Docs / AEO Ultimate for WordPress

What it does

AEO Ultimate makes a WordPress site readable, trustworthy and citable for AI answer engines, then shows you whether it is working. It publishes one connected @graph of JSON-LD schema, a clear publisher and author identity (E-E-A-T), and the discovery files AI agents look for, such as llms.txt. It controls which AI crawlers may read your site. It asks the AI engines your customers’ questions and records whether they cite you. And it finds the answers your site is missing, often in documents you already have.

It runs alongside Yoast SEO, Rank Math or AIOSEO without duplicating what they already output. Your SEO plugin keeps doing titles, meta descriptions and redirects; AEO Ultimate adds the layer that AI answers are built from.

Built for SEO professionals

• Free: no license, no paid tier, no feature locked away.
• Bring your own key: AI features run on your own Anthropic, OpenAI, Google, Perplexity, xAI or Mistral key. You pay the provider directly, at their rates, and choose the model.
• Bulk: fill every missing alt text, social description and FAQ, Service or Contact schema across a site in one pass, and generate product schema for a whole catalogue.
• Semi-automated: the plugin detects what each page is, adds schema only where it is missing, and suggests the E-E-A-T details you have not filled in. AI-powered bulk actions show the number of billed calls and ask before running, so there are no surprise bills.
• Modular: 29 modules, each switched on or off on its own, so a site runs only what it needs.
• Client-ready: Access & Roles lets an SEO manager or a client’s marketing team use the tools without being a full administrator, and the CSV export gives you the raw data for your own reporting.

The signals it builds and monitors

• Structured schema markup: present, correct for the page type, and connected into one graph
• E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness): who publishes the site and who wrote each page
• Entity identity and clarity: your business, people, products and services as entities linked to Wikipedia and Wikidata
• Content freshness: dates an engine can read and a visible Last Updated line
• Social graph coverage: Open Graph and Twitter/X Cards on every page
• AI bot access: which crawlers may read, cite or train on your content, and what a no-JavaScript crawler actually sees
• AI citations: whether eight AI engines (ChatGPT, Gemini, Claude, Perplexity, Grok, Le Chat, Muse (Meta AI) and DeepSeek) cite you, name you without a link, or leave you out, and who they cite instead
• AI traffic: which AI bots crawl your pages, and which AI engines send you visitors
• Content gaps: how AI retrieval cuts your pages into passages, which answers your site is missing, and which of them your own documents already hold

New in 2.28: eight AI engines, and where the buyer searches from

Version 2.28 adds two AI engines to AI Visibility, lets every run search from a real place, and makes the RAG Engine’s suggestions sharper.

• Muse (Meta AI): Meta’s assistant, through the Meta Model API with web search. Muse answers slowly, so its questions are sent in the background and collected as they finish; the rest of the run never waits for it.
• DeepSeek: through DeepSeek’s API with web search. DeepSeek support is new: it works, and is still being tested. Most of DeepSeek’s use is outside the US, led by Asia, with Russia and India near the top, and much of India searches in English. If your buyers are mostly in the US you can leave it off; if you sell internationally, especially into India, it is worth tracking.
• Search location: Each run also searches from a place: automatically your business address (Local Pack) or store country, or any country, state and city you choose, so a Tampa shop is measured the way Tampa buyers see the answers. Engines whose APIs take a search location use it; the others are told where the user is. Runs are only compared with runs from the same place.
• Gave no sources: an answer that did not cite you and cited nobody else is marked as such, per engine and per answer, so “not cited” is never mistaken for a competitor winning.
• Questions about your business: in the RAG Engine’s Next questions, follow-ups about you (turnaround, warranty, shipping, rush service, reviews) are grouped by topic with the page that should answer them, or a note that you have none yet.
• Pages you still rank for: Search Placement flags searches whose page is now in the Trash or unpublished, with a one-click restore.
• Markdown versions of your pages at /page.md, and FAQ schema drift detection when a page’s visible FAQ and its FAQPage schema no longer match.

New in 2.26: buyer journeys in AI Visibility

People don’t ask an AI assistant one question and leave. They ask as themselves, with the assistant knowing something about them, and they keep going: the first answer leads to the next question. Version 2.26 brings both of those into your AI Visibility checks.

• Buyer personas: ask a run as one of your buyers, such as “machine shop owner”, “manufacturing engineer” or “first-time homeowner”. The persona reaches the engine as background about the user, the way the consumer apps personalise their answers, and your question is sent exactly as written. Type your own personas, or let the plugin suggest them by reading your main pages. If you know your buyers better than any page does, type them in; if you would rather not use personas at all, switch them off and every run asks as nobody in particular.
• Follow-up questions: every answer comes back with the questions that engine expects the buyer to ask next, predicted by the same engine that answered.
• Follow the conversation: optionally ask the engine its own top follow-up, in the same conversation, one or two turns deep. You see whether you are still cited further into the buyer’s research, or whether you appear in the first answer and disappear by the second. Each question’s whole conversation finishes before the next question starts, so a run you stop halfway still holds complete conversations.
• CSV export: a run, or every run you keep, in one file: each question, its full answer and verdict, followed by its follow-up questions and any later turns. Every row carries the persona, the site and its business type, plus pattern columns worked out for you: where a brand or subject falls in the question, model numbers, places, and whether the question is looking for a business to hire or for an answer. Exports from many client sites stack into one sheet, ready for a spreadsheet or for your own AI analysis.
• Next questions in the RAG Engine: the follow-ups the engines predicted, checked against your pages and your documents. Where a document answers one, you get the passage to publish; where nothing on the site answers it, you get it on a list of content to write. A follow-up that the conversation went on to ask shows what each engine then did, for example “When asked: Gemini left you out”.

AI Visibility: are the engines citing you?

AI Visibility asks eight AI engines (ChatGPT, Gemini, Claude, Perplexity, Grok, Le Chat, Muse (Meta AI) and DeepSeek) the questions your customers ask, on your own API keys with each engine’s web search switched on, and records what each one said. It is the closest thing to measuring whether any of your AEO work is paying off.

The questions

Questions are built from data the plugin already holds, so runs stay comparable week to week and engine to engine: your company and its policies, the brands you track, your categories and product types, individual products with their variants and prices, and your posts. You see the full list before anything is asked, untick what you don’t want, and can have awkward template phrasing rewritten into natural questions. You decide how many questions each level gets with the Light, Standard or Deep presets, or set each number yourself.

What each answer is scored on

• Cited: the answer linked to something you own, your own domain or a profile you registered, such as your LinkedIn company page. The board shows which surface was cited, so “my site got picked up” reads differently from “only my profiles did”.
• Named: the answer mentioned you without a link. The exact passage where it named you is shown, so you see how you were described, not just that you were.
• Absent: neither. The board shows who was cited instead.
• Unavailable: no answer was recorded, for example a refused or failed call. It is never counted as absent.
• Accuracy: where the plugin knows the true answer (your refund and shipping policies, where you ship, variant options, prices, stock), it checks whether the engine got it right, got it wrong, or didn’t say.

What you do with the results

• Who gets cited instead: every domain the engines cited in your place, per engine and per scope (the whole site, a category, a brand, your posts). Seeing who gets named tells you what they publish that you don’t.
• Questions you are not answering: every question where you were left out becomes a brief of what to publish, with a link to the page it belongs on.
• What counts as you: register the profiles and other sites you own, and track product lines or sub-brands in their own right, each with the phrases to watch for and the websites and profiles it owns.
• Trends: cited share per engine across your last runs, and each tracked brand over time. Runs are compared only with runs asked the same way, so a persona run never shows up as a gain or a loss against a baseline run.
• Cited here, visited from there: with Google Analytics connected, AI referral visits sit beside the citation results, so you can see whether being cited is sending people.

Cost and control

Every check is a call to the engine on your key, so the page shows the number of checks a run will make before you start, never a dollar figure, and a daily cap pauses a run instead of letting it run on. Runs keep going in the background if you close the page, one check at a time, and resume where they stopped. An engine that rejects its key is skipped for the rest of the run instead of failing on every question. “Load sample data” builds a labelled sample from your own catalogue, so you can see the board before a single call is spent; sample data never enters a trend or an export.

The RAG Engine: find the answers you’re missing

AI answer engines don’t read your pages top to bottom. They cut the web into short passages, pull back the few that best match a question, and write their answer from those (retrieval-augmented generation). Most businesses already have the answers those engines are looking for, in spec sheets, manuals and FAQs that never made it onto the website. The RAG Engine is a content gap finder built on that fact: it finds those answers, shows you which page is missing each one, and gives you the passage to add.

• Documents: upload PDFs, Word files and text. They are read on your own server and the file is deleted once read; only the searchable passages are kept. Large files upload in pieces and resume after a dropped connection. Tables in PDFs come out as real tables, not scrambled symbols.
• Document labels: each document is marked as a law, code, safety data sheet, contract, manual, spec sheet or other, with its place, edition and what to cite it as. You can correct any label.
• Search Placement: the Google and Bing searches where you rank 4th to 15th that your documents answer and the ranking page does not, with the page to add the answer to. These are your content gaps, taken from your own ranking data rather than guessed. Searches for a business to hire (“near me”, “services”, a town) are set aside, because a document cannot answer them.
• AI Citations: questions from your AI Visibility runs where an engine left you out, beside the passages from your documents that answer them.
• Next questions: the follow-up questions the AI engines expect your buyers to ask next, checked against your pages and documents. Where your documents have the answer you get the passage to publish; where nothing does, you get the content to write.
• Laws, codes and safety data: quoted word for word with their citation, and checked online for changes since your copy’s date. Only the citation is sent for that check, never your document’s text.
• Chunk View: any page split the way a retrieval index holds it, with the passages that fall apart on their own flagged.

Nothing is published for you: every suggestion comes with a Copy button and an Edit page link, and the writing stays yours. Switch it on under Modules → RAG Engine. Read more about the RAG Engine.

Schema and the knowledge graph

The core engine detects the schema already on each page, classifies the page (homepage, service page, article, location page or product), and adds JSON-LD only where it is missing: Organization, Article, NewsArticle, Person, FAQPage, Service, Product, LocalBusiness, ContactPoint, Event, Review, AggregateRating and BreadcrumbList.

Everything on a page is woven into one connected @graph: WebSite, Organization and WebPage cross-referenced by @id, with the author, the products and the services attached to them. An answer engine reads your site as one set of related facts instead of isolated tags: this person wrote this article for this organization, which offers this service. The Knowledge Graph screen defines your entities and links them to Wikipedia and Wikidata.

• Schema Advisor evaluates each page’s schema against its intent and flags what is missing or wrong, without false warnings.
• FAQ, Service and Contact detectors find FAQ sections and accordions, service-oriented pages and contact pages that lack their schema, and can add it, one page at a time or in bulk.
• Schema conflicts: when your SEO plugin and AEO Ultimate output the same type, you see a field-by-field comparison and choose, per type, to keep both or to switch the other plugin’s version off through its own settings hook.
• Bot View fetches any page the way no-JavaScript AI crawlers such as GPTBot, ClaudeBot and PerplexityBot do, and reports which schema and content they can see and which is locked behind JavaScript: tag-manager schema, review widgets, client-rendered apps.

AI Ready: llms.txt and crawler control

llms.txt and llms-full.txt, Markdown versions of your pages (at /page.md, and for AI agents that ask for them), Content-Signal headers, and agent-discovery endpoints: the Agent Skills Index, MCP and WebMCP, OAuth/OIDC discovery, an RFC 9727 API catalog and RFC 8288 Link headers. Your site is also published as an Open Knowledge Format bundle at /okf/, linked from llms.txt.

You decide which AI systems may read, cite or train on your content, with per-bot blocking and rate limits for more than 20 AI crawlers, including GPTBot, ClaudeBot, PerplexityBot and Google-Extended. The AI Prompt Rate Limiter protects your own API budget from bots hammering AI features on your site. Where creating a /.well-known/ file needs server access the plugin doesn’t have, you get copy-and-paste instructions instead of a silent failure.

Company and author identity (E-E-A-T)

Answer engines prefer sources they can identify. The plugin publishes Organization JSON-LD with your legal name, logo, contact points and sameAs profiles, a clean split between article:publisher and article:author in Open Graph, and a sitewide fb:app_id for Meta Domain Insights, with one-click import from Yoast or Rank Math.

Author profiles are published as Person schema with job title, credentials and certifications, areas of expertise, years of experience and social profiles, all edited on the user’s profile screen. An E-E-A-T scorecard covers both your authors and your company and suggests the missing details, which you can apply for yourself. Optional author boxes and hover cards show the same credentials to readers, and a visible “Last Updated” line gives engines and readers an explicit freshness date.

WooCommerce

Every commerce feature exists so an answer engine can quote your catalogue accurately.

• Product schema: every product publishes a connected Product or ProductGroup graph with per-variant offers, identifiers (GTIN, MPN, SKU), images, reviews, shipping from your live zones and your return policy.
• Extension-aware pricing: products from Bookings, Subscriptions, Bundles and similar extensions publish “From” prices or ranges, never a price their own page contradicts.
• Product FAQs: every FAQ source on a product page assembled into one FAQPage node, with free and AI generators for the questions shoppers ask.
• Products dashboard: walks your whole catalogue, names exactly what each product is missing, and generates schema in bulk.
• Google and Bing Merchant Center: compares your catalogue with what each engine lists, surfacing missing products, rejected items and price or availability mismatches.
• Smart Collections: publishes “Starter Kit” or “Under $500” style groupings as real collections with structured data an assistant can quote when someone asks what they need to get started.
• Promotions and price rules: quantity breaks and member prices published as honest price specifications, with coupon codes sent to Merchant Center.
• AI image alt text: fills missing product image alt text in bulk.

WooCommerce’s own Product schema is replaced only with your consent. For stores that want more, AEO Ultimate for WooCommerce is a separate commerce edition; when it runs alongside, this plugin stands down so nothing publishes twice.

Local, sitemaps and indexing

• Local Pack: LocalBusiness schema with a predictive business-type picker, opening hours, service area and geo coordinates, plus a NAP (name, address, phone) audit against Google Knowledge Graph, Google Business Profile and Bing Maps, and the directories your industry should be listed in.
• Sitemaps: a cached XML sitemap and a Google News sitemap with NewsArticle schema. If Yoast, Rank Math or AIOSEO already provide a sitemap, you are told, so you don’t end up with two.
• IndexNow: Bing, Yandex and other IndexNow engines are told the moment you publish or update. The verification key is generated and hosted for you, and local and staging sites are never submitted.
• Not Indexed: finds published pages missing from Google’s index and flags the likely cause, including duplicate meta descriptions.
• Smart 404 Rescue: ignores bot probe traffic, sends real visitors to the closest published page, corrects broken internal links at their source with an undo log, and turns Search Console 404s for moved pages into permanent redirects. No per-404 logging, so no database bloat.

Images and social cards

One screen manages Open Graph and Twitter/X cards across every page: titles, descriptions, share images, card type and creator, with twitter:creator filled from your author profiles. Descriptions and alt text can come from content you already have or be generated with AI, one page at a time or for every page that is missing them.

Command Center: measure it

Google Search Console, Google Analytics 4 and Bing Webmaster Tools in one view, with Google PageSpeed audits (Lighthouse score and Core Web Vitals).

• AI Crawler Watch logs which AI bots read your site, which pages they fetch and what they waste requests on.
• AI referral tracking counts the human visits that arrive from ChatGPT, Perplexity, Claude, Gemini and Copilot, per page and per engine, with no cookies and no per-visitor data.
• Funnel Audit connects that AI attention with your internal links. AI engines tend to cite informational posts rather than the pages that sell, so it finds money pages that are buried or orphaned, generic anchor text, missing schema on conversion pages, and cited posts with no path to a page that converts.
• A traffic-leak scan finds high-traffic pages that lose mobile visitors to slow loads, and a content scan flags inline scripts, base64 images and other page-builder leftovers.

More tools

• AEO Score: your site graded from 0 to 100, with every lost point linked to its fix.
• Reviews: collect and display customer reviews with AggregateRating and Review schema, scoped automatically to your LocalBusiness, products, services or organization.
• PR Bridge AI: turns a press-release post into AP-style copy and sends it to wire services, with a media report for tracking distribution.
• Access & Roles: give marketing staff or clients access to the tools without making them administrators.
• Diagnostics: a screen that records errors, including fatal ones, so a problem on a live site can be found instead of guessed at.

All 29 modules

Each is switched on or off under TWT AEO → Modules. For a walk through every screen, see Every screen.

Your keys, your models, your data

Schema, llms.txt, crawler control, sitemaps, E-E-A-T and the RAG Engine’s document search all work without an API key. The AI features (AI Visibility, AI writing, image descriptions, document labels and the law check) run on your own key for the provider you choose, and you pick the model for each provider under Settings → AI Models, including models newer than the plugin.

Nothing is sent anywhere until you configure a feature and use it. Uploaded documents are processed on your own server and the file is deleted once read. Every outside service the plugin can contact, what it sends and the provider’s terms are listed on the WordPress.org page.

Works alongside your SEO plugin

AEO Ultimate is designed to run with Yoast SEO, Rank Math, AIOSEO and WooCommerce. It detects the SEO plugin you use, reads what it already outputs, and adds schema only where it is missing. Where both would output the same type, you choose which one wins, type by type, and a module that overlaps with what your SEO plugin already does can simply be switched off. You stay in control at every step.

Three ways to set up

1. Autopilot (recommended): AEO Ultimate configures itself from your existing site data. It detects FAQ schema, imports your company information and social profiles, and pre-fills your author box. You review what will run, then click Go.
2. The Basics: two quick steps for your business name and contact details, social profile URLs, your Google Analytics 4 measurement ID, IndexNow and the XML sitemap.
3. Expert: full control, step by step: complete company identity, every social platform, author schema and credentials, module selection and API keys.

After activating, go to Settings → Permalinks and click Save Changes once so /llms.txt and the sitemap resolve. The getting-started guide walks through the rest.

Requirements

Go deeper

• Every screen: what each screen does and when to use it.
• FAQ: API keys, compatibility, performance and privacy.
• Troubleshooting: schema not appearing, caches, permalinks and the Diagnostics screen.
• Source on GitHub and the full changelog.

Support is available through WordPress.org, the full documentation, and via Slack, WhatsApp, or email.