A Moment of Reflection
A few years ago, we lost a heavyweight in the SEO community when Bill Slawski passed away. If you never heard of Bill, it wouldn’t surprise me much. He held a niche position in the industry as the primary contributor of deep analysis, interpretation, and writing about Google and other search engine patents and whitepapers. He had a talent for taking these complex technical documents and making them accessible and understandable for the broader SEO community. Unless you truly geeked out on the internal workings of search engines and were constantly trying to stay a step ahead of the competition, you would most likely avoid the complex nature of his posts.
His findings were foretelling though. Despite the fact that Google would repeatedly claim that they didn’t use one metric or another within their algorithm, Bill would routinely find patents and other documentation suggesting they did. I was reminded of this as I wrote a previous post concerning the methodical rollout of Google’s algorithm updates. His work helped shed light on what was actually going on behind the scenes at Google and, for me, it was an immense help in understanding the nature and direction of search as we knew it.
Lies! Lies! And More Lies!
While Bill was never an “Ah-ha! I have proof!” type of guy, much of what he covered as possibility came to fruition in May of 2024, when a leak of Googles Internal Search API documentation became public. In it, over 14,000 potential ranking attributes, many of which contradicted or provided more detail than Google’s previous public statements.
While the leaked documents do not reveal the exact weights of these factors, the key ranking signals and features discovered include:
1 – User Behavior and Click Data
The documentation strongly suggests that Google uses user-click data as a ranking signal, which Google representatives have historically downplayed.
- NavBoost and Click Signals: Features related to the NavBoost system, which uses click logs to boost or demote rankings, were present. Metrics include:
- goodClicks and badClicks: Likely indicators of user satisfaction or dissatisfaction with a search result.
- lastLongestClicks: Tracking the longest period a user spent on a page before returning to the search results.
- unsquashedClicks: Clicks that aren’t normalized or filtered out.
- Chrome Data: The documents reference the use of data from the Chrome browser (chromeInTotal) for site-wide traffic and engagement metrics.
2 – Site Authority and Quality
The leak confirmed the existence of internal metrics that align with the concept of domain or site authority, which Google had previously stated they do not use.
- Site Authority: Metrics like siteAuthority and hostAge were revealed, suggesting that the overall credibility, history, and age of a website influence the ranking of its individual pages.
- Sandboxing: An attribute called hostAge is used to “sandbox fresh spam in serving time,” which supports the long-suspected “sandbox” effect where new websites face an initial hurdle in rankings regardless of content quality.
- Originality/Quality: Features that measure Content Originality and quality scores (pageQuality) suggest a focus on unique and non-easily replicated content, potentially using models to estimate the “effort” put into a page.
3 – Links and Authorship
- Backlinks: Links, and specifically the PageRank of a site’s homepage, are still explicitly referenced. The documents also mention demotion factors, such as anchorMismatch (when the link’s anchor text doesn’t match the target site).
- Author Authority: Metrics related to author authority and reputation were referenced (authorReputationScore and modules like WebrefMentionRatings), indicating that the expertise of the content creator is a factor.
4 – Content and Page-Specific Factors
- Content Freshness: Google tracks multiple dates associated with a document, including bylineDate (explicit date on the page), syntacticDate (date from URL/title), and semanticDate (date derived from content analysis), reinforcing the importance of fresh and updated content.
- Page Titles: The documents contained a module suggesting that the quality of page titles can have site-wide implications for ranking.
- Whitelisting: References were found for whitelisting certain domains/authorities for sensitive topics like elections (isElectionAuthority) and COVID-19, confirming that certain sources receive preferential treatment.
Those in the Know
Bill Slawski was widely regarded as a legend in the industry for sharing his knowledge, helping to shape the practice of SEO as it is known today. But the primary reason I bring him up is that, with his passing, there seems to be little emphasis in the SEO community to follow in his footsteps and regularly dive into the patents that Google, Bing, or any other search-oriented giant is regularly churning out. And, while I am certainly just as guilty in my laziness, I figured I’d offer some sources to go to if you’re into this type of thing.
The best places to find information on patents filed by major companies like Google, Bing (Microsoft), Facebook (Meta), and YouTube (Google) are public patent databases.
Here are the most reliable and comprehensive resources, along with tips on how to search:
1 – Google Patents
- What it is: A free search engine hosted by Google that indexes patents and patent applications from over 100 patent offices worldwide, including the USPTO, EPO, WIPO, and more.
- How to search: This is often the easiest place to start.
- Go to Google Patents.
- Use the Advanced Search feature and look for the “Assignee” field.
- Enter the company’s name. You may need to try variations, such as:
- For Google/YouTube: Google LLC or Google Inc.
- For Facebook: Meta Platforms, Inc. or Facebook, Inc.
- For Bing/Microsoft: Microsoft Technology Licensing, LLC or Microsoft Corporation
2 – United States Patent and Trademark Office (USPTO)
- What it is: The official source for all U.S. patents and published applications.
- How to search:
- Use the Patent Public Search tool on the USPTO website.
- In the Basic Search or Advanced Search, look for the field specifically for Assignee Name or Applicant Name (since the company is typically the assignee/applicant).
- Use the same company name variations as noted for Google Patents.
3 – World Intellectual Property Organization (WIPO) – PATENTSCOPE
- What it is: WIPO’s search service that provides access to the published International PCT applications and national patent documents from many participating offices. This is excellent for checking global filings.
- How to search:
- Go to PATENTSCOPE.
- Use the Advanced Search or Field Combination search.
- Search in the Applicant or Assignee field for the company name.
4. European Patent Office (EPO) – Espacenet
- What it is: A free, worldwide patent search system offering access to over 140 million patent documents from around the world.
- How to search:
- Use the Smart search or Advanced search options.
- Search using the Applicant field for the company name.
These databases allow you to search specifically by the entity (company) that owns the patent, which is known as the Assignee.
Looking Forward
Just as Bill regularly updated us on how search was evolving, utilizing these databases can help you understand where search—especially in the era of artificial intelligence—is heading. It just takes a little work to get your head around the concepts you find in the documentation.
Leave a Reply