The load time is not an issue really since it is just text data and only about 20 products. We have a nifty setup actually, we use Algolia to display the products on the category page, we are just making a double call to Algolia before the page render and taking their json array and modifying it into what is needed.

If you run this page through the Structured data testing tool, https://www.ebay.com/b/Fine-Jewelry/4196/bn_2408477  this is what one of us is proposing. See how the tool does not break each into a product node.

At the other time this is a page that the other is proposing,  https://www.apple.com/shop/mac/mac-accessories  It breaks the products into nodes. (albeit there are errors that can be fixed with the pricing)

I think this better illustrates the issue.

We are adding json+ld data to an ecommerce site and myself and one of the other people working on the site are having a minor disagreement on things. What it comes down to is how to mark up the category page. One of us says it needs to be marked up with as an Itempage, https://schema.org/ItemPage  The other says it needs to be marked up as products, with multiple product instances in the schema, https://schema.org/Product

The main sticking point on the Itemlist is that Itemlist is a child of intangible, so there is a feeling that should be used for things like track listings or other arbitrary data.

I use GT Metrix. If you sign up for a free account you can throttle the connection to mobile network speeds, http://screencast.com/t/LtNMGetvuu  They seem to have the best on the market for doing tests like this.

Jon,

I have never done it with a robots.txt, one easy why that I think you could do it would be on the page level. You could add a noindex nofollow to the page itself.

You can generate it automatically too and have it fired depending on the url by using a substring search on the url as well. That will get them all for sure.

It is either or really. Its not hard to tell that a site is running wordpress, it is pretty hard to hide. At the same time the admin directory does not have any public facing files for non-logged in users as well. I just use a good Wordpress security plugin, good server practices and leave it at that.

Good pick up on the Walmart issue. I actually took that screen shot when I was making that post, so it kind of confused me to how it changed so quickly. I did a bit more digging and figured out what they are doing. It basically looks like their canonical on their category pages is broken. See this video, http://screencast.com/t/Azyjjuxor8  So when you change pages, the canonical stays the same as your entry page and the rel=next/previous does not change either. But that would lead me to suspect if Google is crawling the site, they see the canonical pointing to the top level of the category. But interestingly enough if they ever did a fetch as Google on a paginated page, or if the crawler started with page 5 that would lead to some weird results Google would have to deal with. So effectively I would say their is broken.

It also seems like Rand might have changed his tune a bit since that article. When I did an interview with him, he said to canonical back to the main category page in faceted navigation situations. "Having 28 separate category pages for women’s pants where the only variations are size and color is probably not a great idea. In faceted navigation scenarios like that, I’d make the color and size filtered pages use a rel=canonical back to the non-faceted women’s pants page."  https://www.prestashop.com/blog/en/seo-expert-series-rand-fishkin-of-moz/

So for a better answer it might depend on how the category pages are structure with the navigation as well.