The Top 5 Takeaways From Google’s Search API Leaks

Transcription: 

Alright, now that the dust is settled, I wanted to go over some of our top five takeaways from the Google Search API leaks.

First, a big preface: just because something is included in the API leaks does not necessarily mean it is used as a ranking factor. However, these are some of our biggest takeaways and some of the biggest changes that we’re going to be making to our strategies moving forward based on the information found in the API leaks.

The first takeaway, and probably the biggest, is that Google is clearly using clicks more aggressively than we previously thought as part of the ranking factors. If we look at references to Navboost, Google’s system for getting feedback on click data, it’s referenced 130 different times in this documentation. There are all types of variables: clicksBad, clicksGood, clickInterval, totalClicks. It’s very clear that they’re using clicks more heavily than we ever thought to get feedback on search results. So, it appears that optimizing for clicks might have stronger impacts than we previously thought. Navboost is the system they are using to inform all of this, and it’s referenced a lot in the Google documentation.

The second takeaway is that freshness really matters. This is something we’ve known at Go Fish Digital. We’ve done different tests around this and seen very strong results from updating freshness. This was verified in the Search API leak. In Mike King’s analysis, he found that Google uses three different variables: bylineDate, semanticDate, and more, heavily leaning on the dates used in content. Our belief is that if Google thinks the content is outdated, it can’t trust the content. For example, Google wouldn’t trust an article from 2020 to rank for “best smartphones”; it would need an article from the present day. So, freshness really matters, and we see evidence of that in the API leaks as well.

The third takeaway is that we believe Google is using more than just an individual piece of anchor text to judge the context of links. This is something our founder Dan Hinckley found. There are two different attributes: fullLeftContext and fullRightContext. The description talks about how the full context of the text around the anchor text is considered. Google might be using the sentence before and the sentence after the anchor text to understand its context. So, even if you use generic anchor text like “click here,” Google might still figure out its context.

The fourth takeaway, which I found super interesting, comes from an article on Search Engine Land that unpacked the Google API leaks. They found that Google uses a urlHistory variable. Andrew Ansley found that Google might keep 20 different versions of a given webpage in the index. If a document has more than 20 changes, they only keep the last 20 changes. This means Google might keep historical references of your content on their servers. At Go Fish Digital, we’ve seen Google using old title tags we had moved on from, which this explains: Google kept a historical record of the URL and maybe, based on click data, used the old version.

The final takeaway is that Google almost certainly uses some type of topical authority and overall site authority metrics. We’ve seen this in variables like siteAuthority. Despite Google saying they don’t use Domain Authority, here it directly references site authority. Google seems to care about a site’s core area of focus, referenced in variables like siteFocusXcore and siteEmbedding. These indicate Google wants you to claim expertise in a certain vertical and not stray too far from it. Embeddings convert your text into numeric values, which could be used to measure how far your content deviates from your topical authority.

Those are our biggest takeaways from the Search API leaks. I’d love to hear any insights you have found. These are some really interesting insights, and I’m excited to see the new testing and findings as people continue to uncover these leaks and discover new insights.

About Chris Long

Chris Long is the VP of Marketing at Go Fish Digital. With 10+ years of SEO experience, Chris works unique problems and advanced search situations to help clients improve organic traffic through a deep understanding of Google's algorithm and web technology. Chris has advised on the search strategies for clients such as GEICO, Adobe & The New York Times. He is a contributor for Moz, Search Engine Land, and Search Engine Journal. He is also a speaker at industry conferences such as MozCon, SMX East, BrightonSEO and more.