BrainTrust Query: What are we going to do with all that RFID data?
By Bill Bittner, President, BWH Consulting
(www.bwhconsulting.com)
People are having lots of discussions about RFID technologies and the challenges associated with capturing the serial numbers of containers encoded in their RFID tags. Most of the talk has been about getting the hardware to accurately capture every single unit.
Read rates for the RFID tags vary widely depending on the combination of product characteristics, tag manufacturer, reader manufacturer, and environment. Liquid and metal containers introduce transmission problems; tags of different shapes respond differently; readers with different propagation patterns achieve different response rates; and RF interference from equipment and nearby readers distort results.
But in addition to the operational challenges, people are asking a more fundamental question: “How do we use all this data?” Just capturing “events” – often defined as units moving across designated boundaries, such as from the backroom to the selling floor – will result in huge amounts of data. What is the purpose of capturing all this and how will it help run our businesses?
The obvious answers are track and trace. Pretty much everyone understands this to mean determining where something has gone or where it has been.
Loss prevention is another factor, as RFID can be used to detect “lost” containers or identify them as “stolen property” and make them valueless.
But beyond this, how are we going to integrate the new information that comes from “the container’s talking” into our existing stock ledger and locator applications?
Moderator’s Questions: Will RFID data engender what amounts to a second set of books?
Sorry for all the questions, but the more I think about this subject, the more arise:
- Does integrating data mean stock ledgers will no longer track inventory based on accounting standards such as FIFO or LIFO and now use specific unit
accounting? - Does it mean that physical inventories and stock locator systems will have to be modified to balance with the inventory generated from tags?
- How will this differ if read rates are less than 100 percent?
- If you have an inventory that, by definition, is not 100 percent accurate, how do you use the information it provides?
- Will there come a day when the RFID inventory will be considered more accurate than all the bookkeeping inventories and become the driving force behind
inventory control?
Answers, or even idle speculation, much appreciated. –
Bill Bittner – Moderator
When point of sale systems first became popular, retailers had internal debates about accepting the POS data versus continuing to rely upon traditional financial data. Numerous retailers tracked the data differences between the “financial system” and the “merchandise system.” For example, the financial system might deduct shrink from inventory as a percentage of daily sales, yet no merchandise system deducts shrink from inventory that way. Financial systems emphasize dollar amounts, and merchandise systems emphasize item counts. If the RFID read rate is accurate enough, it could be used for financial purposes in some applications, assuming the RFID tags are secure and viruses are prevented. Of course, that last sentence has 4 major qualifications.
If the question implies that there is a concern over the amount of data amassed as a result of RFID at the item level, fear not. First, many retailers are already keeping POS transaction-level data for at least a year, just in case. All we would be doing to that data as a result of RFID, is to append the RFID identifier of the item involved in that transaction; no great increase in volume.
Other transactions resulting from RFID item-level tagging would be non-POS inventory movement transactions. These transactions, except for chain-of-custody, or pedigree type transactions — such as those retained in the pharmaceutical industry — can be discarded after a user-determined period of time. There is no burning need to keep historical data on inventory movement past a minimal amount of time. Retailers would keep the data, but only if they chose to keep the data.
I don’t think there is anything about which to worry. Besides, if worse comes to worse, data storage space is continually getting cheaper, smaller, faster and better…
You know, I don’t think of it that way. RFID data has to go through several layers of business rules before it becomes “information” – like, inventory level or inventory status. Will RFID data replace FIFO? I don’t think so. It should, theoretically, make it easier to identify which units were actually first, though.
Read accuracy aside (which is actually dismissing a lot, I know), it’s those business rules that are challenging. People talk a lot about all the data the RFID generates, but if we architect it properly, most of that data will stay at the edge and have a very short lifespan. Business rules, agents, etc. will monitor that data to either translate it into business events, or elevate the urgency of a condition. That doesn’t replace existing stock ledgers, it becomes another input into them. But our enterprise application software has to make some big changes to accommodate, and service oriented architectures have to become a lot more mature and pervasive before this kind of sophisticated relationship between the edge and the enterprise can happen.
Read accuracy aside, all the technology pieces are there – they’re just not strung together well yet, and some of the pieces are not very mature. This is RFID the evolution, not the revolution.
As a researcher, I’m looking past this second set of books, which will be necessary. I’m thinking of the day when we can track the units back to the consumer’s homes to determine length of usage. This way we could follow-up with product attribute and other critical information. And don’t whine about privacy issues. We all know that doesn’t exist in our brave new world anymore.
Here are some thoughts to build on Bill’s observations and questions. It seems to me that in the emerging RFID world, inventory will be defined in terms of flows versus stock, and accounting quality POS data will be the Holy Grail. LIFO and FIFO will become less relevant terms as we move to true average cost.
In this Brave New World, the focus will be on driving the velocity of inventory turns, and as we do this, supply and demand chains will be in perfect integration. It’s an exciting vision but one for which Bill has raised some important questions, including implicitly, how do we verify this is the right thing to do from a financial/business point of view?
At this point, we are preoccupied with the effectiveness of technology which seems to me to have the cart a little ahead of the horse when we don’t know exactly why we’re moving to RFID and how to judge when we get there. What might help is a mega-industry look at the economic benefits such as what was done by Kurt Salmon for ECR or McKinsey for the UPC scanning code.
There is an analogy that can be drawn between webpages and RFID data.
To get to a webpage, you turn everything into packets, you transact around packets, you filter packets, you log packets. Those logs are so full of data that logging is the best you can do. You can’t be event driven at the level of the packet. But, you can audit logs when problems arise, as in when you get hacked, and you can discover how the problems happened.
Webpages used to be files that you served. There might be a truckload of files being served, but they were flat files. Then, webpages started being generated from databases, so the stuff being served was relational. Then, webpages started to do personalization, but in batches, now in real-time as the pages were requested. The latter level of personalization means that the content is in a data warehouse. A data warehouse correlates well to a warehouse or the grocery shelves. But, in the end, the webpage is delivered.
Having captured all that data means that problems can be uncovered and processes changed to prevent reoccurrences of the problem in the future. These changes can be evangelized and adopted across a company or industry. Improvements can happen fast.
Having all that data doesn’t have to affect the accounting system at all. The accounting system is an abstraction anyway, like the webpage. Some researchers can take a look at whether reality or the accounting system answers a particular question better or not. Where the answers are better, opportunities for change exist. Adoption will take a while. But, without all that data, you can’t even begin to ask the question, “What should we change?” So having the data is a strategic option. An option that if not taken by your company, it will be taken by some other company. So the data will grant competitive advantage to first movers, and the data will give rise to vendors that can exploit that data.
Drowning in all that data might mean just logging it and knowing where to look. It might not need to end up in a database or a data warehouse. Database administrators still use blobs to store untagged, undifferentiated data where the only context is that of the blob, as in everything in here is about the strawberry jam that touched the logistics system last Tuesday. Getting into useable form is an extra step. But, the capacity costs would be different if that data was easily meaningful.
That meaning is hard to reach is a different issue from not being able to reach that meaning at all. When system boundaries move, when a system produces new never had before data, consumers spring up, as do opportunities.
There are some serious issues which still need to be addressed with RFID and personal privacy and security. Everyone seems to be focused on how we should be amassing, storing and using this purchasing and marketing behavior, when we have yet to effectively address the problems with gaining consent to identify, store and use the data which RFID is empowering the retailer with. This is no longer just scanner data, since products can be tracked anywhere in the store area and purchasing behavior and identification can be done even if the customer decides not to purchase the item. Without purchase, there is no agreement from the customer to accept the store’s terms and conditions, and all of the aggregate data cannot be used without this. We are jumping the gun here, since personal privacy, personal security, and the legal implications which RFID raises really need to be addressed first, prior to implementing it in the retail environment.