Google changes Local OneBox Ranking algo! Or do they?

With the recent upgrade to the Google Local Onebox, Google has introduced a significant change to their main search engine result ranking of these local results.

I have previously noted that there was a discrepancy between the Local Onebox rankings and the ranking results in Google Maps. With this update, it appeared at first glance that the rankings of the Onebox results were closer to the results in Google Maps. Based on this very small sample below I concluded that Google Local OneBox was now returning the same results as Google Maps.

Here are the comparative results the main results page Local Onebox vs Google Maps ranking from December 17, 2006

Google Organic Onebox Local Results Google Maps Results
1)Anchor Bar – A.Hyatt Hotels & Resorts: Hyatt Regency Buffalo
2)Kuni’s Sushi Bar B.Adam’s Mark Hotels & Resorts
3)Hyatt Hotels & Resorts: Hyatt Regency Buffalo C.Buffalo Marriott Niagara
D.Anchor Bar
I. Kuni’s Sushi Bar

Here are to current results from February 3, 2007:

Google Organic Onebox Local Results Google Maps Results
1)Hyatt Regency Buffalo A.Hyatt Regency Buffalo
(716) 856-1234
2)Adam’s Mark Hotels & Resorts B.Adam’s Mark Hotels & Resorts
(716) 845-5100
3)Sheraton C.Sheraton
(716) 681-2400

Matt McGee from Small Business SEM was kind enough to point out that this was not universally true (ah, the benefit of peer review). But Matt contended that more Local OneBox results were unlike their Google Maps counterpart than were similar.

I immediately withdrew my hasty conclusion and went back to the drawing board. Was Matt correct or was I?

It was obvious that there were still differences between the results. For example, on the search restaurants Pittsburgh Pa you see different results between the Local OneBox Results and the Google Maps Results:

Google Local OneBox Results Google Maps Results
A. Omni William Penn Hotel www.omnihotels.com530 William Penn pl, Pittsburgh, 15219 –(412) 281-7100 201 reviews – Get directions – More information A. Omni William Penn Hotel – more info »530 William Penn pl, Pittsburgh, PA(412) 281-7100 – call –
B. Primanti Brothers www.primantibros.com46 18th St, Pittsburgh, 15222 –(412) 263-2142 76 reviews – Get directions – More information B. Hilton Hotels – more info »600 Commonwealth Place, Pittsburgh, PA(412) 391-4600 – call –
C. Mallorca Restaurant www.mallorcarestaurant.com2228 E Carson St, Pittsburgh, 15203 –(412) 488-1818 33 reviews – Get directions – More information C. Wyndham Garden Hotel University Pl – more info »3454 Forbes Ave. University Place, Pittsburgh, PA(412) 683-2040 – call –
D. Primanti Brothers – more info »46 18th St, Pittsburgh, PA(412) 263-2142 – call –
K. Mallorca Restaurant – more info »2228 E Carson St, Pittsburgh, PA(412) 488-1818 – call –

To attempt to establish an answer, I sampled 10 additional cities to determine if there was a large scale trend in the results indicating that there was greater similarity between the Local OneBox and the Maps results than in the past (although I have little baseline data).

After my first three samples I was feeling pretty good about my conclusions. But the results soon started shifting towards Matt’s conclusion. In the end? Of the 12, 6 results were exactly the same between the Local Onebox and Google Maps, 6 results were different.

Here are the results:

Google Search Local OneBox Results Google Maps Results using Local results for restaurant near link Amount of Difference
san francisco restaurant san francisco ca restaurant Same
chicago IL restaurant chicago IL restaurant Same
dallas tx restaurants dallas tx restaurants Same
atlanta restaurant atlanta restaurant Much Different Local F, C & G were A, B, C on OneBox
knoxville restaurants knoxville restaurants Much Different Local L, G & J were A, B, C on OneBox
rochester ny restaurants rochester ny restaurants Same
cleveland restaurants cleveland restaurants Same
olean ny restaurants olean ny restaurants Mostly Same Local A, C, B were A, B, C in OneBox
restaurant Minneapolis MN restaurant Minneapolis MN Much different Local L, F & A were A, B, C in OneBox
restaurant Detroit Mi restaurant Detroit Mi Much Different Local B, L & E were A, B, C in OneBox

Certainly, it is strange that Google still offers up different top 3 results on their main search results page than on Google Maps results. It is also interesting (although perhaps not important) that they do it roughly half of the time. The question is why?

It is possible that two distinct algorithms are at work and switch between cities periodically. That seems complicated to me. Perhaps they are testing to see which generates the least complaints or perhaps they are trying to overcome the inherent unfairness of only listing 3.

Another possibility is that Google Organic calculates the results slightly differently and uses ratings and other indicators of relevancy. One way to test that would be to evaluate a different industry that has no reviews on-line and see if they still present results differently.

Certainly, if Google is presenting “the most relevant results” it is hard to argue that different results are equally relevant.

Update 2/8/07: I think that Bill Slawski has identified the nature of the problem in his recent post on Location Prominence.

Update 2/9/07: Matt asks the Big Why question about this in his recent post: Why Does the Same Local Search Produce Different Results?

Please consider leaving a comment as your input will help me (& everyone else) better understand and learn about local.
Google changes Local OneBox Ranking algo! Or do they? by

12 thoughts on “Google changes Local OneBox Ranking algo! Or do they?”

  1. Mike,

    The table up there with the bold — where Hyatt, Adam’s Mark, and Sheraton are listed — based on my experience, I would say that’s the exception, not the rule. That’s the first example I’ve seen where the Onebox results are an exact match of the Maps results. I haven’t had a perfect match on ANY test queries I’ve done.

  2. Yes, I see that …I haven’t tested it widely….but it does appear to better align than before although what the differences are and why certainly need some looking at…it is possible that relevance and ratings still factor into the results on the onebox and not in Maps…

    The problem is I don’t have a large enough sample from previous research to see exactly how it has changed. I am going to have to dig into and see whats what,

  3. One difference that might be involved could be the that the boundaries of the maps being shown are different, which is a factor in deciding which locations to show..

  4. Hi Mike,

    It is complex, and I think that it is changing and evolving over time. My post about differences between Onebox and local results based upon location prominence only touches upon some of the issues involved.

    I know that some topics/subjects that people searc for involve showing maps at a greater or smaller scale.

    I also know that some locations have a greater density of businesses than others, and that can influence the scale of the map.

    The concept of which centerpoint to use has and can change over time – center of a specific zip code, center of a range of zip codes, center of a smaller map in OneBox, center of a larger map in Local, distance from a specific location when address information is very specific, use of latitude/longitude as a center point, use of GPS as a center point. Different queries may determine that different ones are used. Different locations – rural, suburban, urban – may determine which ones are used.

    The concept of distance from a centerpoint is also something possibly subject to change. Do you measure it by how the crow flies, or by the mileage if you have good road information, or by driving times if you can estimate those – and if you have different information – more complete or less complete – for some areas, do you mix that information up?

    When you have incorrect or inconsistent location information for a business, is there a lessening of the ranking for a site, that ties it to a specific location, which makes it might not rank as highly.

    When you have rating informationfor a business, or review information, you probably don’t treat all reviews equally – so a review from Yahoo might be worth more than a review from the Newark Post, or from the World Weekly News. So a site with a single review from an excellent resource that is editorially controlled – like one of the food critics from the New York Times – might carry more weight than a user’s review from Judy’s Book.

    I’m just painting with broad strokes at some of the complexities. And I think that Google is experimenting as we go, to see what seems to get the best reactions based upon click-throughs and non click-throughs, time spent on pages, similar searches in the same search session (or online session), and many other things. With millions of results to interpret, and the time to slowly make changes, they are going to cause those of us trying to follow along some headaches along the way.



  5. I’m seeing some maps/onebox showings that seem predicated on an alpha directory.

    listings seem to be made based on alphabetical order. Anyone else seeing anything like that?


Leave a Reply

Your email address will not be published. Required fields are marked *

Comments links could be nofollow free.