In A First, Google Does Real-time Joint Retail Trials

In the 15 or so years that we've had E-Commerce, the industry has seen quite a few improvements, but nothing that radically changed the way people shopped or retailers sold. Local inventory search, which today is not even in its infancy (not really even embryonic; it's more like a zygote), is likely to be the first truly dramatic shift.

Consider a quest for a stuffed animal shaped like a buffalo or a very specific part for a long-discontinued product. Assume that the consumer needs it today, so overnight shipping isn't an option, or perhaps the consumer needs to examine it prior to purchase. The capability for a search engine to have a current and comprehensive list of products (SKU and sub-SKU) for every retailer in a given geography has huge potential to reshape E-Commerce. If nothing else, it could be some very good news for the smallest of retailers today, which hardly ever see any search engine revenue.

Last Thursday (March 11), Google made a major—albeit extremely preliminary—move into local inventory search through a deal with a handful of major chains: Best Buy, Sears, Williams Sonoma, Pottery Barn and the Vitamin Shoppe. The beta test has its accuracy issues, missing many products that are definitely in stock at those chains.

It also gets too specific too quickly. For example, we ran a test seeking a good HDTV. It found quite a few specific models and configurations. But it forced us to select a specific model, brand and configuration rather than allowing us to say, in effect, "Find me the nearest store that has any." The ones it found were all much farther away than the Best Buy around the corner (and, yes, we checked. That local Best Buy had several HDTVs in stock).

But those bugs will get worked out. Indeed, from Google's perspective, that's the whole point. Instead of working out this process internally and then bringing in retailers—or the reverse, with retailers figuring it out first—Google's objective is to run the earliest stage tests with various chains "so we're both learning how to do this in parallel," said Paul Lee, Google Product Search's business product manager.

"There is some conception, historically, that there is a linear model where the retailer has to get everything done right. Then, eventually, the search engines can piggyback," Lee said, suggesting that the simultaneous approach should sharply accelerate learning. First, the sequential approach would theoretically take much longer, as the retailer would have to figure it out and then explain the discovered process to the search engine. But realistically, the retailer will likely not have considered some things the search engine lives and breathes, thereby forcing the process to be re-created. If the search engine went it alone, a similar lack of experience with retail logistics would cause the same problem. By doing it together, the learning curve should be much faster and the robustness of the final product stronger.

The Google test is being done solely in mobile, leaving the desktop for much later, if at all. It's not even offered for all of Google's mobile platforms: Only iPhone, Palm WebOS or any Android-powered device. Sorry Blackberry users!

But the challenge for the retailer to have a comprehensive and current list of every product down to the sub-SKU level—not merely that it has 52 stuffed animals in stock from the Acme Toy Company (if it's good enough for Wile E. Coyote, it's good enough for us), but that it has three shaped like buffalos—is huge. The potential rewards are equally big, though, with each and every product capable of acting as its own tiny little search engine ad for the store.Large chains typically like to stock the most popular items, while smaller chains—and certainly family-owned single-location stores—gravitate toward niche or unusual items. But think about it. When some small antique shop gets an unusual item, it's generally hoping to attract a consumer who is browsing rather than a browser that is consuming. The alternative is to try and match that unusual product with a customer who really needs it.

"Well, somebody will want it," says the merchandise buyer. True, perhaps, but how to connect the two? What are the odds that one of those people who needs that item will happen to drive by that store and decide to walk inside. Even if that happens, what are the odds that the item will be displayed right then or that the customer would even notice it? Local inventory search, once enough products are in its clutches, has massive potential.

Before that can happen, there are quite a few cost issues that have to be dealt with. Even though some services are offering to accept the feeds from retailers for free, the cost of getting their product databases in shape to offer this kind of feed is overwhelming. "What does it cost a small independent retailer to take advantage of that service?" asked Dan Butler, the VP/Retail Operations for the National Retail Federation. "Many of them struggle to just keep their sites up and running."

Technically, a retailer wouldn't have to have a Web site to use such a service. But Butler found that quite unlikely. "If they have don't a site, they wouldn't have the mindset" to organize, catalogue and keep all of their product info current.

For that and 50 other reasons, there's little debate that the smaller merchants will be the last to get involved in local inventory search. "For one thing, there's a certain amount of activity that is happening with the larger retailers, a more robust API system that allows them to more easily deliver that kind of information," said Ron Levi, VP of product at, which specializes in Web product search.

Levi's own engine is also just toying with this kind of information. "It's limited. It's not been extended to all inventory, and it's not perfectly accurate. But it's a step in the right direction," Levi said.

Even if you were to combine all of the products searched by Google and other search engines, it barely would cover one-thousandth of one percent of the products in the inventory of the major retailers plus millions of not-so-major retailers. How much would a consumer be willing to pay if your software discovered the specific product they sought was a mere 10 minutes away, if they only knew where to look?

To make this capability viable and truly useful to residents of one region (defined as, perhaps, an area within a 45-minute to a one-hour drive), the engines would have to realize at least a one- and perhaps a two-percent complete database. More is better. But once the one-percent barrier is broken, the chance of actually finding the desired item becomes relatively likely.

Levi suggests that some products will be very difficult to catalogue without a lot of manual typing--items such as handmade goods, custom clothing, bait and tackle, etc. "Most of those are not going to have a UPC. That makes it a much greater challenge to normalize them," he said.