So, Google’s Internal Engineering Documentation Leak containing Google Search’s Content Warehouse API has hit certain websites and forums. Save yourself some time trawling through pages of updates on the latest Google document leak and dive into some initial takeaways…


The things that made me go “huh?”

I was first alerted to this from ipullrank’s fantastic initial dive into the google document leak, and while this document doesn’t tell us how much Google is valuing a ranking signal, it does give us insights into areas of the algorithm that we have arguably never had before.

The document is pretty dense and arcane, but I wanted to share some initial highlights from it that made me go “huh, that’s interesting”. So, without any more preamble, here are my top 4 (not 3) highlights.


1. Domain Authority, by any other name…

For years Google has been adamant that they don’t use Domain Authority to rank a site or assess it’s quality. It turns out this may have been a little bit of Google word play or obfuscation, as the term they use in their internal document is “Site Authority”.


2. What is the value of a click…

Another statement many prominent Googlers have provided many times is that they don’t use clicks as search ranking signals. Now, while that is their official line, thankfully the SEO community has largely been dubious of this – especially after Rand Fishkin conducted an experiment during his time at Moz that seemed to indicate the opposite – that clicks do impact rankings.

So not only does this indicate that clicks do impact rankings, but that different types of clicks play a part:

  • The last time a link was clicked
  • How long people stay on the page after clicking a link
  • Good clicks (for good people)
  • Bad clicks (from naughty bots)

I guess this means it’s time to polish up those meta descriptions and play around with your page titles to try and attract that click from the SERPs.


3. Twiddlers are not just for thumbs…

The use of the term “twiddler” made me go huh – and this was something that’s been highlighted as important to the various “Boost systems” that are in the documents. In a nutshell, a twiddler is a filter that runs after the main search algorithm and can adjust the score of web pages on the fly. 

For example if Google wanted to show more independence day content and news around July 4th for July 4th search terms, then they could use a twiddler for a specific time period to do this, then turn it off and revert the search results pages to what they were before.

iPullRank suggests that often the experiments Google runs before making major algorithm changes are run as twiddlers, which would make sense as they can be turned off without needing to make widespread changes to the core algorithm.

If you want a more in depth overview of twiddlers, this post by Rank Me Amadeus/Julian Redlich is a great place to start.


4. Format and more…

The final point from the google document leak that made me go “huh” are a few snippets that are related to content formats and length. So while I wouldn’t advocate for wholesale changes to current content, a few things to consider for new content are:

  • Google cuts off pages that are too long – so you need to make sure the most important content is coming first or at least early on in your pages
  • Short content earns a score based on originality – So you need to ask yourself, what am I adding to the conversation here, and is what I’m saying said elsewhere already?
  • Page Titles matter – they are still measured against queries, so put your core target term in them, though there is also no character count limit on them (even if they look ugly in the SERPs)


So what?

Firstly, this document does mean we can’t always trust what Googlers are telling us. Secondly, it shows the importance of running your own tests on content and page structure to find the best formats to rank pages.

Ultimately though we are in roughly the same place we were before, with a little more information, so we should:

  1. Create great content people want to interact with
  2. Consider what’s ranking already – and look to understand what will succeed
  3. Earn links (quality relevant ones!)
  4. Play with titles and descriptions to earn a click.

More insights and findings will no-doubt surface over the next few weeks and months, as well as a potential response from Google. If any more interesting topics or insights come up, we’ll be sure to cover them here.