Richard Hearne, https://xitter.cf/t.co/Snrr3a1mO1: these days mainly interested in Technical SEO and Machine Learning.
Phuket
Joined December 2007
- Tweets8.1K
- Following451
- Followers1.2K
- Likes6.3K
Good sleuthing:
so, ChatGPT released sample code for their web crawler. Guess what, it completley strips out all Schema. So let's put that myth to bed. H/T no twitter having Rodrigo Stockebrand for this github.com/openai/gpt-oss/bl…
PSA: if you're using GSC Impressions with individual XML Sitemaps, the Impressions data Google is showing are aggregate for the site.
Previously, they returned the Impression count for the URLs in a sitemap. This seems to no longer be the case.
Bug or feature? I'm not sure.
Not trying to flex at all David. It's a weak argument based on very limited cases where HTML sitemaps *might* have some slight edge over XML sitemaps. I didn't see any context about small sites in this thread, but even then I'd take XML over HTML all day long. YMMV of course.
Without telling me you've never worked on a large multi-million url site. Never mind a massive 8- or 9-digit url site...
So they have systems to identify this, yet they are asking - cap-in-hand - for users to report issues for further investigation? Their ranking data has been useless for some time now due to this issue. Any claims of mitigation are at odds with reality inside their product.
Fair play to Nozzle publishing this. Not really in their own interest, and the other tools seem pretty tight lipped about it. But hopefully Nozzle will get the attention they deserve for their research.
Some very serious data analysis showing just how much poisoning is going on:
nozzle.io/research/poisoned-…
TL>DR;
27.5M SERPs analysed
660K+ keywords with poisoned SERPs
49 countries - Austria had 18.8% of scraped SERPs show poisoned results.
8.0% of SERPs poisoned
Most likely a third party product summarizing the reviews, and I'm pretty confident the bad reviews being left in is a conscious decision. Very easy to train or prompt out the bad in this response.
Nice that they kept it honest - not only the good, but also the bad highlighted. Probably also a fantastic way to capture product feedback. Is it indexable, and does it rank? Also wonder if it works well for AI?
I think you and I are doing something very different. I need to dog out my old tests and screenshots. The one major difference I see immediately is that I always tested rendered real content instead of dynamically painting elements. But we seem to have very different outcomes.
On the flattened DOM - this I was aware of previously, and has been the case for a decent time IIRC.
Thinking a bit more about this overnight. Google recently changed their docs to say that they changed their max HTML size to 2MB (before 15MB). I wonder now if they resize the window to any height so long as its content is within the 2MB limit?
Ok. Good to know. I think this is new behaviour since I last tried it. I'll have to get testing again to see how this works now.
Why is innerHeight changing after 5s? I think you may be doing something which isn't typical of what WRS does. I tested this by loading a very long page and then outputting read innerHeight at top of page. Then just read this from the screenshot. No dynamic loading involved.