Quick answer: We measured the marketing sites of nine Shopify SEO apps on 23 September 2026 — sitemap size, structured data, llms.txt, AI-crawler rules, hreflang, home page weight and response time. Eight of the nine publish in one language with no hreflang at all. Seven publish an llms.txt. Only three take any position on AI crawlers in robots.txt. Two ship a home page with no structured data whatsoever.

Why audit the vendors

An SEO app is a claim about competence. The cheapest way to test a claim like that is to look at what the vendor does on the one site it fully controls, where there is no merchant theme to blame and no platform to work around.

This is not a product comparison. Nothing below says which app fixes more issues on a Shopify store, and a company can run an ordinary marketing site while shipping an excellent app. What it measures is narrower and more checkable: does the vendor apply to its own domain the practices its product sells?

Our own site is in the table. It is not the best performer on every row.

The data

Every figure comes from a public HTTP request made on 23 September 2026.

Site Sitemap URLs llms.txt GPTBot named JSON-LD types hreflang Home HTML
avada.io 3,303 yes no 17 0 242 KB
yoast.com 2,133 yes no 12 0 218 KB
outrank.so 782 yes yes 9 0 443 KB
rankengine.app 706 yes yes 11 8 90 KB
storeseo.com 463 yes yes 10 0 250 KB
tiny-img.com 332 yes no 5 0 55 KB
seoant.com 99 yes no 6 0 226 KB
boosterapps.com 29 no no 0 0 69 KB
pluginseo.com 0 no no 0 0 3 KB

What the numbers say

Eight of nine publish in one language. Only one site in the set declares any hreflang alternates at all. Shopify merchants are not only English-speaking, and every one of these vendors sells into the same global marketplace. A zero here is not a bug — a site with no translated pages has nothing to declare — so it is a statement about publishing strategy, and the strategy is uniform.

llms.txt is already table stakes. Seven of nine publish one. Anybody still describing it as an edge is describing 2025. It costs nothing, no engine has committed to reading it, and at this adoption rate its absence is more noticeable than its presence.

Most of the category has not decided about AI crawlers. Six of nine robots.txt files never mention GPTBot. That leaves the default in force, which is access allowed — so the outcome is the same as an explicit welcome, arrived at by not choosing. Three sites name the crawler. All three allow it.

Two home pages carry no structured data at all. One of those also serves a 3 KB home page with no H1 and no reachable sitemap. The richest markup in the set belongs to avada.io, with seventeen distinct schema types on one page — which is its own kind of problem, because schema that describes nothing a searcher asked for is weight without benefit.

Home page HTML spans two orders of magnitude. 3 KB to 443 KB for the same job. The heaviest page in the set belongs to a vendor selling content optimisation.

What we would change on our own site

Publishing the table obliges us to read our own row honestly.

The sitemap says 706 URLs. Google Search Console, last updated 18 September 2026, reported 7 indexed. Sitemap size is a measure of publishing, not of performance, and ours is the clearest illustration in the table: a site can be technically clean, fast, structured and multilingual, and still be almost entirely absent from the index.

The reason is not on this table. It is authority — 43 referring domains, which is not enough to get 700 pages crawled and kept. Every column above is necessary. None of them is sufficient.

How to run this on your own site

Each column is one request and no tooling:

  • Sitemap URLs — fetch robots.txt, follow the Sitemap: line, count the loc elements. If it is a sitemap index, follow one level down.
  • llms.txt — request /llms.txt and check for a 200 with content, not a soft 404 that returns the home page.
  • AI-crawler position — read robots.txt and look for the crawler by name. No mention means allowed.
  • Structured data — parse every JSON-LD block on the page and list the @type values. Our schema markup generator writes the blocks; the Googlebot simulator shows what a crawler receives.
  • hreflang — count link elements with rel="alternate" and an hreflang attribute in the head.
  • Home page weight — the byte length of the HTML response, before images.

The measurement is the easy half. Deciding which numbers matter for your store is the part worth spending time on, and for a Shopify store the honest starting point is usually a full technical audit rather than any single column.