Posted by randfish
In early June of this year, SEOmoz released some ranking correlation data about Google’s web results and how they mapped against specific metrics. This exciting work gave us valuable insight into Google’s rankings system and both confirmed many assumptions as well as opened up new lines of questions. When Google announced their new Places Results at the end of October, we couldn’t help but want to learn more.
In November, we gathered data for 220 search queries – 20 US cities and 11 business "types" (different kinds of queries). This dataset is smaller than our web results, and was intended to be an initial data gathering project before we dove deeper, but our findings proved surprising significant (from a statistical standpoint) and thus, we’re making the results and report publicly available.
As with our previous collection and analysis of this type of data, it’s important to keep a few things in mind:
With those out of the way, let’s dive into the dataset, which you can download a full version of here:
Interestingly, the results we gathered seem to indicate that across multiple cities, the Google Places ranking algorithm doesn’t differ much, but when business/query types are considered, there’s indications that Google may indeed be changing up how the rankings are calculated (an alternative explanation is that different business segments simply have dramatically different weights on the factors depending on their type).
For this round of correlation analysis, we contracted Dr. Matthew Peters (who holds a PhD in Applied Math from Univ. of WA) to create a report of his findings based on the data. In discussing the role that cities/query types played, he noted:
City is not a significant source of variation for any of the variables, suggesting that Google’s algorithm is the same for all cities. However, for 9 of the 24 variables we can reject the null hypothesis that business type is a not significant source of variation in the correlation coefficients at a=0.05. This is highly unlikely to have occurred by chance. Unfortunately there is a caveat to this result. The results from ANOVA assume the residuals to be normally distributed, but in most cases the residuals are not normal as tested with a Shapiro-Wilk test.
You can download his full report here.
Next, let’s look at some of the more interesting statistical findings Matt discovered. These are split into 4 unique sections, and we’re looking only at the correlations with Places results (though the data and report also include web results).
With the exception of PageRank, all data comes via SEOmoz’s Linkscape data API.
NOTE: In this data, mozRank and PageRank are not significantly different than zero.
All data comes via SEOmoz’s Linkscape data API.
NOTE: In this data, all of the metrics are significant.
All data comes directly from the results page URL or the Places page/listing. Business keyword refers to the type, such as "ice cream" or "hospital" while city keyword refers to the location, such as "Austin" or "Portland." The relatively large, negative correlation with the city keyword in URLs is an outlier (as no other element we measured for local listings had a significant negative correlation). My personal guess is nationwide sites trying to rank individually on city-targeted pages don’t perform as well as local-only results in general and this could cause that biasing, but we don’t have evidence to prove that theory and other explanations are certainly possible.
NOTE: In this data, correlations for business keyword in the URL and city keyword in the title element were not significantly different than zero.
All data comes directly from Google Places’ page about the result.
NOTE: In this data, all of the metrics are significant.
Our hope is to do this experiment again with more data and possibly more metrics in the future. Your suggestions are, of course, very welcome.
As always, we invite you to download the report and raw data and give us any feedback or feel free to do your own analyses and come to your own conclusions. It could even be valuable to use this same process for results you (or your clients) care about and find the missing ingredients between you and the competition.
One of the topics that emerged from Pubcon was “Should SEO’s Focus on Where Google is Heading”, and I’m going to agree with Aaron that focusing on short term algorithmic holes isn’t a smart thing for most people (churn and burn folks–you keep on keeping on). I agree that most publishers should focus on where Google is going. However, the one thing I think publishers need to be aware of and be wary of is Google’s transition to becoming an answer engine.
When I refer to Google trying to become an answer engine, what exactly do I mean? I mean that Google will provide the answer right on the SERP itself if possible and, more frequently, from a Google-owned or Google-maintained property. What exactly do I mean by that? I would be willing to bet that at least one Googler is hunched over a monitor somewhere trying to figure out how to convert voice searches into standardized results. Get out your best Jean Luc Picard impersonation, grab your android phone, and say “COMPUTER … Show me airline prices from Los Angeles to Las Vegas on March 15th.” Now imagine that Google, using its recently acquired ITA travel data, could show you the 5 cheapest flights without needing to send you to the airline, travelocity, or any of the other intermediaries. Good for Google, good for the user … but scary if you are a publisher.
Google has been moving in this direction for years with queries like [what is george washington's birthday]
There’s no need for the person performing that query to visit any website because Google became the answer machine. Earlier this year, they began making inroads in commercial searches for things like [mortgage rates]
Google’s latest incursion into becoming the answer machine came from its local results when they began stealing … err aggregating … reviews from other sites and mixing them with their own on place pages.
IMHO this represents a clear and present danger to every web publisher. For a while, Google will be content to let publishers keep serving the information that Google hasn’t figured out how to gather efficiently/profitably, even if that means referring users to low quality, demand media style pages from About.com and eHow.com. However there’s no doubt in my mind that once Google thinks they can do better, they will scrape your data and throw you under the bus without a second thought … cause it’s all about the users, right?
The one exception that may leave you a leg to stand on is if you are a brand and are building some sense of brand loyalty. If users type in [<brand name> + <keyword phrase>] Google will show less “Google answers”. For example [george washington's birthday wikipedia] or [bank of america mortgage rates] contain none of the Google properties. Of course, it would seem to me that this is a massive conflict of interest as far as Google is concerned, but I’m not a legislator, so what do I know.
The days of being a pure affiliate and building sites without any thought to branding are coming to a close. They will never disappear completely, but there will be less of them. The purely keyword-based traffic without a hint of branding is going to become more competitive and, in some cases, you will be competing with Google itself or with Google owned properties like Boutiques.com. Heed these warnings Caesar and fear the Ides of March …
photo credit: Michal Osmenda
Generally I have not been a huge fan of registering all your websites with Google (profiling risks, etc.), but they keep using the carrot nicely to lead me astray. … So much so that I want to find a Googler and give them a hug.
Google recently decided to share some more data in their webmaster tools. And for many webmasters the data is enough to make it worth registering (at least 1 website)!
When speaking of keyword search volume beakdown data people have typically shared information from the leaked AOL search data.
The big problem with that data is it is in aggregate. It is a nice free tool, and a good starting point, but it is fuzzy.
There are 3 well known search classifications: navigational, transactional, and informational. Each type of query has a different traffic breakdown profile.
Further, anecdotal evidence suggests that the appearance of vertical / universal results within the search results set can impact search click distribution. Google shows maps on 1 in 13 search results, and they have many other verticals they are pushing – video, updates, news, product search, etc. And then there are AdWords ads – which many searchers confuse as being the organic search results.
Pretty solid looking estimates can get pretty rough pretty fast.
If there is one critical piece of marketing worth learning above all others it is that context is important.
My suggestions as to what works, another person’s opinions or advice on what you should do, and empirical truth collected by a marketer who likes to use numbers to prove his point … well all 3 data sets fall flat on their face when compared against the data and insights and interactions that come from running your own business. As teachers and marketers we try to share tips to guide people toward success, but your data is one of the most valuable things you own.
In their Excel plug-in Microsoft shares the same search data they use internally, but its not certain that when they integrate the Yahoo! Search deal that Microsoft will keep sharing as much data as they do now.
There have been some hacks to collect organic search clickthrough rate data on Google. One of the more popular strategies was to run an AdWords ad for the exact match version of a keyword and bid low onto the first page of results. Keep the ad running for a while and then run an AdWords impression share report. With that data in hand you can estimate how many actual searches there were, and then compare your organic search clicks against that to get an effective clickthrough rate.
Given search personalization and localization and the ever-changing result sets with all the test Google runs, even the above can be rough. So what is a webmaster to do?
Well Google upgraded the data they share inside their webmaster tools, which includes (on a per keyword level)
Even if your site is rather well known going after some of the big keywords can be a bit self-defeating in terms of the value delivered. Imagine ranking #6 or #7 for SEO. Wouldn’t that send a lot of search traffic? Nope.
When you back away the ego searches, the rank checkers, etc. it turns out that there isn’t a ton of search volume to be had ranking on page 1 of Google for SEO.
With only a 2% CTR the core keyword SEO is driving less than 1/2 the traffic driven by our 2 most common brand search keywords. Our brand might not seem like it is getting lots of traffic with only a few thousand searches a month, but when you have a > 70% CTR that can still add up to a lot of traffic. More importantly, that is the kind of traffic which is more likely to buy from you than someone searching for a broad discovery or curiosity type of keyword.
The lessons for SEOs in that data?
Search is becoming the default navigational tool for the web. People go to Google and then type in “yahoo.” If you don’t have a branded keyword as one of your top keywords that might indicate long-term risk to your business. If a competitor can clone most of what you are doing and then bake in a viral component you are toast.
Arbitraging 3rd party brands is an easy way to build up distribution quickly. This is why there are 4,982 Britney Spears fan blogs (well 2 people are actually fans, but the other 4,980 are marketers).
But if you want to pull in traffic you have to go after a keyword that is an extension of the brand. Ranking for “eBay” probably won’t send you much traffic (as their clickthrough rate on their first result is probably even higher than the 70% I had above). Though if you have tips on how to buy or sell on eBay those kinds of keywords might pull in a much higher clickthrough rate for you.
To confirm the above I grabbed data for a couple SEO tool brands we rank well for. A number 3 ranking (behind a double listing) and virtually no traffic!
Different keyword, same result
Link building is still a bit of a discovery keyword, but I think it is perhaps a bit later staged than just the acronym “SEO.” Here the click volume distribution is much flatter / less consolidated than it was on the above brand-oriented examples.
If when Google lowers your rank you still pull in a fairly high CTR that might be a signal to them that your site should rank a bit higher.
Enough about our keywords, what does your keyword data tell you? How can you better integrate it to grow your business?SEO Book.com – Learn. Rank. Dominate.
Posted by randfish
As of yesterday, both Bing and Google have confirmed (via an excellent interview by Danny Sullivan) that links shared through Twitter and Facebook have a direct impact on rankings (in addition to the positive second-order effects they may have on the link graph). This has long been suspected by SEOs (in fact, many of us posited it was happening as of November of last year following Google + Bing’s announcements of partnerships with Twitter), but getting this official confirmation is a substantive step forward.
In addition to that revelation, another piece of critical data came via yesterday’s announcement:
Danny Sullivan: If an article is retweeted or referenced much in Twitter, do you count that as a signal outside of finding any non-nofollowed links that may naturally result from it?
Bing: We do look at the social authority of a user. We look at how many people you follow, how many follow you, and this can add a little weight to a listing in regular search results. It carries much more weight in Bing Social Search, where tweets from more authoritative people will flow to the top when best match relevancy is used.
Google: Yes, we do use it as a signal. It is used as a signal in our organic and news rankings. We also use it to enhance our news universal by marking how many people shared an article.
Danny Sullivan: Do you try to calculate the authority of someone who tweets that might be assigned to their Twitter page. Do you try to “know,” if you will, who they are?
Bing: Yes. We do calculate the authority of someone who tweets. For known public figures or publishers, we do associate them with who they are. (For example, query for Danny Sullivan)
Google: Yes we do compute and use author quality. We don’t know who anyone is in real life
Danny Sullivan: Do you calculate whether a link should carry more weight depending on the person who tweets it?
Google: Yes we do use this as a signal, especially in the “Top links” section [of Google Realtime Search]. Author authority is independent of PageRank, but it is currently only used in limited situations in ordinary web search.
We now know that those link sharing activities on Twitter + Facebook are evaluated based on the person/entity sharing them through a score Google calls "Author Authority," and Bing calls "Social Authority."
We can probably predict a lot of the signals the search engines care about when it comes to social sharing; some of my guesses include:
We can probably also take a stab at some of the signals Google + Bing use for Author/Social Authority in the context of the sharing/tweeting source:
These signals represent my opinions only, and while it’s very likely that at least some are being used, it’s even more likely that there are many more that aren’t listed above. Over time, hopefully we’ll discover more about the impact of social sharing on web rankings and how we can best combine SEO + social media marketing.
To me, the most exciting part about this is the potential to reduce webspam and return to a more purely editorial model. While people often link to, read and enjoy sources that link out manipulatively, very few of us will be likely to follow a Twitter account, friend someone on Facebook, or "like" something in a social site that’s inauthentic, manipulative or spammy.
The social graph isn’t necessarily cleaner, but the complexity of spam is far lower.
Here’s to the evolution of organic marketing – search, social, content, blogs, links – it’s all coming together faster than ever before, and that’s a very good thing for holisticly minded web marketers.
Google is growing *far* more complex.
Page x can rank based primarily on the criteria for page y from that same site. So if you analyze the data behind the page which is showing up in the search results, in some cases you will be looking at the wrong data sets!
Google has typically displayed page titles in the search results. However there are now numerous exceptions to that.
Google has aggressively pushed into regionalization and localization, but sometimes they miscategorize a website’s target market or a user’s location … delivering irrelevant search results.
Sometimes Google pulls data from 3rd party data sources and lists that next to your listing. I mean, sure they have used DMOZ historically, but why exactly are they showing my site as having Russian text on it?
As Google grows in complexity, the number of bugs in their system multiply. Sometimes you don’t rank because you screwed up. But sometimes you don’t rank because Google screwed up. Typically Google sees minimal difference either way, as there will always be another website to fill up the search results. But as a business owner, when Google gets it wrong you can be screwed pretty bad, particularly if you stock physical inventory and have to tightly manage your supply chain & cash flow.
Recently Google created a news source attribution tag. If it works, it might be a good idea. But (even outside of spam) there are ways it can backfire.
Consider Google’s webmaster verification tags. One of our customers had an outing with an old webmaster who in turn did a sneaky change of location inside of Google Webmaster Tools over the weekend. After seeing traffic fall off a cliff, we figured out what happened & registered the site in Google webmaster tools. There are instructions on how to remove the former registered user, however the option does not appear in my client’s account.
The redirect will allegedly be in place for 180 days! The only way to get around it is to ask for a review by the Google engineering team.
In the past SEO was primarily about encouraging (and, perhaps, in some nefarious angles, coercing) Google into giving you a little more love. But now a big part of SEO is identifying & fixing Google bugs. And this is before you consider things like universal search, custom vertical result sets, search personalization, social layers, traffic shaping, changing crawling priorities, sites getting hacked, filters hit by aggressive SEO or competitive sabotage, more random filters, changing related words in search vocabularies, using search query chains, Google Instant shifting keyword demand profiles, search engines competing against SEOs by monetizing the organic search results, basic/straightforward SEO concerns, technical crawling & indexing related issues, improper server configurations, new additional signals being added to search – like sentiment analysis. etc.
A big part of brand building in terms of SEO is not only to help you build up passive organic links (and to be able to charge a premium for your product), but it is also something which helps establish a bit of a competitive moat from algorithmic errors & makes it harder for Google to accidentally dump you. Further, if you have brand & catch a Google bug there is little risk to asking for a review. But if you do not have brand then even asking for a review could be risky.
Anyone who tells you that ‘SEO is easy’ across the board is either ignorant or exceptionally ignorant.SEO Book.com – Learn. Rank. Dominate.
Interesting little snippet from Mr Cutts:
“Matt recommends SEOs do not “chase the algorithm” and instead try to predict where Google will be going in the future”. Matt was addressing PubCon.
Good advice, methinks.
Trying to predict where Google is going is something we do a lot of at SEOBook.com. Whilst no one has a crystal ball, it’s good practice to keep one eye on the search horizon.
So, where do we think Google might be heading?
Easy one, huh.
Their biggest competitors appear clueless when it comes to search. Bing may make some inroads. Maybe. It’s hard to imagine anyone eating Google’s lunch when it comes to search, for many years to come.
Is Facebook a threat? I doubt it. Search is difficult, and I can see no reason why Facebook – which has a media focus – could own the search channel any more than Yahoo could.
Search is, after all, an infrastructure problem. Google’s infrastructure would be very difficult to replicate.
A search result set only really contains spam if the Google users think it contains spam i.e. they don’t see the answer they were expecting.
The fact a website may fall outside Google’s guidelines might get competing webmasters’ knickers in a knot, but it probably doesn’t matter that much to Google, or anyone else.
Even though Matt Cutts says Google will devote more resources to this, I suspect Google’s efforts will largely remain focused on outright deception i.e. misrepresentation, hijacking and malware.
We can forget the San Fran techno-hippy ethos of the web. It will not be a free-for-all democracy, if it ever was. History shows us that power tries to centralize control in order to maintain it.
Google may try to keep users on Google for longer. They do this by owning more and more verticals, and extracting data and reformatting it. When they send visitors away from Google, they’ll try to do so more and more on their own terms. Watch very carefully what type of sites Google rewards, as opposed to what they may say they reward.
Expect less competition in the market as a result. Some people are already getting angry about it.
Google follows users. So do Facebook. Anywhere your users are, you’ve got to be there, too. On Google Maps. On YouTube. Wherever and whenever. Think beyod your website. Think in terms of getting your data out there.
As Rich Skrenta pointed out in a recent interview:
Social media can drive tons of attention, awareness and traffic. But the search box is the best way to navigate to stuff you want. Now what will drive those results – if I type in “pizza”, what should I get? The answer can be very different depending on whether the results are coming from the web, Yelp, or Facebook. So I guess my answer is that I still see search being the core way to navigate, but I think what gets searched is going to get a lot more structured and move away from simple keyword matches against unstructured web pages
Microsoft Research found that people tend to organize their memories in geographic terms i.e. where they were when something happened.
If you want to know where Google is heading, then watch Marissa Meyer. Marissa has been responsible for much of what you see in Google in terms of how it is organized. Marissa has just moved to head of Geographic and Location Services.
Google Earth. Google Maps. Google Local. Google Street View. Mobile location data and targeting. Expect more data to be organized around locality.
“…but this changes everything…”
SEO hasn’t changed all that much in years. We still find an audience (keyword research), we publish content, we build links to the content, and then we repeat it all over again.
The changes come around the edges, especially for big companies like Google. There is a lot of risk to Google in making radical changes. Shareholders don’t like it. Why risk breaking something that makes so much money, and is so popular?
The biggest changes in the way we do things on the web are probably going to come from the upstarts. They’re probably hard at work in their garage right now.SEO Book.com – Learn. Rank. Dominate.
Ben Edelman did it again
This time he highlighted how Google hard codes their search results:
[When] we roll[ed] out Google Finance, we did put the Google link first. It seems only fair right, we do all the work for the search page and all these other things, so we do put it first… That has actually been our policy, since then, because of Finance. So for Google Maps again, it’s the first link. – Marissa Mayer
If they gain certain privileges in the marketplace by claiming to not abuse their power and that their algorithmic results are neutral, but those algorithmic results may be pushed below the fold, then is it “only fair” for them to put themselves in a default market leading position in any category they feel they can make money from by selling ads in? Or is that an abuse of power?
As Google adds features, collects more data, puts ads everywhere, and pushes into being a publisher on more fronts, at some point there will be a straw that breaks the camel’s back. Big money is paying attention and the list of “evidence” grows weekly. Sometimes they still think like a start up. And that will lead to their demise.
It might not be anytime soon, but eventually they will hit a whammy.SEO Book.com – Learn. Rank. Dominate.
For the past few years, right before the prime 4th quarter holiday selling period, Google pushes out an update that causes an upheaval in the organic SERP’s. This has lead to talk of there being a Google Holiday Update Conspiracy, with the goals of making SEO look bad to drive up Adwords spending. Is there any truth to these rumors or is is just people upset about rankings loss? Let’s take a deeper look at the issue …
First let me preface this with a little background. When I worked as an in house SEO for a large regional retailer many moons ago we had a policy: no new programming goes live on the website between November 1st and January 1st. That was when the company made its greatest profit and, like any smart business, we didn’t want to “break” the checkout process or have some other programming issue mess things up. Later on, we became concerned about not messing with our search engine rankings as well. We wanted to make sure we had our best foot forward, and this is a sentiment shared by businesses today … including Google.
Google has the same goals you do, giving their customers (ie users) the best product (the best SERPs) during a time of peak search volume. If more people believe they will get better results, more people will use Google and the advertisers will get more exposure for their ads. Your goal of maintaining your rankings during the holiday period doesn’t play a role in Google’s decision making. Google doesn’t really care about your rankings … only you do. Google just wants to give users the overall best results.
So is this a conspiracy on Google’s part to make SEO’s look bad, or to drive up revenue? I’m going to uncharacteristically disagree with Aaron and Peter and say … No I don’t think the holiday update is designed to shake up SEO or drive up adwords revenue. Those are just nice side bonuses
As I said before, Google knows search volume increases right before the holidays, and they want to make sure they are putting out the best product they can (their SERP’s). Your specific website’s ranking or traffic doesn’t play a role in the decision making process. From Google’s perspective, they want to eliminate spam and be as resistant to manipulation as much as possible, so it makes sense that some “bad,” “aggressive,” or “guideline violating” tactics will get negated. The sad fact is that, with any change, there will likely be some collateral damage of “innocent” websites in the short term, but that will get sorted out down the road. I know that’s not much consolation for a business owner who has seen a 90% drop in traffic, but it’s not personal. There’s no shortage of websites to fill the SERP’s. Sometimes the players just get shuffled around.
Does Google engineer these changes to occur right before the holidays to drive up ad revenue? I don’t think so. Google made instant search to give users results quicker and hopefully have a better user experience. The fact that it helped them make millions of extra dollars in revenue is just a bonus. Google makes its decisions based on what it thinks is best for users … not publishers … The sooner you understand that concept the better. From Google’s perspective, Google’s job isn’t to drive traffic for you to your website, Google’s job is to give users the best results. It’s your job to build a brand or create some other unique service offering that makes people look for you … not just for the generic keywords.
When Bing launched, one of the interesting things they did to make the organic search results appear more relevant was to use link anchor text to augment page titles (where relevant). This would mean if people searched for a phrase that was mostly in your title (but maybe your page title was missing a word or 2 from the search) then Bing might insert those words into the page title area of your listing if they were in some of the link anchor text pointing into your page.
Before being switched over to Bing, Yahoo! would sometimes display the H1 heading as the clickable link to your site (rather than the page title). Bing also uses on page headings to augment page titles.
Historically if Google has thought it would appear more relevant to searchers, sometimes they have shown a relevant machine generated piece of your page displaying those keywords in context rather than the meta description in the snippet, but typically Google has been far more conservative with the page titles. Sometimes Google would list the ODP title for the title of a page, most until recently they have generally typically just listed the page title as the clickable link to your site.
Recently Google has grown more experimental on this front, being willing to use link anchor text and on-page headings as part of the listing. In addition, if the page title is short, Google may add the site’s name at the end of the title.
Here is an example in Google of the page title being replaced by part of an on-page heading & also showing the site’s name being added to the end of the link
And here is the associated on-page heading for the above
I have also seen a few examples of the link anchor text being added to the page title in Google, however it was on a client project & the client would prefer that I didn’t share his new site on an SEO blog with 10′s of thousands of readers.
Last November Matt Cutts recently did a video on the topic of Google editing the page titles for relevancy & how it was a fairly new thing for Google. Even back then Google was quite conservative in editing the clickable link … I think they have only grown more aggressive on that front in the past month or so.