|"Location" results - relevance drop since Dewey update|
The reason for Google's high search engine market share is based on the relevancy of its search results.
Following its Dewey up-date it seemed to me that results relevancy had deteriorated dramatically. My impression was that the inclusion of a location word in the search phrase particularly scrambled results.
So I ran an audit comparing Australian pages search results on Google and Yahoo.
I assessed the top 10 search results for 16 different search phrases (160 search results). All search phrases included "Sydney" with either "PR" or "public relations" in combination with "company", "companies", "agency" or "consultants". "Online" was added to some of these phrases. In other words, all were very specific for someone looking for the services of a Sydney based PR company.
The results (for 1 Aug 08):
PR Companies listed:
For Google, that's an average of less than one PR company in the top 10 results over the 16 phrases.
All PR company pages listed:
Google 64 (40%)
Yahoo 106 (66%)
Non PR site pages listed:
Google 96 (60%)
Yahoo 54 (34%)
"Non PR" sites on Google included arts, media, commercial printers, unrelated directories, associations, jobs and edu sites.
My impression is that Google's results relevancy has crashed across many other search categories since the Dewey upgrade.
Has anyone else run similar audits?
Hello John, and welcome to the forums.
I agree that Google is struggling with queries that contain a location keyword. There are a couple factors here - one of which is an ongoing challenge with identifying the location of a site in the first place.
A second factor is Google's mini-campaign this year to increase "diversity" in the top search results. That may well be a significant factor that your audit is picking up. In other words, they don't even intend to be "literal" in the results they return for many queries. As Google expressed openly, they hope to return results for what you mean and what you need, not merely for what you typed.
There's a lot more to a search engine than local results - and Google has not shown a significant loss of market share since the dewey non-update in April. I'd estimate at least 100 updates and tweaks since then, and the Google audience remains massive and loyal.
[edited by: tedster at 4:56 pm (utc) on Aug. 3, 2008]
I want to make further comment, because I changed John's original title for this thread. It was originally "The Death of Google" - that's quite dramatic, and it would almost definitely have increased clicks into this thread. However, I felt that the original title was also a bit deceptive to the readership, since the audit John made is very small and confined to a very specific, local result.
A severe and extended drop in relevance across the board might BEGIN to hurt Google - but even then it would be a long path from the intial hurt to "death". And it's very unlikely that the slippage in user satisfactione would go unnoticed and uncorrected for very long.
In this case, where we are talking about Local Search, we're looking at a frontier area and not the core of their search product. And yes, this frontier is still wild and woolly! It's hard to "slip" in relevance while a benchmark is still being established.
Apologies for the overly dramatic title.
I’m aware of Google’s intention to “diversify” results and acknowledge this is a miniscule sample, hence the request for corroboration by others.
What prompted the post is that in every “service” industry I’ve ever audited in search engines (over 10 years of audits), the inclusion of a location word to cull out irrelevant results has been a major factor in search methodologies.
My impression is that Google used to deliver a lot more relevant results to these types of search.
The questions I pose are:
Has Google gone too far in its diversification program?
When people start to realise result relevance is suffering, will they switch search engines?
Thanks for the welcome to the forum.
John - Interesting that you've used this particular search, because it's a search that I've frequently used as a test search for Google (with the full spelling, not the abbreviation, and with a different city. I didn't tack company onto the end). IMO, Google's results are significantly better than they used to be on this search.
It used to be that the search was significantly skewed by the presence of sites belonging to trusted organizations, companies, and institutions with cityname in their name (and thus in their inbound anchor text).
Just the mention of "public relations" on the home page of such a site was enough to put it ahead of many cityname public relations companies.
Partially, this was due to what I perceived to be a flaw in Google's algo.
In part, though, when I looked more closely, I saw that it was also due to the emphasis on graphics over text in the PR field. It appears that both Google and the PR companies have gotten smarter.
So, granted that Google has had to go through some byzantine efforts to combat spam, and risk throwing out the baby with the bathwater at every turn... nevertheless, on this one search, anyway, it looks like the baby has survived. That may not be true in Sydney.
My feeling has always been that while Yahoo often looks like it satisfies queries better than Google does, or at least more literally, when you look at the results more carefully you see that Google does have a better perception of page and link quality. I can get a local site up to the top in Yahoo much more easily than I can in Google.
[edited by: Robert_Charlton at 5:47 am (utc) on Aug. 5, 2008]
Robert - I just ran some of the "PR" & "public relations" phrases with local city names in Google UK and Canada. They don't seem to be delivering the same level of unrelated search results as Google Aust.
I also tried, "accountants" & "plumbers" in combo with local cities with similar results.
Google Aust throws up many more job sites, major media and general directory sites than the others.
My impression is that the Dewey upgrade has added more weight to "site importance" or "off page" factors. Perhaps with our small web population an effect like this could be precipitated.
I wonder whether Google makes algo adjustment at a country level to redress possible issues like this?
I think right now and for the next several years, good results for local will be a challenge. Many local service providers that you would ideally want in the results are smaller companies who have a limited web presence that is unlikely to be very useful or SEOed. I have found it practically impossible to get much more than a phone number for a variety of contractors and services in my area. Honestly, the best local results tend to come about when there are national companies with local reach. Mom and pop are just too busy working to figure out the Interweb.