As the Director of SEO R&D for Slingshot SEO, an Internet startup born in the cornfields outside the great city of Indianapolis, Indiana, I’m fascinated by and driven to understand the complexity of Google’s “computationally intense algorithm.” I’m also fascinated by what Google engineers will say when their Public Relations department isn’t around. Because of these fascinations and drive, I’m constantly monitoring what search engineers are working on in their respective fields and then incorporating aspects of their developments into SEO best practices. In the list below, I’ve gathered some interesting quotes from Google engineers and have made some observations (some while wearing a tinfoil hat) about what they may mean for SEO. Beginning with quotes and observations about speed, I’ll then move on to ones that involve “user data.” Finally, I’ll do some wild speculation on what I think is in store for the future of image recognition. Hopefully, the SEOmoz community and its collective intelligence will take these quotes and further “distill” them into insights we can all use.
Knowing that speed is important to Google’s search experience, it’s no wonder that it was one of the original tenants of their search philosophy. Because of this attention to speed, Google performs related tests to refine the way users interface with their product.
Given the recent announcement that Google is now using “Site Speed in Web Search Ranking(s),” these quotes from one of Google’s VPs helps illustrate my point that following what Google engineers are saying can often provide clues to what Google is working on for the future of search. Because of these past quotes about response time and given Google’s recent announcement, it would have been wise to have included something as simple as “page load speed” in every on-page SEO audit since at least mid 2009. Wrapping up Google’s thoughts on speed, and demonstrated by the quote below, it’s interesting to know that Google will even impose speeds on their users.
Given this quote, we can gather that Google has the ability to throttle usage at will. In terms of SEO, and with my tinfoil hat placed firmly on my head, the ability to throttle results could mean many things. For example, given a query, if Google knows a certain demographic converts fractionally more reliably after a certain amount of time, it therefore makes sense that they would consider implementing a strategy around this data to maximize relevancy and/or profit. Speed isn’t the only thing Google engineers talk about.
This quote supports my theory that Google is weighting co-occurrence terms and synonyms of a given search query and not just the exact or derivative phrases. I would assume that this would count for on-page content, as well as in-bound link anchor text and image alternative text. These co-occurrence terms and synonyms may communicate to Google that content containing them in combination with exact phrases is more relevant than content or in-bound links simply containing exact or derivative keyword phrases. Google can easily identify these terms for most search queries as demonstrated by their related searches at the bottom of the SERPs (search engine results pages).
At first, this statistic might be hard to believe. However, I think it speaks again to the importance of focusing on quality, keyword and co-occurrence rich content generation through blogs, news and forums. Furthermore, constantly adding unique content to your website will help your site win longtail search phrases, or search phrases with three or more words that are less competitive than shorter phrases. Given the engineer’s ideas about analyzing their own usage data, and the recent development in Webmaster Tools that now includes the CTRs (click through rates) on search terms, it’s becoming increasingly more difficult to discount Google’s interest, capability, and motivation to capitalize on user behavior.
The inclusion of “user data” as a ranking factor is a hot button issue in the SEO world. Given this, there are compelling arguments and refutations. However, Udi Manber has repeatedly implied that Google takes user behavior into account when it tweaks its algorithms for enhanced relevancy.
Knowing that Google has the ability to predict search behavior, some “tinfoil hat” observations can be made about “seasonality” as it pertains to user data. For example, do search results change in Mexico around Spring break when there’s a massive influx of tourists? Does a result for a popular, yet seasonal, tattoo shop appear all year long for “Panama Beach tattoo shop?” Surely, there are many potential ways Google could handle this type of query. Accounting for past user data may be one of them.
It would seem that Google does factor CTRs into their search algorithm. Whether or not they are a ranking factor is debatable. However, if for some reason your website makes it to the first page of Google for a competitive search phrase and no one clicks on it, then why should it remain on the first page? The take-away here is to focus on creating effective page titles and meta description tags. Your page titles and meta descriptions should contain exact keyword phrase matches, benefits and calls to action that would entice a searcher to click on your listing. Also, I would suggest that you make sure to keep your page titles within the 70 character limit so they look clean and professional. Short, keyword rich URLs have also been shown to increase CTRs for listing.
This may come as no surprise to most; however, I find it very telling of the future of how Google will look at external factors that influence rankings. Links within blog posts will continue to be positively weighted. More importantly, this quote speaks of plans within the Googleplex to incorporate the social graph in the algorithm to affect web search and other search verticals. Considering that the social graph is largely “nofollow,” it’s then likely that there must be other factors surrounding social links; e.g. number of followers/friends, CTRs on past links, etc. Regarding image recognition, represent an area of search that I believe will become increasingly more relevant as engines begin to reliably and accurately assign authority and trust to images that are related to the written content surrounding them. Google claims to be making advancement in this area.
Using lessons from quantum research to more efficiently and effectively recognize complex patterns; e.g. images. If page load speed is a ranking factor today, then I would argue that image recognition optimization may become a factor in the future (if it’s not already one). The reason I believe this is because sites that are able to more accurately display the content they wish to represent, even at an image quality level, should be more authoritative. In theory, one reason Google may be telling sites to speed up their pages; i.e. compress them more, is so they can fit higher quality/accurate/more content in their datacenters. With image recognition optimization in mind, it would be interesting to see what SERPs would begin looking like with pages that have the most contextually accurate and authoritative images of “Water Lilies” or “Advil.” Furthermore, what do odd concepts like “wild boar insurance” look like to a search engine?
Personally, it was fun finding these quotes, riffing on them, and putting them together for the SEOmoz community. Similarly, I encourage the community to occasionally check in on what Google’s engineers are saying and am looking forward to continuing this conversation down in the comments.
