When Wal-Mart's much-watched, ambitious RFID pilot project kicks off next month in Dallas metro-area stores, don't be surprised when it fails. Save your shock for the slim possibility that it will succeed.

I say this because Wal-Mart's stated goal for its pilot is "100 percent readability of pallet tags through dock doors and 100 percent readability of case tags on distribution center conveyor belts."

Not a chance.

The passive Class 1 RFID tags and the readers that use radio frequency signals to get information from the labels aren't capable of achieving 100 percent success in noisy warehouses loaded with signal-distorting metal objects. And the relatively weak signals from the Class 1 readers will have loads of trouble figuring out the contents of cases of, say, Clairol shampoo or Palmolive dish soap, since those signals don't propagate well through liquid.

I'm not the only RFID doom-and-gloomster. Gartner issued a report earlier this year warning its subscribers to "prepare for disillusionment with RFID." Analyst Jeff Woods concluded that, for the moment, "radio frequency identification cannot live up to the promises that have been made for the technology."

The statistics tell the story
But you already know how disappointing RFID has been. A mere 8 percent of 50 companies tpolled by ABI Research claimed to be very or somewhat satisfied with their RFID programs. Nearly 20 percent said they were somewhat or completely dissatisfied.

Nonetheless, half of 30 manufacturers surveyed by Accenture said that they "expect high returns on RFID investments." You know they're in for a rude awakening.

You'd think the big problems dissatisfied users face would be in back-end integration. Not so. It's the performance of the tags and readers that irk them. That and their cost.

So, while I'm grateful that Wal-Mart has an RFID FAQ page on its Web site, it will be a long, long time before we need to know that people can either keep or discard the harmless RFID tags they find on goods purchased at a Sam's Club or Wal-Mart Supercenter.

Despite the hoopla and the apparent need, RFID's best chance for success won't be in retail. I'm betting it will be in manufacturing, particularly among makers of big-ticket items. That's because the RFID labels and readers to be used there will be either Class 3 or 4 technology. Those RFID tags include a battery with an estimated three- to five-year life span and readers that use a strong enough signal to work well in a real-world warehouse and everyday manufacturing floor. They'll even figure out what's inside a case of shampoo or a pallet with kegs of beer.

Adding memory to the label
One company working toward that goal is Intelliflex in San Jose. Intelliflex hasn't announced its dexistence to the world yet. Its venture capital backers consider it to be in "stealth mode," with its dproducts still in early beta. But Ashish Asthana, vice president of marketing and product strategy, took a moment to discuss the advantages of his firm's technology.

With an integrated battery on an RFID tag, he says, you get to add memory to the label. With memory, he suggests, application writers can expand beyond the simpler expectations retailers have about using RFID to improve inventory control.

"In manufacturing, you can do things with configuration control throughout your supply chain," says Asthana. You can even make an RFID tag do things like report the temperature of an item, he adds.

David Ezequelle, venture partner at Alloy Ventures, which is one of Intelliflex's investors, adds that in the future, battery- and memory-assisted RFID tags and readers might be used not just to identify an object's contents, but to locate it, too. For example, he says, a hospital could tag its costly instruments and use a triangulation of readers to find them when they go missing, as happens so often in busy, sprawling hospitals where everyone is sharing pricey gear.

Sure, Class 3 and 4 RFID products are much more expensive; tags are US$25 to $30 versus 50 cents or so for the Class 1 labels that Wal-Mart's partners are applying today. But the objects tagged with these expensive labels have a far, far greater value than consumer products found on a retailer's shelves.

Let's see the top end
So I say, let's forget Wal-Mart for a while. We can wish the retailer and its suppliers well in their efforts to solve a tough problem at the lowest possible cost. But I doubt that many companies will be able to apply many lessons from their experiences, at least for a while.

What we need are examples from, say, automakers that can use Class 3 and 4 tags to alert inventory systems that the expensive products from their Tier 1 suppliers are waiting to be offloaded from trucks; or from the transportation industry, where railcar owners can keep track of rolling stock and their contents and schedules. The best RFID work will be done at the high end.

"It's all about ROI," Ezequelle says. "No one's going to invest in it without money coming out the other end."