Yipit Blog

The problem with local search is a data problem. For local shopping, Yipit’s focus is on inventory information in local businesses (currently furniture in new york). Also critical is enhanced context in location. For the last hundred years, the leading local search solution (the yellow pages) provided a list of business contact info, organized by city. 

However, high-value local recommendations often require granular local context (e.g., “two blocks south of Penn Station,” “in Columbus Circle,” etc.). Whether you’re in New York or anywhere else, consumers often seek more than a street address.

Our friend Ahmed Farooq of iBegin has developed iBegin Places, an open solution for enhanced location information, including neighborhoods, public transporation and landmarks. 

This is a valuable resource for the local search community, and it’s optimized for our usability (and cross-platform compatibility):

all places (and subsequent ‘collections’ of ‘places’) are easily exportable via GeoRSS, Google Maps JS, JSON, and/or KML.

Similar data is being developed by several excellent by firms. iBegin Places, however, is free to use and open to contribute. We’re excited to see the local search community take advantage of this resource, and most importantly, continue to add valuable location context.

Local search results have been disappointing users for years. The main reliable source of data about local businesses remains the antiquated yellow pages’ 300-or-so categories. If you want to search by a more specific category, the results are spotty at best. So, what’s the solution?

Well, there are two proposals: 1) local user reviews / content and 2) more accurate business data. Clearly, both will be necessary but many new local search companies seem to putting their chips behind proposal #1. We at Yipit are focusing on proposal #2. Who’s right? That will hopefully be debated at a panel Yipit proposed for SXSW’s interactive festival. I say hopefully because the panels are determined, in part, by a vote. So, if you care to support Yipit and the local search industry, please submit your vote for our panel: The Local Search Solution - Context or Accuracy? I’m not going to say that this is more important that voting for the national election, but I’m also not going to say that.

As for the panel speakers, we are pleased to announce the following highly-respected members of the local search community have agreed to participate on the panel we suggested:

• Sebastian Provencher, Praized, author of the Praized Blog
• Perry Evans, CEO of Local Matters, co-founder of MapQuest
• Peter Krasilovsky, The Kelsey Group, author of Local Onliner

In addition to our panel, I would encourage everyone to also vote for another excellent panel proposal from Chris Tolles of Topix.com:

• Think Globally, Post Locally: The Emerging Power of Local Voice

In an effort to curtail manipulation of local search results, Google has recently articulated to local businesses that listing multiple business names at a single address is no longer acceptable. While this is a positive overall step for local search, the Manhattan local market presents too many obstacles to this approach for addressing mapspam.

Image by Jorge Gobbi

In a city of skyscrapers, single addresses often host hundreds of businesses. Making matters worse is a predominance of mail forwarding centers that provide an authentic-looking mailing addresses for only a few bucks a month. In New York City, address redundancies are not the exception, they are the rule.

There are several scenarios for why businesses might share an address, including:

1. Legitimate businesses sharing a single office building.

2. Legitimate service providers (e.g., contractors, transportation, etc.) who obtain a Manhattan forwarding address to improve their local SEO. For examples, look up Yipit’s address on Google Maps and see our thousands of “neighbors.”

3. Businesses with an actual local retail presence that use multiple names and web presences. See #1 and #3 in a search for modern furniture.

4. Businesses with no local retail presence appearing local by using a forwarding or otherwise fabricated address.

But shouldn’t checking for multiple businesses solve all of these scenarios, and weed out the good from the bad? Not in Manhattan, where the vast majority of businesses share a building. But shouldn’t suite, floor or office numbers correct for this? Perhaps, but they’re rarely available. Analyzing a sample of Manhattan yellow page data reveals the following:

• % of businesses located at an address occupied by at least one other business: 83.2%
• % of those businesses whose address includes a suite, floor or office number: 31.7%
• n = 7,425

In less dense areas, automatically eliminating address redundancies can go a long way. In Manhattan, multiple businesses that share the same address cannot be eliminated prima facie, so you need a local authority to determine the correct answer.

Based on our review of the Manhattan furniture market, we found local search engines’ current furniture local search results largely disappointing. Depending on the service we used, searches for “furniture ny,ny” returned many actual furniture stores, though few major retailers. Searching for “couches ny,ny” returned only bars, restaurants, hotels and no furniture stores. When a user goes beyond the broad furniture category and searches for a product type (like “couches”), search engines break down completely.

We analyzed the field of leading local search options including: Google Maps, Yahoo Local, Ask City / Citysearch, YellowPages.com and Yelp. We performed two searches, “furniture ny,ny” and “couches ny,ny” and looked at the top ten results. We excluded YellowPages.com from the survey because the results, ignoring the advertisements, were just listed in alphabetical order with no attempt at classifying based on quality.

The following tables summarizes our results. For each search engine’s top ten results, the first column of numbers represent the percentage of stores that sell any furniture and the second column indicates the percentage of stores that carry a wide selection.

Top 10 Results for “furniture ny,ny”:

Top 10 Results for “couches ny,ny”:

Here are some selected observations regarding the quality of the results returned by search engine.

Google Maps. Of the big three, Google Maps did the best with the furniture search returning eight out of ten actual furniture stores (though one of the results had an old address). The other two results were “A Salt and Battery” (their #3 result and an excellent fish and chips restaurant) and a furniture store that closed six months ago. However, only two of the eight furniture stores offer a wide selection of furniture pieces. Unfortunately, when it came to the couches search, Google broke down; they returned mostly hotels and one web design shop but no furniture stores.

Yahoo Local. We were most disappointed by Yahoo Local’s results. For a major, yellow pages defined category like furniture, just 2 actual furniture stores were returned, neither of which were major furniture stores. Also, their #4 and #7 results were the same business. As for couches, they returned four actual furniture stores but none had a sizable selection of couches.

Ask City / Citysearch. When Ask combined with Citysearch to provide a new local search engine, we thought they would be a strong competitor in the market. Unfortunately, they do not seem to have pulled it off. Ask City returned only five actual furniture stores (they also returned two stores that have been closed for months). Of those five furniture stores, only one was a major furniture store. When it came to the couches search, where we thought they could have an advantage with Citysearch’s database, they were unable to return any actual furniture stores.

Yelp. Yelp surprised us the most. They successfully returned ten furniture stores. While only four of those were major retailers, the stores they returned seemed to be driven by quantity and quality of user reviews. However, they do not seem to be able to handle more specific searches. Searching for couches returned only one actual furniture store with the remaining results being bars / lounges.

Conclusion. We were surprised by the low quality of the local search results we received. For a general category like furniture, we expected much better results (20% high quality results is disappointing). When it comes to going beyond one of the broad categories and searching more specifically (e.g., “couches”), it is clear that search engines do not have an answer. When product-type data is not readily available online, search engines will try to use whatever other context they have around the business (web page results, user reviews) to try to find relevant results. But, ultimately, that leads to highly inaccurate listings. We continue to believe that accuracy is one of the major components of building a successful local search engine and do not think current local search data / algorithms can provide it.

See Yipit results for Furniture New York and Couches New York.