In April, 2000, Vicinity (where I was employed) and Northern Light Search released Geosearch, the first large scale deployment of a geo-enabled search engine. (The NL corpus was about 300 million pages.) The basic idea is that Geosearch would transform addresses and some phone numbers on web pages into a form that could be efficiently represented in search engine indexes. This allowed NL to do true proximity searching, retrieving relevant pages with addresses/phone numbers within a geographic region. Geosearch was on the air for two years. Google Local appeared in 2003, and is similar to Geosearch, except that Google Local adds an additional step, trying to connect the geographic information it finds on web pages with Internet Yellow Pages (IYP) data. Microsoft bought Vicinity in 2002.

The fundamental observation made possible by Geosearch is that local content permeates the web: 20 percent of the pages we examined had either a USA or Canadian address, or a geocodable North American phone number. It is the web itself that is the obvious source for aggregating local content. Not the print YP, not the IYP – a repurposed direct mail mailing list compiled from printed phone directories – not even content from vertical content aggregators. It is the democratically contributed, distributed content that IS the web that will be the richest source of local content in the future. It is the millions of web pages that contain addresses and the associated local content that these addresses usually imply.

Geosearch also allowed us to discern that local content on the Internet is about far more than commerce. What the Internet enables that other distribution mediums do not is timely dissemination of niche information, what I call idiosyncratic local content. The current focus of local search on YP-like data discounts the value of this potentially most attractive piece of local search. My hunch is that if taken together, this idiosyncratic content, each with small individual constituencies, comprises the majority of what people would use local search for. Support groups, local artists, activities that are not yet or not quite mainstream, hobbyist get-togethers. Storefronts don’t change much, social gatherings do.

The “Yellow Pages” is a trusted brand name, and an almost idiomatic expression for any source of easily accessible information on storefront businesses. There is a natural tendency to conflate Local Search on the Internet, still in its nascent stages, with the Yellow Pages, which recently celebrated its 117th anniversary. I think this will change. The proper domain of local search is the geographically-oriented activities of daily life. Local search should tell me what IS in my neighborhood, and what IS HAPPENING in my neighborhood.

Our experience with Geosearch convinces me that the end result for local search will not be that one supercharged proprietary directory will muscle out another supercharged directory. Data is open source. It is good that Yahoo lets businesses enhance their listings. But won’t businesses enhance their listings at Google and Microsoft, too? At some point, aren’t businesses going to say: “Look, this is the information that describes my business, if you want it, here it is, come and get it.” This is how the web works, and I think reasoning that posits that local search is fundamentally different is wrong. There are differences, but nothing that a bit of infrastructure, and structure, won’t fix.

Any local search solution that cordons off YP data from the general web corpus is making too big a sacrifice, and an unnecessary one. There is too much good local content directly on the web.

But Geosearch or Google Local alone – just unstructured content, can’t compete with the YP on two important playing fields. First, YP data is trusted – people rely on it for at least basic information on consumer-oriented businesses. Second, YP data is thin, but it is broad and consistent. You can reason about it because it has some structure. Businesses are categorized with a consistent scheme, such as SIC codes. It is an anomalous but fortunate circumstance that addresses ARE metadata. National Postal Services saw to that, to ensure the efficient delivery of land mail. But an address is not quite enough metadata to suffice for local search. That Google Local cross-checks web content with IYP data is a partial solution at best. The matching is imprecise, and though the IYP doesn’t contribute much information, what it does contribute is metadata.

Rich but anarchic, poor but orderly. Can these two disparate methods for aggregating local content be harmonized? I think so, and all roads lead to metadata.

The first foray into annotating web content was several years ago, by Dan Bricklin when he was at Interland. He proposed the Small and Medium Business Metadata (SMBmeta) initiative (http://www.smbmeta.org) as a way for enterprises to present essential information about themselves. This information would be in consistently named and structured XML files located at the top-level of participating enterprises’ domains. Since SMBmeta files have a consistent structure across websites, search engines could easily find and interpret them.

The question I pose and try to answer in my article is “What would it take for the Internet to duplicate and enhance the functionality of the print and Internet Yellow Pages?” A key challenge is to aggregate information for the millions of businesses that don’t have a web presence. One way to do this is to decentralize the job of collecting this data, to entities that are close to these businesses, such as Chambers of Commerce, or trade organizations. These entities have relationships of trust with both the public and the businesses they represent. They can act as gatekeepers, ensuring the data they collect or certify is authentic. There’s more in the article.

Come to our show in Santa Clara: http://www.kelseygroup.com/dd2005/dd2005_day1.htm

Can you make changes to your Yahoo! yellow page ad at will? If so, would make sense to me as a business owner to drop the paper pages for something I could change to reflect current sales/holidays/incentive programs. (Can’t change my paper ad during the year!)

Added bonus would be the stats/info via a tracking system they could provide. That would be invaluable info to me and not available with the paper pages.

Wonder why they are happy to co-exist….

You might be interested to note that here in Toronto, we’ve launched an offering – CentrSource.com (www.centrsource.com) that directly addresses the challenges and opportunities you speak to. We provide an open market intranet ASP solution (www.responseexchange.com), that any advertiser, agency or media can access to create microsites we call ResponseBoxes. These ResponseBoxes are effectively perishable mini-ecommerce sites that can be associated with advertising in any media, or they can serve as stand alone directory listings. This content can be accessed not only through our website, but also through any advertiser and/or media company that wants to provide a link to us on their website. Why would they do this? Because on clicking on the link, the consumer would be exposed only to the ‘live’ content that is locally relevant to them. This means the consumer in Seattle could be exposed to offers unique to that market, vis a vis a consumer clicking on the advertiser site in Miami. The microsite can be changed on the fly, meaning that merchants can have daily (or hourly) offers.

Our key differentiator is that we not only provide a directory solution, we enable the advertiser to give the tools to the consumer to indicate their place in the purchase cycle – put me on your email list, vs. CALL ME THIS VERY MINUTE!

We too have to beat the YP gorillas out there. Our solution is to pre-vet and load transactional (ie: etailer) websites on our system. For our current beta market in Toronto, we have loaded over 6,000 advertiser websites. These include over 2,000 local advertisers and thousands of international, US and Canadian websites. We have now started the process of going back to these advertisers to direct them to our self service pay for performance system. They only pay when the consumer clicks on a specific action item designed by the advertiser – and you will see we present many opportunities for the consumer to do so.

What you’d probably be most interested in though is our view regarding the CEO Consumer, the empowered user for whom we built a new online utility – CentrSource.com. You can find out more about our point of view – and our business -at http://centrsource.typepad.com.