RFID Special Report, Part II: 20/20 In 2020 — Inventory Accuracy Takes Center Stage

“Learn to crawl and walk before you try to run” is something we frequently advise our retail clients.   That advice is particularly helpful when we are working with them on roadmaps for harnessing the RFID data produced in their stores and supply chains.  This report will outline some smart tactics at each stage of the progression from decision making processes that rely heavily on human input to those that capitalize on advances in automation and artificial intelligence.

Information Overload

It has long been established that RFID provides retailers with significant operational benefits.  Our retail clients are moving beyond the core foundational capabilities and are eager to fully capitalize on the new forms of data their RFID systems are generating.  This is key to extracting every last dollar of ROI.  Whereas their initial focus may have been on reactive measures, like tackling out-of-stocks or allocating into Shrink, our clients are now focused on using data proactively to change what they sell and how they sell it.

We typically devote attention at the outset to providing our clients’ national Store Operations teams (and the individual stores themselves) with good KPIs and dashboards to help them remotely monitor the condition and performance of each store, district and region.  Are there executional issues?  Are they elevating product availability?  Which stores or which employees might need re-training?  Does the communication program for the entire fleet of stores need to be tweaked?

Inventory Visibility is another immediate priority.  This includes decisions on how to use RFID Operational Ledgers, compared to traditional Financial Ledgers.  From here, the data can be expanded to support additional functions within the business.  Finally, it is important to create early alignment between a retailer’s RFID initiative and its Automation, Machine Learning, and AI initiatives.  

Determining  precisely which data to incorporate into each dashboard is a typical initial step.  Below are some examples of reporting and metrics that we have recommended to our clients.  This is not an exhaustive list, but rather a sample:

Store Operations

As retailers transition from pilots to full rollouts, a base for success is maintaining visibility into every store’s operational execution. We often tell our clients, that the “hard part” of any program success is the “soft stuff.” Motivating store personnel with relevant data to build actionable goals supports helps them adjust to new methods and practices.  Almost all employees within the store will see changes in their day-to-day activities.  Change management is an ongoing process. 

Examples of relevant metrics are:  Compliance and Quality of Cycle Counts; On Floor Availability (display compliance, size assortment, or the availability of the most current seasonal Floor Set); Labor Efficiencies (e.g. Receiving or Processing, Cycle Counting, Omni-Channel Decline Rates).  Program performance rises when this type of dashboarding is used to celebrate and reward success, at individual stores and also at the district and regional levels.

Visual Merchandising

Operational metrics can also expand into Visual Merchandising functions in the store.  This is especially exciting as historically Visual Merchandising functions in the organization are just that ---- they are visual.  Those groups currently have little data to provide long term indicators regarding how specific stores are (and should be) setting up their shop floors.  Examples here include: Capacity metrics; Assortment Quantities versus Buy Plans or Minimums; and Display Compliance (including choices of which items are displayed on Mannequins).

Asset Protection

Ideally, a retailer’s AP team will start using RFID data from stores to gain insights into in-store losses. This data can be used to dynamically adjust each store’s targeted asset protection strategy within the year, rather than wait for the next annual physical count of inventory.

Some retailers decide to conduct physical counts in certain stores within their fleets more than once per year. RFID data helps retailers reduce and ultimately eliminate these intra-year physical counts.  RFID data also provides valuable evidence and clues that allows them to fight theft throughout the year more successfully.

Examples of information worth tracking include: Unexpected Losses; Trend Level Loss; Highest Loss Products.  A good metric for the supply chain is In-Transit Loss.

Merchandise Planning & Allocation

Applying RFID data to optimize Planning and Allocation decisions is typically a core component of a retailer’s business case.  Allocation accuracy, reducing store out-of-stocks, transitioning from a Push to a Pull allocation model, and reducing overall chainwide inventory levels.  These are some of the key ways retailers are attempting to drive value. 

Some examples here include: Distortion between RFID and the Financial Ledger (Overstocks, Understocks, “Zero on Hands”); Out-Of-Stock Reporting; Inventory Distortions. 

On floor availability has a bearing on markdown decisions and other methods of inventory consolidation.  Retailers can crack the code using RFID data.  Planning teams will have interest in the same Visual Merchandising examples listed above, gaining better insights into store fixture and space capacities.

Brains vs. Brawn

Over the past few years considerable attention has been paid to Automation, Artificial Intelligence and Machine Learning.  The excitement is understandable.  Just like RFID, these technologies are changing the way retailers conduct their businesses.  But the interrelationship between these four technologies is not well understood.  It deserves a closer look.

Retailers who use RFID typically rely on reporting and dashboarding tools to put the system to greatest use.  These tools streamline the work of employees in stores, in the field, and at headquarters too.  Automation, Machine Learning and AI will allow retailers to achieve this with even less human involvement.  And those decisions will be smarter decisions.

There is a fair bit of confusion when it comes to the discussion of Automation, Machine Learning and AI.  For the purpose of our conversation, we will define Automation as robotic process automation (i.e. Bots).  We will define Machine Learning as prescriptive analytics and decision engines.  And we will define AI as deductive analytics.

Bots are software robots that mimic human actions --- essentially the brawn.  By contrast, AI is concerned with the simulation of human intelligence by machines ---- the brain. AI and Machine Learning are often used interchangeably, but they aren’t the same thing.

When machines carry out tasks based on algorithms in an “intelligent” manner, that is AI.  Machine Learning is the subset of AI that focuses on the ability of machines to receive a set of data and learn for themselves, changing algorithms as they learn more about the data the machines are processing.

In the next section we will outline how RFID data can be used to drive new efficiencies and value within retail using Automation, Machine Learning and AI. 

Automation: 30 Hands Are Better Than One

The ability of retailers to process and utilize the large swaths of rich data their RFID programs create is today constrained.  After all, a retailer can’t afford to hire an army of employees, in its stores and at headquarters, to evaluate and transact on this data.  The answer is simple:  Bots. 

We have helped several retail clients identify ways to deploy Bots effectively.  Below is one simple example.

Prioritized Receiving

Optimizing the scheduling of labor to receive shipments effectively can be tricky.  And once the receiving process starts, completing it on time can be challenging too.  There is usually only so much processing that can be done before a store opens.  And on when shipments are unusually large, many items needed on the floor remain unprocessed in cartons. This is due to a combination of factors, including: limited staffing, limited visibility, and limited time.

But what if it was possible to do an analysis of what the most important items in an incoming shipment included, at the box level, and make a processing recommendation to the receiving staff?  With the availability of RFID data and Bot automation, this is now possible.

The type of Bots we have recommended to our clients leverage various data to make these recommendations, including, but not limited to:

1.      Which products arriving in a shipment are required on the sales floor due to the product not being represented on the floor (as of that morning)?  This can be done at the Style/Color/Size level.

2.      Which products are arriving in that shipment that are deemed to be Hot Sellers that merit priority processing ahead of other product?

3.      Which products are arriving in that shipment that are part of a new Floor Set (and are therefore required on the sales floor immediately)?

4.      Which boxes contain priority items? 

5.      Which boxes contain the highest number of priority items?

A single Bot is capable of accessing multiple systems, pulling the relevant information, cross-referencing, and performing this analysis, in near time, every morning.  This can be done for each store in a retailer’s store fleet.  Producing a prioritized carton processing list ensures that even if the Receiving Team can’t process every box in a shipment by the usual deadline, at least the most important products in that shipment will have made it to the sales floor before the first customer walks through the front door.

Artificial intelligence

AI investment is clearly spiking.  Juniper Research predicts retailers will spend $7.3 billion on AI by 2022, up from approximately $2 billion spent in 2018.  According to Gartner, 85 percent of customer interactions will be managed by AI by 2020. 

RFID will help retailers get more out of these investments.  Algorithms powered by AI can seamlessly recommend options that would typically be much too laborious (and costly) to identify manually. AI adapts to its environment and gets better the more you use it.  It needs data. The more data the better. And RFID is excellent at generating data!

RFID solutions that rely on the use of handheld devices for data capture can generate location-specific (i.e. zonal) information as often as once a day. Real-time “Hands Free” RFID solutions go beyond simply providing zonal information.  They can also identify, in real-time, an item’s whereabouts in the store, often as accurately as “plus or minus” one foot. This opens the door to product adjacency analysis, path-to-purchase analytics, and many other “brick-to-click” analytics.

There are countless ways that RFID data can empower AI algorithms.  Some examples include:

·        Optimizing Fulfillment Of Omni Orders (especially when retailers have established a Fulfill From Store capability) 

·        Optimizing a store’s “four wall” inventory based on floor capacity, sales, velocity, sales forecasts, trends

·        Avoiding out-of-stocks and unnecessary markdowns — e.g. by featuring and promoting underselling products held in reserve that otherwise would later have to be discounted

·        Making product placement and visual recommendations based on analysis of how similar products have performed in the past on the floor or on mannequins or in display windows

·        Optimizing replenishment priority based on product profitability and “must-win” items

·        Optimization of shopping personalization by unifying online data and in-store data to create a single view of customer shopping behaviors and preferences

·        Product design feedback and optimization based on Fitting Room analytics (including Try/Buy ratios) and product return rates

We are confident that retailers who invest wisely and focus intently on these opportunities will be well rewarded.

John-Pierre Kamel is a recognized leader in the Canadian and U.S. RFID communities. He has nearly 20 years of enterprise strategy and solutions integration experience. Prior to joining RFID Sherpas in 2010 he led the RFID practices of VeriSign and Bell Canada. Earlier in his career, John-Pierre was the Canadian Lead for the Mobility Solutions Practice of Capgemini.

Marshall Kay has advised the presidents of several leading apparel, footwear and consumer product companies. Before founding RFID Sherpas in 2007, Marshall led the North American RFID practice of Kurt Salmon Associates. He began his management consulting career at A.T. Kearney. Marshall has authored numerous reports on RFID and collaborative commerce.

Kristen Munroe Kristen’s deep understanding of retail processes and change management is informed by her 18 years of retail experience, most recently with Ralph Lauren. For eight years she has had national Retail Operations Director responsibilities. She has also led stores at the district level and served as National Director of Training & Recruitment. Kristen’s RFID experience dates back to 2013.

RFID Sherpas is a retail consulting practice that assists global brands and retailers at all stages of their RFID journeys. The firm is vendor agnostic and does not re-sell RFID hardware, software or tags. The firm’s services include: executive workshops, business case evaluation, process optimization, solution architecture, vendor selection, negotiation support, project management, analytics and reporting.