Dan Shure posted on his podcast his audio notes from after a private dinner with Gary Illyes of Google from April 4th 2017. Basically, they had a dinner with several SEOs, after the dinner when Dan got into his car, he took audio notes of what he remembered from that dinner.
I listened to the whole podcast and while some things are new, I don't think any of this is something that is secret and Gary would be upset with this being revealed. Either way, he published the podcast and people are chattering about it in public, so it is good to cover.
You can listen to the podcast over here. Here are Dan's raw notes:
- AMP does not pass the mobile-friendly test.
- Hidden JS content
- Purchasing an old domain with old content
- Changing the context and content of old URLs
- Losing star snippets can indicate a sitewide issue
- The structured data testing tool uses the same exact system as their algorithm
- Panda is a sitewide signal, and they have an actual Panda score
- 302s vs 301s and is this why Wired lost their traffic when going to https? (very surprising answer here)
- Ignoring partial manual action penalties
- What links are ALWAYS ok to remove or disavow (and not)
- The old “topical authority” algo
- Adding new low-quality URLs to a site
- Knowledge boxes in the US but not Canada
- How does Google know when a search like “ring” has a brand intent?
- Google’s concept of a “site” (subfolders vs subdomains – what is a “site”?) and domain clustering
- Sites that don’t need to worry about links in any way whatsoever
- DA is OK as a metric, but one thing to be aware of
- What is going to replace all manual actions
- Sitewide signals were updated in the Feb 2017 update
- Google does not click behind tabs etc (when fetching via JS is required) because they don’t want to end up in a secure or private area, or in a .php file etc
- How to rank for big head keywords like “meat”
I personally thought the Panda score stuff was not real. He said Panda is sitewide and the site gets a Panda score. He said "If the Panda score goes below a certain point that could lower your site quality which then could take away your snippets." This was all about 4:45 into the podcast. But we covered the concept of Panda scores a couple times in the past and I am not sure how relevant it is today with Panda kind of built into the core algorithm.
The Private Voice Memo I Made After Dinner with Gary Illyes [in 2017]
— Dan Shure (@dan_shure) July 29, 2019
- Google had a "topical authority" algo, replaced by Himmingbird
- The real reason Wired lost traffic when switching to https
- Partial Manual actions automatically expire
- LOTS morehttps://t.co/A85wwFNcH6
Here are some things I don't know if we heard before but I can be wrong:
At 6:00 in the podcast: Prior to May of 2016 Google had some type of spam precaution that prevented fully passing signals from http to https. It was resolved in 5/2016 (& it seems Gary helped w/this). BTW, the Wired situation had nothing to do w/302s vs. 301s. Was something else. pic.twitter.com/LhqzI6oR1P
— Glenn Gabe (@glenngabe) July 31, 2019
At 12:33 in the podcast: Google removed an older topical authority algorithm since Hummingbird began to handle what that algo used to cover. Hummingbird understands longer-tail queries better & helps Google understand the meaning behind queries. pic.twitter.com/w4POZbU4Wq
— Glenn Gabe (@glenngabe) July 31, 2019
In fact, Glenn Gabe has a Twitter stream of what he found interesting in the conversation.
Keep in mind, Dan could have misheard, misunderstood or misrepresented some of this conversation. Dan is a really good guy, so if he did, it was not intentional. Either way, this is a very interesting podcast and honestly, he has an amazing podcast you should all probably subscribe to.
Forum discussion at Twitter.
