Chat with us, powered by LiveChat
Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor. Lorem ipsum dolor sit amet, consectetur.
Contact us now +1128 5255 8454
[email protected]

Instagram Feed


Using Search Patents to Hack Google’s Secret Metrics

Most of the time, SEO teams have to play a guessing game—can we predict how Google’s algorithm works, or what they’ll do next? While SEOs have a good intuitive sense of what ranks well, and conduct a lot of research of our own, the algorithm remains top secret. We never get to “look behind the curtain.” But what if we did—or more accurately, what if Google is already giving away its secrets in public documents?

That’s more or less the idea behind a great new piece from Barbara Starr, who has been sleuthing through patents for Google’s search technology. Google files a lot of these (no surprise there), and they contain a tantalizing inside look at how exactly Google determines relevance.

To be clear, none of these patents give away everything. But there’s a perennial tension in patents between keeping trade secrets secret, and giving away enough specifics to make sure your idea is protected. Barbara decided to dig through these with a specific question in mind: how is Google choosing which data to pull out and highlight in knowledge graphs and info panels? And how can you make your website the one to get featured?

Structured Data

This question is important for two reasons:

  1. Google increasingly provides “quick answers” to questions so that users don’t even have to click on regular search results. If your site never gets into their info panels and sidebars, you may get overlooked.
  2. Even among organic results, Google may choose a few to highlight (such as in a “three pack” of local business listings). Being among the featured few means more visibility.

To choose what to feature and how, Google relies on “structured data”—not just relevance but the way different pieces of information relate to each other. For example, a search for “US president” will show an info box about Barack Obama, with related info (and a picture) drawn from the web. That’s fundamentally different than just a list of relevant websites, and structured data is what makes that possible.

Startling Discoveries

In the patents, Barbara has found extremely detailed info on the types of metrics that guide this structured data search process. Unsurprisingly, these are much more elegant than just how many times a keyword is mentioned, or who’s linking to whom. She identifies four key metrics that influence what get featured in those info panels. My favorite of these is the “Fame” metric.

The Fame metric is actually a special application of the Contribution metric. Contribution measures influence. It evaluates a rich cross-section of data sources such as social media influence, online reviews, references in books and scholarly articles, etc; all potentially weighted by the importance of the source. (Google’s patent suggests that reviews by professional critics carry more weight than consumer reviews, for example.)

“Fame” is a submetric that applies this to people. It is an incredibly powerful barometer with a lot more behind it than just a Klout score. This is why a search for Tom Hanks not only shows a sidebar about the actor, but correctly believes that a list of his movies would be a relevant thing to include there. Each of those movies is a Contribution factor involved in his Fame score.

Making the Metrics Count

All of this is fascinating to us search nerds, but is there any way to put this info to work for a business? The answer is both yes and no.

To Barbara, these search metrics are actionable intel. She notes that in today’s searches,

“…specific regions of the search results are defined or templated in some manner, and ranking/ordering for each varies by context or domain… From an SEO point of view, this means that optimizing a company’s website or web presence will be based on targeting these templates, each of which may well have their own ranking algorithm based on context.”

But that amounts to trying to not only rank for certain topics, but to also do so predictively in a way that lands you in an info panel. Given that only one source will likely be featured in such panels, and that the panels only appear in certain types of queries, that is a very high-level game with a tiny chance of payoff.

For most SEO clients, you don’t need to get featured in a knowledge graph; ranking well in local results and for your keywords is enough to keep traffic and customers flowing in. It would be nice if your picture of a hamburger was the featured image for “hamburger” searches worldwide, but it’s far from the most effective way to drive traffic.

To stay in top of your SEO game, bring in qualified experts who understand how the field is changing and what counts as high-impact optimization. At EverSpark Interactive, we start off each client with a comprehensive SEO technical audit that looks at over fifty factors influencing your ranking. Then we work only on the factors that will have the biggest result.

Want to find out more? Contact us today and get a FREE consultation.