---
title: "Yoast, Rank Math, AIOSEO or No SEO Plugin: What 212 WordPress Sites Give AI Agents"
slug: wordpress-seo-plugin-agent-readiness-corpus
published: 2026-11-20T03:14:50.462163+00:00
updated: 2026-11-20T03:19:04.972854+00:00
author: "M Asif Rahman"
author_url: https://ar.bd
category: "AI Readiness"
tags: check:C2, check:C3, check:B2, platform:wordpress, WordPress, SEO plugins, llms.txt, corpus study, AI readiness
description: "212 WordPress sites split by SEO plugin: plugins add JSON-LD and often llms.txt, but none adds named AI bot rules. Check-by-check data from 20 Nov 2026."
url: https://aiscan.site/blog/wordpress-seo-plugin-agent-readiness-corpus
---

**Direct answer:** Across 212 WordPress homepages in our scan database on 20 November 2026, an SEO plugin made a clear difference on exactly the checks it has a feature for: JSON-LD was present on 153 of 155 sites with a detected plugin against 32 of 57 without one, and llms.txt passed on 54.8% against 40.4%. On the checks no plugin ships by default, named AI bot rules in robots.txt and Markdown copies, sites with Yoast or Rank Math did no better than sites with no plugin at all.

The comparison posts that rank for "Rank Math vs Yoast" now carry an AI section, and they disagree with each other about which tier ships llms.txt. None of them counts what the plugins actually put on live sites. We have a scanner that records which SEO plugin a WordPress site runs and grades 21 AI-readiness checks on the same page, so this article does the count: 212 WordPress sites, split by the plugin our scanner detected, check by check.

## Quick summary

| Question | Answer from 212 WordPress sites | Source |
|---|---|---|
| Which plugins showed up? | Rank Math 78, Yoast 63, none detected 57, AIOSEO 8, ThinkRank 5, SEOPress 1 | Our scan DB, latest scan per host, 20 Nov 2026 |
| Does a plugin add JSON-LD? | Yes: 98.7% of plugin sites vs 56.1% with none detected | C3 evidence string, same query |
| Does a plugin add llms.txt? | Partly: 54.8% pass C2 with a plugin vs 40.4% without | C2 status, same query |
| Does a plugin add named AI bot rules? | No: 26.5% pass B2 with a plugin vs 29.8% without | B2 status, same query |
| Does a plugin add a Markdown copy? | No: 23.2% vs 26.3% pass C1 | C1 status, same query |
| Highest mean score? | AIOSEO 74.6 and ThinkRank 74.2, both too few sites to rank | Mean of `score` per group |
| Most common plugin, mean score | Rank Math 68.5, Yoast 66.6, none 63.2 | Same query |

The short version for a site owner: your SEO plugin probably handles structured data already, may or may not have llms.txt switched on, and almost certainly says nothing to GPTBot, ClaudeBot or PerplexityBot by name. That last part is a robots.txt edit you make yourself.

## How we measured it

**The population.** AIScan stores every scan it runs. For this study we took each host's latest site scan on rubric `2026.08.2`, removed our own domain aiscan.site, and kept the hosts our platform detector marked as WordPress. On 20 November 2026 that gave 869 hosts in total and 212 on WordPress (24.4%), scanned between 25 August and 20 November 2026. The queries ran at about 03:00 UTC on 20 November against our production database and returned only aggregates: counts, means and medians per plugin and per check. No individual site is named here, and none was fetched again for this article.

**The plugin label.** The scanner reads the served HTML of the homepage and matches it against a fingerprint for each SEO plugin it knows. That has three consequences you should keep in mind while reading the tables:

1. "None detected" includes sites running a plugin we do not fingerprint, such as The SEO Framework or Slim SEO, and sites where a cache or minifier stripped the plugin's comment.
2. A site that runs two SEO plugins still gets one label.
3. The label says which plugin is installed, never which of its features are switched on. A Yoast site with llms.txt off and a Yoast site with it on look the same to the detector.

**Small groups.** AIOSEO (8 sites), ThinkRank (5) and SEOPress (1) are too few to rank. They appear as counts in the tables and are left out of every percentage comparison in the text.

**Disclosure.** ThinkRank is our own WordPress SEO plugin, built by the same team as AIScan. It sits in the tables because the detector found it on five sites, and it is not ranked against anything.

**The sample.** These are sites somebody chose to scan, most often their owner checking their own site. That makes the group more interested in AI readiness than the average WordPress install, and it means the numbers describe this sample, not WordPress as a whole.

## The master table: 212 WordPress sites by detected plugin

| Detected plugin | Sites | Mean score | Median | C2 llms.txt pass | C3 pass | B2 AI bot rules pass | C1 Markdown pass |
|---|---|---|---|---|---|---|---|
| Rank Math | 78 | 68.5 | 67.5 | 37 (47.4%) | 57 (73.1%) | 19 (24.4%) | 22 (28.2%) |
| Yoast SEO | 63 | 66.6 | 65 | 37 (58.7%) | 39 (61.9%) | 15 (23.8%) | 11 (17.5%) |
| None detected | 57 | 63.2 | 59 | 23 (40.4%) | 23 (40.4%) | 17 (29.8%) | 15 (26.3%) |
| AIOSEO | 8 | 74.6 | 74 | 8 | 7 | 4 | 0 |
| ThinkRank (ours) | 5 | 74.2 | 65 | 3 | 3 | 2 | 2 |
| SEOPress | 1 | 80 | 80 | 0 (partial) | 1 | 1 | 1 |

Two things stand out before any detail. The gap in mean score between Rank Math and no plugin is 5.3 points, smaller than most plugin marketing would suggest. And the column where a plugin should matter most for AI agents, the named bot rules, is the one where the no-plugin group is ahead.

## Where the points go: dimension scores and levels

A mean score hides which part of the rubric a plugin moves. AIScan groups its checks into five dimensions and scores each from 0 to 100 on the sites where it applies. Here are the three large groups side by side.

| Dimension (mean score, 0 to 100) | Rank Math (78) | Yoast (63) | None detected (57) | Gap, plugin minus none |
|---|---|---|---|---|
| Discoverability (robots.txt, sitemap, links, 404s) | 93.9 | 92.8 | 87.6 | about +6 |
| Content (llms.txt, structured HTML, Markdown, feeds) | 61.7 | 58.8 | 54.5 | about +6 |
| Bot access (named AI rules, Content Signals, Web Bot Auth) | 26.9 | 25.0 | 32.2 | about -6 |
| Commerce, where it applies (44, 27 and 22 sites) | 22.4 | 17.9 | 20.0 | none |

The pattern from the check tables holds at dimension level. Plugins lift discoverability and content by roughly six points each and leave bot access where it was, slightly below the no-plugin group. Bot access is the weakest dimension for every WordPress group in the sample, and on Yoast sites the mean is a quarter of the available points.

Levels tell the same story from another angle. AIScan assigns a level from 0 to 5 by score band.

| Detected plugin | Level 0 to 2 | Level 3 | Level 4 | Level 5 | Scored exactly 100 |
|---|---|---|---|---|---|
| Rank Math (78) | 7 | 35 | 25 | 11 | 1 |
| Yoast SEO (63) | 3 | 36 | 20 | 4 | 3 |
| None detected (57) | 13 | 27 | 7 | 10 | 4 |

A plugin mostly lifts sites out of the bottom levels: 23% of no-plugin sites sit at level 2 or below, against 9% of Rank Math sites and 5% of Yoast sites. It does far less at the top. The no-plugin group has a larger share at level 5 than Rank Math (17.5% against 14.1%) and more perfect scores than either plugin group. A site at the top of this rubric has usually done the robots.txt and Markdown work by hand, and the SEO plugin is not what got it there.

The capabilities dimension (API catalogs, MCP cards, OAuth discovery) applies to only some sites, so it is left out of the table: it applied to 43 Rank Math sites, 14 Yoast sites and 6 with no plugin, which ties into the one result we cannot yet explain, further down.

## llms.txt (check C2): the plugin feature shows up, unevenly

According to the [llms.txt proposal](https://llmstxt.org/), fetched from llmstxt.org on 20 November 2026, a site publishes a Markdown file at `/llms.txt` with an H1, a short summary and lists of links. Our C2 check fetches that path and passes a file that answers 200 with an H1 and either sections or links.

| Detected plugin | Sites | C2 pass | Partial | 404, no file | Other failure | Median size of files served |
|---|---|---|---|---|---|---|
| Rank Math | 78 | 37 | 6 | 29 | 6 | 18,352 bytes |
| Yoast SEO | 63 | 37 | 3 | 23 | 0 | 3,301 bytes |
| None detected | 57 | 23 | 6 | 28 | 0 | 5,763 bytes |
| AIOSEO | 8 | 8 | 0 | 0 | 0 | 73,027 bytes |
| ThinkRank (ours) | 5 | 3 | 0 | 2 | 0 | 2,381 bytes |

Yoast sites pass most often of the three large groups, 58.7%, and their files are small: a median of 3,301 bytes. Rank Math files are more than five times larger at the median, which fits a generator that lists more of the site by default. Every one of the eight AIOSEO sites passes, with a median file of 73 KB, which looks like a full content dump. Eight sites cannot tell you whether that is AIOSEO's default or the habit of eight owners.

Rank Math has the most hosts with no file at all, 29, and six that fail for another reason (blocked or erroring responses rather than a clean 404). A plugin that can generate the file does nothing until somebody turns it on.

**What each plugin says it ships, checked on the day.** The scan cannot see a plugin's settings, so we read each vendor's own pages on 20 November 2026:

| Plugin (version on wordpress.org, 20 Nov 2026) | llms.txt | Named AI crawler rules | IndexNow | Source |
|---|---|---|---|---|
| Yoast SEO 28.9 (10+ million installs) | Yes, free: "Built into Yoast, free for everyone" | "Bot blocker" for GPTBot, CCBot and Google-Extended, Premium | Premium | [yoast.com llms.txt page](https://yoast.com/features/llms-txt/), [wordpress.org listing](https://wordpress.org/plugins/wordpress-seo/) |
| Rank Math 1.0.282 (4+ million) | Yes, an "LLMS Txt" module under Rank Math SEO, Dashboard; tier not stated on the page | Not listed | Instant Indexing module | [Rank Math knowledge base](https://rankmath.com/kb/llms-txt/), [wordpress.org listing](https://wordpress.org/plugins/seo-by-rank-math/) |
| All in One SEO 5.0.5.1 (2+ million) | Yes, "llms.txt Generator" in the free listing | robots.txt editor, no per-bot feature listed | Yes | [wordpress.org listing](https://wordpress.org/plugins/all-in-one-seo-pack/) |
| SEOPress 10.4 (300,000+) | Yes, "Native llms.txt support", new in 9.8 | robots.txt editor | Yes | [wordpress.org listing](https://wordpress.org/plugins/wp-seopress/) |
| ThinkRank (ours) | Yes, generated for posts, pages and products | Allow or block each AI crawler, per agent | Not listed | [thinkrank.ai](https://thinkrank.ai) |

According to each vendor's own pages, verified on 20 November 2026, every plugin in the table can produce llms.txt. The pass rate tells you how often owners switched it on and kept it valid, not whether the feature exists.

**A correction to our own scanner.** While checking these pages we found that AIScan's fix guidance for C2 gets this wrong. Asked for a Yoast site, it says Yoast "does not generate llms.txt", and for a Rank Math site it says "Rank Math cannot generate llms.txt". Both statements are contradicted by each vendor's own documentation as read on 20 November 2026. We logged it as a scanner bug on 20 November 2026; until the guidance is corrected, use your plugin's own toggle before you reach for a hand-made file. If you would rather build the file by hand anyway, our [llms.txt generator](https://aiscan.site/llms-txt-generator) writes one from your sitemap.

For what goes inside the file and how large it should be, our [census of llms.txt size and structure](https://aiscan.site/blog/llms-txt-size-structure-census) measured 427 files.

## JSON-LD and meta descriptions (check C3): the clearest plugin effect

C3 reads the homepage HTML and records four parts: how many `<h1>` elements, whether there is a `<title>`, a meta description and a JSON-LD block. Our [study of structured HTML on 855 homepages](https://aiscan.site/blog/structured-html-homepage-corpus) found that JSON-LD is the part sites miss most often. On WordPress the plugin decides it almost entirely.

| Detected plugin | Sites | JSON-LD present | Meta description present | Exactly one h1 | No h1 | Two or more h1 |
|---|---|---|---|---|---|---|
| Rank Math | 78 | 76 (97.4%) | 76 (97.4%) | 59 (75.6%) | 14 | 5 |
| Yoast SEO | 63 | 63 (100%) | 56 (88.9%) | 40 (63.5%) | 17 | 6 |
| None detected | 57 | 32 (56.1%) | 31 (54.4%) | 35 (61.4%) | 18 | 4 |
| AIOSEO | 8 | 8 | 7 | 8 | 0 | 0 |
| ThinkRank (ours) | 5 | 5 | 5 | 3 | 2 | 0 |

Every Yoast homepage in the sample printed JSON-LD. Rank Math missed it on two. Without a detected plugin, close to half the homepages had none, and the same share had no meta description. This is what the vendors' own feature lists predict: [schema.org](https://schema.org/) structured data and the meta description are the core of what a WordPress SEO plugin does, and the numbers show that they do it.

The `<h1>` column barely moves. A plugin does not write your heading; the theme does. Rank Math sites had one h1 a little more often than the others, but 14 Rank Math homepages had none at all, and so did 17 Yoast homepages. If your C3 result is partial on a WordPress site with a plugin installed, the cause is almost always the theme's homepage template (a logo in place of an h1, or a hero block that uses a paragraph). Seven Yoast homepages had no meta description; the scan records that it was missing, not why.

## Named AI bot rules (B2) and Content Signals (B1): no plugin effect

B2 passes when robots.txt names AI crawlers individually (GPTBot, ClaudeBot, PerplexityBot, Google-Extended and others) instead of leaving them to the `User-agent: *` group. According to [RFC 9309](https://www.rfc-editor.org/rfc/rfc9309.html), the robots exclusion standard, a crawler obeys the most specific group that matches its name, so a named group is the only way to give one AI crawler different rules from the rest.

| Detected plugin | Sites | B2 pass | B2 fail | B1 Content Signals pass | D1 robots.txt pass |
|---|---|---|---|---|---|
| Rank Math | 78 | 19 (24.4%) | 58 | 9 (11.5%) | 70 (89.7%) |
| Yoast SEO | 63 | 15 (23.8%) | 47 | 4 (6.3%) | 57 (90.5%) |
| None detected | 57 | 17 (29.8%) | 39 | 11 (19.3%) | 42 (73.7%) |
| AIOSEO | 8 | 4 | 4 | 1 | 7 |
| ThinkRank (ours) | 5 | 2 | 3 | 2 | 3 |

Sites with Yoast or Rank Math have a working robots.txt more often (about 90% against 74%), and they are no more likely to name a single AI crawler in it. The no-plugin group even leads on Content Signals, the `Content-Signal:` line Cloudflare writes into managed robots.txt files.

The reason sits in WordPress core. When no physical robots.txt file exists, WordPress answers `/robots.txt` itself through [`do_robots()`](https://developer.wordpress.org/reference/functions/do_robots/), which, as its source on developer.wordpress.org shows, writes `User-agent: *`, a Disallow for the admin path and an Allow for `admin-ajax.php`. SEO plugins add a sitemap line and give you an editor for that file. None of the four large plugins adds named AI groups by default. According to Yoast's own wordpress.org listing, fetched from wordpress.org on 20 November 2026, its per-bot "Bot blocker" for GPTBot, CCBot and Google-Extended is a Premium feature, and it is a blocker: it writes Disallow rules. If you want those crawlers allowed by name, you write the groups yourself.

Our earlier piece on [security plugins that block AI crawlers](https://aiscan.site/blog/wordpress-security-plugin-blocks-ai-crawlers) covers the opposite failure, where a firewall plugin turns AI crawlers away before robots.txt is ever read.

## Markdown copies (check C1): nobody's default

C1 passes when a page offers a Markdown version, either through `Accept: text/markdown` or a `.md` URL. Rank Math sites pass 28.2%, sites with no plugin 26.3%, Yoast sites 17.5%, and none of the eight AIOSEO sites. Among the plugins in this study, only ThinkRank lists a Markdown copy of each post, and it lists it as a Pro feature. The sites that pass C1 mostly get it from somewhere else, typically a CDN rule such as Cloudflare's Markdown for Agents or a dedicated plugin, which is why the column does not follow the plugin label.

## Sitemaps, feeds and the rest

| Check | Rank Math (78) | Yoast (63) | None detected (57) | What a plugin does here |
|---|---|---|---|---|
| D1 robots.txt with a Sitemap line | 70 | 57 | 42 | Adds the sitemap line to the virtual file |
| D2 XML sitemap | 72 | 60 | 48 | Replaces the core sitemap with its own |
| D3 Link header or head link | 77 | 57 | 49 | Core already prints REST API links |
| E5 RSS or Atom feed | 63 | 48 | 46 | Core prints the feed; plugins rarely change it |
| E1 clean 404 for a missing page | 67 | 49 | 50 | Theme and host, not the plugin |

The sitemap checks favour plugin sites by about ten points, which matches what each plugin documents. The feed and 404 rows do not move, because WordPress core and the host handle them.

## One result we cannot explain yet

P4 looks for OAuth discovery files at `/.well-known/oauth-authorization-server` and `/.well-known/oauth-protected-resource`. On these WordPress hosts it passed on 39 of 78 Rank Math sites, 11 of 63 Yoast sites, 5 of 57 with no plugin, and 4 of the 5 ThinkRank sites. The Rank Math passes are spread over 37 different domains and 14 different scan days, so they are not one site scanned many times, and 34 of them return a clean 404 for missing pages, so they are not soft 404s answering every path. The scan stores only the status codes, not which plugin served the file. We are not attributing it to Rank Math until we can see the response, and it does not change any other result in this article.

## Should you switch SEO plugins for AI readiness?

On this evidence, switching between Yoast and Rank Math for the sake of AI agents buys you very little. The two large groups sit 1.9 points apart on mean score, and close on most checks. The larger gaps are llms.txt (Yoast ahead by 11 points), C3 and Markdown (Rank Math ahead by about 11 points each). The llms.txt and Markdown gaps come from what owners switched on, and the C3 gap comes mostly from the h1 and the meta description, which depend on the theme and on whether the owner filled in the homepage description. A migration costs a weekend of redirects, settings and schema checks. Turning on the llms.txt switch in the plugin you already run costs a minute.

Switching makes more sense in three situations:

1. **You have no SEO plugin at all.** This is the group that loses the most: 25 of 57 homepages without JSON-LD and 26 without a meta description. Any of the plugins in this study fixes those two parts of C3 on install.
2. **You run two SEO tools that fight over the same files.** A site with one plugin for titles, another for schema and a snippet plugin for robots.txt has three writers for files that need one owner. Consolidating into one plugin, whichever it is, removes the conflicts.
3. **You want per-agent robots.txt rules from a settings screen rather than a text editor.** Among the free tiers we read on 20 November 2026, only ThinkRank lists that; Yoast lists it as Premium and as a blocker; the others give you an editor and leave the groups to you.

If none of those applies, keep your plugin, run the scan, and fix the three or four checks it names. Our [state of AI agent readiness study](https://aiscan.site/blog/state-of-ai-agent-readiness-2026) points the same way across all platforms: the checks that separate the upper levels (an API catalog, an MCP server card, Markdown copies) are files and headers an owner adds once, not features a platform switches on.

## What to do with your plugin, step by step

**Start with a scan.** Run `npx aiscan-cli yoursite.com` in a terminal, or paste your homepage at [aiscan.site](https://aiscan.site/), free and with no account. The report detects your SEO plugin and grades C2 (llms.txt), C3 (title, meta, h1, JSON-LD), B2 (named AI bot rules), B1 (Content Signals), C1 (Markdown), D1 and D2. Everything below maps to one of those checks; the [content checks page](https://aiscan.site/docs/checks/content) and the [WordPress platform page](https://aiscan.site/docs/platforms/wordpress) explain each in detail.

**If you would rather run one plugin for all of it, ThinkRank is the one we build.** [ThinkRank](https://thinkrank.ai) generates llms.txt from your live posts, pages and products, writes per-agent Allow or Block groups into robots.txt from a settings screen, and handles titles, meta and schema in the same plugin. The practical reason it fits this problem is that it replaces the "one plugin for SEO, a snippet for robots.txt, a file upload for llms.txt" stack with one plugin, and it imports existing settings from Rank Math, Yoast, AIOSEO and SEOPress, showing what it found before it writes anything. It is ours, so weigh that; the steps below work without it.

| You run | C2 llms.txt | B2 named AI bot rules | C3 if partial |
|---|---|---|---|
| Yoast SEO | Turn on llms.txt in Yoast's site features (free) | Yoast SEO, Tools, File editor: add the groups below | Check the theme's homepage h1 |
| Rank Math | Rank Math SEO, Dashboard, enable the LLMS Txt module | Rank Math SEO, General Settings, Edit robots.txt (Advanced Mode) | Check the theme's homepage h1 |
| AIOSEO | Its llms.txt Generator | Its Robots.txt Editor | Usually passes already |
| SEOPress | Its native llms.txt setting | Its robots.txt editor | Check the theme |
| No plugin | Upload a file from the [generator](https://aiscan.site/llms-txt-generator) | Upload a physical robots.txt | Add a plugin or theme schema; this is where you lose the most |

The block to paste for B2, if you want these crawlers allowed by name:

```txt
User-agent: GPTBot
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Google-Extended
Allow: /
```

Change `Allow` to `Disallow` for any crawler you want to keep out. Then scan again: B2 should pass on the next run. For the full setup on WordPress, our [AI readiness setup guide](https://aiscan.site/blog/ai-readiness-setup-guide) walks through every check, and the [platform league table](https://aiscan.site/blog/platform-league-table-agent-readiness) shows where WordPress stands against Next.js, Astro and Shopify.

## What AIScan cannot see here

The scan sees the output, never the settings. It cannot tell you that your plugin's llms.txt toggle is off, only that `/llms.txt` returned 404. It does not record which plugin wrote a robots.txt group, so a group added by hand and one written by a plugin count the same. And it grades the homepage plus a handful of fixed paths, so a plugin that adds schema only to posts, or only to products, will not show here.

## Common mistakes we see in these results

1. **Assuming the plugin did it.** 29 Rank Math and 23 Yoast sites in the sample had no llms.txt although their plugin can produce one.
2. **Treating a working robots.txt as an AI policy.** About nine in ten plugin sites pass D1; about one in four names an AI crawler.
3. **Blaming the plugin for a missing h1.** The theme writes it. 31 homepages with Rank Math or Yoast had no h1 at all.
4. **Stacking two SEO plugins for one feature.** Two plugins writing robots.txt or JSON-LD on the same site produce two answers; pick one owner for each file.
5. **Trusting fix guidance that predates the feature.** Ours did, as the llms.txt section above explains.

## The next action

Scan your WordPress site at [aiscan.site](https://aiscan.site/) or with `npx aiscan-cli yoursite.com` and look at four checks first: C2 (llms.txt), B2 (named AI bot rules), C3 (structured HTML) and C1 (Markdown copy). Fix C2 with your plugin's own switch, fix B2 with the robots.txt block above, then scan again to confirm. The rest of the checks, ordered by how much they move the score, are on our [guides page](https://aiscan.site/guides).

