On May 27th, Rand Fiskin at SparkToro broke the news that he’d been forwarded 2,500 pages worth of API (Application Programming Interface) documentation from Google’s search division. After gaining background from anonymous sources, as well as the leaker identifying himself as Erfan Aizmi, the documents were deemed authentic.
The documents pointed to 2,596 modules with 14,014 attributes (API features) that appear to come from Google’s internal “Content API Warehouse.” As mouthwatering as this sounds to SEOs who want to know the eleven herbs and spices that make up the precise signals and weights of Google’s PageRank algorithm, it falls far short of unveiling the secret formula to Coca-Cola.
Being code documentation, the information contained within the leak is highly technical in nature - and perhaps even above my paygrade.
A deep-dive by Mike King of IPullRank says:
“The leaked documentation outlines each module of the API and breaks them down into summaries, types, functions, and attributes. Most of what we’re looking at are the property definitions for various protocol buffers (or protobufs) that get accessed across the ranking systems to generate SERPs (Search Engine Result Pages – what Google shows searchers after they perform a query).”
If I’m correct, it shows people how the computers at Google access certain repositories of information or code to execute the SERP - not really how the SERP is defined itself. However, it did reveal a couple of aspects of Google’s algorithm that are worth re-evaluating.
MAKE A REFERRAL - GET $50!
〰️
MAKE A REFERRAL - GET $50! 〰️
Clicks May Be More Important Than We Think
Google has long squashed the idea that clicks are not a ranking factor in search. According to other leaks and testimony from the US Department of Justice Anti-Trust trial, Google does employ a “implicit user experience” ranking algorithm called NavBoost, which ranks content higher based on actual clicks versus expected clicks. These signals are separated into badClicks, goodClicks, lastLongestClicks and unsquashedClicks.
Let’s say a well-edited and researched document ticks all the EEAT boxes. At the end of the day if human users prefer another piece of content, it makes sense to rank that content higher as user experience matters most. NavBoost is mentioned 84 times in the documents as a heading, which implies the algorithms contained within are important. How these work, are weighted, or tracked, etc. we cannot say.
Domain Authority - Possibly A Thing?
Google themselves have asserted they don’t use “website authority” scores, a metric designed by Moz.com, in their search rankings. The documents reveal a Compressed Quality Signal called siteAuthority() - how this is computed, measured, or implemented is anyone’s guess.
Chrome Is Used For Ranking
As mentioned before, NavBoost helps rank content based on click and user behaviour. Some user behaviour - though we aren’t privy to what extent - is harvested using cookie history, logged-in Chrome data, and pattern detection algorithms. This is known as the ChromeInTotal module.
This has been known as far back as 2016, as real-time Chrome data would be used in their RealTime Boost Signal. If you are queasy at the idea of Google tracking this much information about your browsing habits, then I encourage you to switch to Firefox.
Other Modules in the Leak
Other modules in the leak that hint at the presence of certain ranking factors are timeliness, by looking at dates in the byline (bylineDate), URL (syntacticDate) and on-page content (semanticDate); titlematchScore which suggests that site titles are a major ranking factor; determines the focus topic of a page using a comparison of page embeddings (siteRadius) to the site embeddings (siteFocusScore); and also stores domain registration information (RegistrationInfo.)
What Does This Tell Us About SEO?
On the surface - not a hell of a lot. This leak may be akin to finding out that KFC uses salt shakers, flour stations, and deep fryers in the making of their fried chicken, but not the specific ingredients nor the requisite quantities to replicate their secret recipe. It also tells us things we already knew but Google were coy over, such as Domain Authority and using Google Chrome data to reinforce their SERPs. I’ve used a gang of weasel words to report on this topic, and with good reason: we may have a few of the puzzle pieces but we have no idea if they fit together, let alone what the entire picture is supposed to look like.
It may have given some, but not total credence to long-held suspicions about how Google Search works, but nothing definitive. Though we may not be able to reproduce Coke exactly at home, it doesn’t stop people from trying.
We may not have cracked the code of SEO, but I Sell Words has boosted the SERPs of hundreds of businesses around Australia with quality content. Get your own today!