Risk appraisal and holistic design
An RFID project is similar to any other large IT project and will have the usual risks these attract. However, an RFID project may be sponsored by the business without reference to the IT group, in which case the IT group may not be sufficiently involved, so the experience it already has may not be taken advantage of. As McIntyre points out, you must involve all stakeholders in the project, including the IT group, which will have to implement and maintain much of the supporting technology.
An RFID project will likely be a big project (and big projects are always risky) with high visibility at board level, which increases the risk even further. It will likely cross organisational boundaries, another risk factor, and have both operational and infrastructure implications.
Integration will be a major risk area - the metadata and semantics of the RFID data must be consistent with existing operational data, or you will store up future maintenance problems even if the initial RFID project "works" in the short term. Once you collect data for inventory, management will come up with business intelligence applications for it and this can be another source of confusion. McIntyre says managing integration was a major factor in his system's success: "Our decision from the beginning was not to be a stand alone application," he says. "And that's why we have such a good relationship with TIBCO" (which provides the fundamental integration underlying his different systems).
Operational risk must always be considered - potentially, you might be collecting a lot of RFID data, so the capacity of the existing network may be an issue. And, of course, database volumes may be impacted. McIntyre says his active tag supplier provides a software package which consolidates tag blinks into meaningful transactions: "We're basically keeping all the non-pertinent information off the network and keeping it from being clogged up", he says.
Then there are people risks. RFID can help implement a "big brother" culture, in which people feel under observation all the time - and dysfunctional sabotage is always a possibility. RFID tracking can promote good things, such as equalising of workloads and prevention of petty theft too, but you must get everyone's buy-in and discuss any issues with unions etc. Also, remember to make sure training is available to those involved with any new technology.
There are security risks too, as RFID usually tracks RFID tags, not objects themselves. Is an empty DVD recorder box with an RFID tag indistinguishable from a DVD recorder? Can I invoice for goods when all I deliver is RFID-tagged boxes? RFID tagging can help control pilferage, but it can also enable it, potentially on a larger scale.
All these risks can be addressed, as long as they are identified in time. RFID is a new technology; often, identifying technology risk, and the "good practice" for addressing it, is part of company culture - but company culture may not have caught up with new technology. Once again, don't overlook training (and the availability of external experienced mentors) as a tactic for achieving cultural change.
Which brings us to the question of holistic design. This is where you address risk factors and design a business process that is not just a technology solution, but includes operational and architectural aspects. Unless you analyse and design the manual processes around the RFID technology and processing, how can you validate the interfaces to the automated processes and ensure the risks have been adequately controlled? After all, risk management often comes down to proactive exception reporting, resulting in more or less ad-hoc human decisions; ideally, taken before the symptoms have become severe enough to cause real problems.
Businesses are being persuaded into seeing RFID as a panacea of inventory management and other issues. Whatever the truth of this, companies such as Ford and TNT are running successful RFID projects already, and many developers may be asked to take part in similar initiatives.
There are specific RFID issues - radio waves behave in unexpected ways in the presence of large amounts of metal or water, and you need to decide between passive tags (cheap) and active tags (more flexible), with batteries and a modicum of intelligence. And even cheap RFID tags are orders of magnitude more expensive than simple barcode labels.
Don't overlook the fact that a large RFID project is also a large database technology project and will have all the issues usually associated with an IT project of this size, as well as any specific to RFID. Don't let RFID hype blind you to the issues usually associated with implementing potentially disruptive technology.®