silo-busting’ capabilities is support of ‘virtual distribution centers,’ where the retailer can use a common pool of inventory with defined portions of it reserved for each channel: ecommerce vs. stores vs. wholesale. If the inventory doesn’t sell in one channel, it can be freed up for use in another. This is especially important for seasonal products with short windows of opportunity. This provides the considerable advantages of pooled inventory, while ensuring the retailer won’t be ‘robbing Peter to pay Paul.’

Automating Allocation and Replenishment

Logility also announced their automated allocation and replenishment process. They gave the example of their customer, fast fashion retailer Groupe Dynamite, as a great case study of what’s possible. Montreal-based Groupe Dynamite is adding approximately 30 stores a year, mostly in Canada and the US, but also in the Middle East, which has completely different seasons and styles. Groupe Dynamite has been able to cope with that variety and speed by using Logility Voyager to fully automate about 80% of their store replenishment. The planner doesn’t have to engage with the system — the stores are just automatically replenished three times per week.

Groupe Dynamite wants to sell out within 6-8 weeks and avoid markdowns as much as possible. In the past, they would do a big upfront push of inventory to the stores and a smaller holdback for replenishment. They found that with this approach they were doing a lot of store-to-store transfers because the inventory was not where the demand was. The new system has allowed them to reduce the size of the initial allocations from about 70% of the inventory down to about 50%. Now, with the smaller initial allocation, the remaining 50% is held back and can be allocated based on actual demand. The automated replenishment algorithm will send only the exact sizes needed, rather than the old way of sending a standard set of sizes. This speed, coupled with size precision, has dramatically improved margin contribution and size sell through, while reducing store-level size variance by more than 30%.

Integrated Analytics

Logility is investing heavily in analytics and making them an integrated part of the suite, rather than requiring a separate license. They package up SCORE analytics and retail analytics which can be customized based on service goals, seasonality, number of seasons, and so forth. Their analytics support management by exception, as well as the virtual warehousing mentioned above.

NetSuite

Advanced Order Management

NetSuite has a deep heritage in ecommerce, increasingly integrated with the store retail systems they gained with the acquisition of Retail Anywhere three years ago. Another big step was taken last summer when NetSuite released Advanced Order Management (AOM). This provides distributed order management and rule-based order orchestration, allowing the retailer to decide how fulfillment decisions will be made. These can be based on a range of criteria, such as lowest shipping cost, fastest time, minimizing multiple shipments, as well as more complex algorithms considering staffing costs, capacity (number of associates available), and so forth.

The next release of AOM will include ‘store leveling,’ which optimizes and gets the right quantities of merchandise to the right store. AOM also supports ‘click and collect’ (aka BOPIS — Buy online, Pickup in Store) with task management and item picking navigation for store associates. Rules can be set up about what kinds of orders can be fulfilled by the store vs. the DC; for example, orders that require special packing and handling might only be fulfilled from the DC.

Clienteling Features

NetSuite added more clienteling features, such as reviews accessible to the associate; the ability to filter, search and see the live quantities in the store; and create orders and returns right from the same screen. The associate can see and ship to every address that the customer has entered in their online account and each item in the order can have a different destination. All of NetSuite’s sophisticated ecommerce capabilities are made available to the store associate.

Lucky Brands Gets Lucky Implementing NetSuite

NetSuite issued a press release on how they helped premium denim jeans designer/manufacturer Lucky Brand to quickly transition off their legacy on premise ERP system after Lucky was taken private by a private equity firm. Lucky Brand was able to quickly implement and move on to the foundational system and then build out more advanced capabilities. Lucky uses NetSuite’s OneWorld, leveraging its multi-currency and multi-tax capabilities.

Infor

The GT Nexus Impact

Infor had four really interesting sections at their NRF booth. One of them was GT Nexus, whom Infor acquired last summer. This was a really important and smart move by Infor. Firstly because of the rich set of capabilities that GT Nexus brings in global logistics, end-to-end procure-to-pay/order management, supply chain finance, supplier factory management, inventory management, transportation management, and supply chain intelligence. Secondly because of the tremendously valuable and well-integrated global community of trading partners and service providers (including ocean/ground/air/rail carriers, 3PLs, forwarders and brokers, customs agencies, banks and other supply chain finance providers, insurers, inspectors, and more) that are connected to GT Nexus’ network. Thirdly, and possibly most profoundly for Infor, GT Nexus brings a many-to-many network architecture on which Infor could potentially build other applications. (Note: Infor did not tell me this was their intent, but it seems to me like a great way to get more value from what they acquired). This could be very valuable to support networks and communities of enterprises and organizations, and complex ecosystems such as the DoD.

Dynamic Science Labs

Infor created their Dynamic Science Labs as a Center of Excellence for the ‘science of analytics’ to create IP that can be leveraged across Infor. The Labs work with different teams across Infor to develop new solutions or add to existing ones. They are very customer-driven and won’t start a project unless they have at least one and preferably two or three interested customers. They will then try out different things incrementally to see what works, starting with a simple proof-of-concept. If that proves out, they build a prototype and if all goes well, it finally gets productized. Retail is a somewhat new vertical for Infor, whose roots are deep in manufacturing. They see tremendous opportunity for analytics with all of the rich sources of customer data that are becoming available. The Labs looks like a great capability to help them get there.

The Dynamic Science Lab is also a strong complement to the GT Nexus acquisition. The GT Nexus network brings a rich and deep set of supplier and logistics execution data that could be captured and fed into the analytics machine. Analytics has the potential to enable more precise execution, making prescriptive recommendations based on the state of the entire network including all the parties and nodes, two or three levels deep. A retailer could for example postpone finalizing an order by a couple of days to capture better demand signals and then line up direct shipments to shorten delivery time. This is only one of many supply chain scenarios that could leverage the combination of GT Nexus’ data with the Lab’s analytic capabilities.

Hook & Loop

For over half a decade, Infor has been investing in unifying the user experience across their various product lines. In 2012, they took a bolder move by creating Hook and Loop as a sort of internal creative agency for Infor. Hook & Loop, who started with just five people in 2012, now has about 130 writers (including Pulitzer prize winners), designers, developers, and filmmakers whose sole mission is to improve Infor’s user experience. They will sit with a customer, ask what information they need on a screen, walk through the uses, take out all the clutter, and bring all the points into one simplified and beautiful place. It is an amazing resource for Infor.

These companies represent only four out of the more than 500 companies exhibiting at the NRF this year. However, I believe they embody well the kinds of innovative new capabilities and approaches we are seeing from retail solution providers, indicators of a striking level of dynamism in retail at this juncture.

__________________________________________

¹ Forecasting the impact of returns on saleable inventory levels is further complicated by needing to understand which % of returns are resalable at full price, which portion may be sold at a discount, and which portion will need to be disposed of. — Return to article text above
² Examples of constraining rules for loading a trailer would be things like last-in-first-out sequencing, larger items must go underneath smaller items, don’t overload an axel, etc. — Return to article text above
³ For more on probability forecasting, see Demanding Times: Next Generation Demand Management Practices — Return to article text above

To view other articles from this issue of the brief, click here.