,
14 tweets,
4 min read
Read on Twitter
Thread Warning (1): Here's how the AMA with @methode changed my understanding of search. I'm always happy to learn things and now I have more questions. There's more on Reddit than I will cover, these are just my thoughts before and after. reddit.com/r/TechSEO/comm…
(2) On hreflang before I was mixed. John said it didn't impact rankings but it sure felt like it did. Gary made some comments in a video before where it sounded like they shared signals in the cluster.
(3)After on hreflang: It sounds like it's a double pass where strongest is taken initially on it's own merit (no shared signals in the cluster) then swapped for the most relevant page.
(4) Takeaway (opinion) is unless needed, one language version might be better than a bunch of country-language versions since having one would be stronger. I'm not sure a little localization would offset the lost value here say between US and UK, but definitely not UK and AU.
(5) I did not know this, but it sounds like image recognition used as a relevancy signal. Makes a good case for having relevant images, not random pictures of generic people doing generic things.
(6) They're not going to use any public facing tools to measure site speed. I thought they were going to use CrUX and lighthouse data. Maybe their renderer has some kind of timers/lighthouse it runs? I don't know, but none of what we see will be what they see apparently.
(7) There was a question on noindex, follow. Before @johnmu said eventually these may be treated as nofollow. Gary asked John to clarify and looking forward to that response but Gary said as long as they keep trying to access that page they'll follow the links. (so conflicted)
(8) 302 I knew kept signals at the source, but I thought the new page also built signals. In every example I've looked at (please prove me wrong) the page redirected to, even if it has a canonical to itself is being consolidated with the original source. info:page-B shows me A.
(9) There was some clarification on pages redirected to irrelevant pages (many to one like a homepage) being treated as "soft 404s" as John has said a few times, whereas Gary says the signals do pass on, but...
(10) if it it does become a soft 404 then no signals will pass but it will be reported which I guess means in GSC since my original question was basically how do I troubleshoot this thing when I never see these as soft 404s, so I guess they counted?
(11) And another one where I thought redirects consolidated until they didn't (expired / no longer in place, domain drops, lost in migration, etc), it sounds like that's not really true. I'm hoping for further clarification but he makes it sound like after a period of time...
(12) the signals are all pushed to the new one and if something happened then the source collects signals on it's own again, which to me sounds like it resets to 0 and all the value stays at the page that was redirected to. I have a lot of questions on this now...
(13) Is that a 0 reset? Does it reclaim the original signals and take away from the page that was redirected to? What if I want to change the location to a better match? What if it was never put in place to begin with, is it too late? @methode @johnmu re: reddit.com/r/TechSEO/comm…
(14) and finally things mentioned I'm looking forward to are Gary's writeup on using ML and NLP for SEO and a writeup on faceted navigation (where I feel there are a lot of theories and practices here that are probably bad. lol). And a new word of the day, pornyness.
