Emulate the Internet

The RoI figures on RFID deployments are far from encouraging. Most analysts suggest a focused approach to better RoI, but the catch is in adopting the right approach to infrastructure building and then deploying the right applications, not vice versa.


Wal*Mart put all its weight behind RFID and in January 2005 mandated its suppliers to adopt RFID for pallets shipped in to its retail warehouses. That’s old hat now. More than a year and a half into the deployment, the naysayers are already writing the deployment off. A research by a leading analyst company said that Wal*Mart and other retailers that have pushed RFID might be bitterly disappointed by the results. The company researchers modeled a company-wide roll-out of RFID, used to track cases and pallets of goods throughout the supply chain, at a hypothetical, $5 billion retailer. What they found: In the best-case scenario, RFID investments pay off in nine to 10 years. That's a long time to wait.

The obvious question then is, that is RFID a waste of money? Some analysts suggest that that RFID's deployment won't be as broad as anticipated. It will be highly focused. RFID won't make it into every store. It won't make sense in every store. In other words the research company suggests that it only makes financial sense when used for tracking certain individual items, such as DVDs. If retailers expand their deployment to track everything, it will only elongate the time to positive RoI.

Common Wisdom
Going by the numbers we have today, it would not be too hard to believe what the reports says. But a large number of analysts and informed people, ten years ago, questioned the RoI of a corporate Internet mission because of the limited functionality and associated administrative sweat. And therefore the mass market for residential broadband was inconceivable. RFID projects hold great promise for the future, but face the same concerns that the corporate Internet strategy did over a decade ago. The point that I am trying to make is that CIOs must look at the evolution and the growth of the Internet to take a few lessons. These lessons can then be applied to RFID to foretell its evolutionary path.

Amongst the first candidates to be RFIDed were those applications which were developed around manual data entry or non-serialized bar code tracking. These tasks, if I may call them that, could be upgraded and enhanced with low cost, high quality, automatically captured RFID data. In other words RFID holds immediate interest in areas like store operations, inventory management, electronic proof-of-delivery, and pharmaceutical tracking. However a slew of existing applications ranging from ERP to merchandizing applications are already used to keep track of products or goods. Does this not sound similar to the times when there were no products designed to deliver reliable Internet service and manage bandwidth among multiple business users. And thus out of necessity, many small and medium sized businesses had to work with connections multiplexed though multiple low speed dial-up modems. Similarly the real potential for RFID RoI lies with applications that will be able to leverage the automated tracking of objects at low cost, high volume, and in real time without human intervention.

And its these applications that will make the RFID RoI equation easier to comprehend and faster to turn to black. To be visionary with the Internet 10 years ago was to see the potential for it to spread from universities to enterprise email to today’s Web access via cell phone or e-commerce via wireless at the corner Barista coffee shop. The changes that RFID brings will be as large and life-changing as our mobile phone and Internet technologies have become. RFID networks will evolve from an application layer to become an enabling infrastructure and the ROI will come from both enhancing existing applications and building many new ones that harness this revolutionary data source. Some day RFID tagging may be as ubiquitous as Internet access is today.

The next obvious question is how is all of this possible?

Enabling Infrastructure
The most cost-effective way to deploy RFID on a large scale will be to follow examples set by successful network-building models. Enterprise networks combining ethernet and TCP/IP can easily add more users and adopt new functionality because their layered architecture and support for standards helps to isolate the complexities of implementation, promoting scalability and growth. A layered architecture is fundamental to every successful modern networking technology. This is exactly what the early RFID solutions have lacked, focusing on stove-pipe connectivity instead of network topology. Instead of building a separate foundation for RFID, the RFID network must be welded into the enterprise infrastructure and by standards that are already in place. So the idea is not to look at RFID not as an island of experimentation but as a global plug-and-play solution that can be rolled out enterprise-wide. And this is precisely the anti thesis of products and applications one sees in the market place today. Analyst firms also suggest that the driving mantra for RFID applications must be focus.

Key Benefits
This approach establishes a few key benefits that are otherwise overlooked in most RFID deployments, pilot or otherwise. One, it follows a network centric approach wherein RFID readers belong to the enterprise network. These RFID devices are managed as part of the enterprise, not as separate experimental nodes. The point here is to leverage the existing infrastructure that’s designed for scalability and manageability. Two, this approach to RFID networks builds on standards to promote interoperability and commoditization. This may not be as important today but in times to come, most RFID users will grapple with interoperability issues as a large number of RFID solutions still depend on proprietary tag protocols, reader protocols, and custom-built middleware.

RFID has the potential to emerge as the very fabric on which the new rules of business will be written, the distinguishing factor will be to get the script write.

This is my fourth article on RFID in Dataquest