AEO Ultimate for WooCommerce
Shoppers ask an assistant. The assistant names one merchant.
When ChatGPT, Claude, Gemini, Copilot or Google’s AI answers a shopping question, it doesn’t return ten links. It names one store — and it picks that merchant from machine-readable signals, not marketing copy.
AEO Ultimate for WooCommerce publishes those signals from data your store already has. No new content to write, no feed to maintain by hand.
For sale
Version 1.0.3, sold by Tampa Web Technologies. Everything about the plugin is on this page: the documentation, twelve video tutorials, what it sends where, and every change in every release.
30-day money-back guaranteeA full refund within 30 days of your date of purchase.
Version 1.0.3
WordPress 6.9+
WooCommerce 8.0–11.2
PHP 7.4+
HPOS & checkout blocks
GPL-2.0+
New in 1.0.3
Ready for AI shopping agents, not just AI answers
AI assistants no longer only recommend a store; they shop in it. Shopify gives every store a built-in door for them. WooCommerce does not, so AEO Ultimate checks the doors you open yourself, and can let agents check out through the two protocols they use: Google’s Universal Commerce Protocol (UCP) and the Agentic Commerce Protocol (ACP) used by ChatGPT.
Agentic Checkout: UCP and ACP from one core
Off until you switch it on. Every price, tax line, discount and shipping rate comes from a real WooCommerce cart and your live shipping zones. A total that does not reconcile to the cent is escalated to the buyer, never adjusted.
Paid through your own Stripe account
With the WooCommerce Stripe gateway set up, UCP agents pay with Google Pay and ACP agents with Stripe Shared Payment Tokens, charged for exactly the total shown. The order lands in WooCommerce like any other. No card data passes through the plugin.
A safe hand-off otherwise
Without Stripe, or when a card needs 3-D Secure, the agent sends the buyer to your own checkout with the cart already filled.
ACP only with a key
Nothing ACP is published until you create an agent key on the Agent Ready page. Keys are stored as hashes and shown once. If another UCP or ACP plugin is active, the module will not switch on, and names it.
Agent Ready: can an agent get in and buy?
Checks what closes the door: Coming soon mode, robots.txt rules against search and shopping agents, firewall plugins, the Store API, a Merchant Center feed, policy pages and shipping zones.
Discovery files checked against the spec
Fetches /.well-known/ucp and /.well-known/acp.json the way an agent would and checks them against each protocol’s own spec: HTTPS, no redirects, caching headers, required fields.
Inventory an agent can trust
Stock management off, products not tracking stock, variations sharing one count, parents contradicting their variations, paid orders that never reduced stock: every affected product listed.
Agents that sign their requests
AI agents using Web Bot Auth are recognised in AI Traffic, Store API carts included. Nothing is blocked.
Guides: Agent Ready · Agentic checkout
For store owners and reviewers
Everything about the plugin, in one place
Every link below is public. Nothing here needs an account.
Documentation
The WooCommerce documentation hub: setup, every module, and how each score is worked out.
Video tutorials
Twelve walkthroughs recorded on a real WooCommerce store, each with a written guide and a transcript.
Every screen
A tour of every admin screen in the plugin and what it does.
FAQ · Troubleshooting
Common questions, and what to check when something does not show up.
External services
Every outside service the plugin can contact, what it sends, when, and the provider’s terms and privacy policy.
Privacy and data · Changelog
What the plugin stores, and every change in every release.
Plugin facts
Version
1.0.3
Requires WordPress
6.9 or later (tested to 7.1)
Requires WooCommerce
8.0 or later (tested to 11.2)
Requires PHP
7.4 or later
HPOS (custom order tables)
Declared compatible
Cart & Checkout blocks
Declared compatible
Licence
GPLv2 or later
Usage telemetry
None
Updates
From our licence server, once a key is entered
Publisher
Tampa Web Technologies
One connected @graph, not a pile of blocks
Every product page emits a single connected graph instead of competing JSON-LD blocks — all cross-referenced by @id so a crawler reads your store as one web of facts.
WooCommerce’s own duplicate Product and breadcrumb markup is suppressed automatically, so you ship one clean block.
What one product page emits
OnlineStore
WebSite → WebPage
Product or ProductGroup + variants
Offer / AggregateOffer
BreadcrumbList
The details that decide whether you’re quoted
Product data comes from core WooCommerce, correctly — which mostly means refusing to publish things that aren’t true.
An empty price is not a price of nothing
A product with no price set publishes no offer at all, rather than an offer saying it costs zero.
The price a shopper actually sees
Tax-inclusive figures, so schema matches your storefront and your Merchant Center feed.
Backorder is a real answer
All three stock statuses map properly — including on backorder, which most plugins flatten.
Variants keep their range
Variable products emit an AggregateOffer with a real low and high price, not a guess.
“Will it fit?” is answerable
Weight and dimensions publish as measured quantities with your store’s unit code — no reading the description.
Shipping from your live zones
Rates and destinations read from real WooCommerce shipping zones, and your returns page is linked as the return policy.
The whole gallery, captioned
Angles, packaging and dimension charts publish as ImageObject nodes. Filename-derived titles like “image-1235” are dropped — they look like metadata while saying nothing.
Identifiers from the right field
GTIN comes from WooCommerce’s own product identifier field, and voice passages are marked speakable.
Your schema never goes stale
Structured data is rebuilt from live WooCommerce data on every page load — never frozen into a saved copy. Change a price, sell the last unit, add a variation or edit a shipping zone, and the published schema follows immediately.
Merchant Center parity, both engines
Compares your catalogue against what Google and Bing actually list — surfacing missing products, rejected items and price mismatches, with an integrity score.
Everything it does
Each feature, as the plugin’s own readme describes it. Open any one for the detail.
Store Trust scorecard
Every online store handles payments and personal data, which puts it in Google’s strictest “Your Money or Your Life” category — trust gaps do not merely rank lower, they are held back to protect shoppers. The scorecard audits the three pillars that matter for commerce:
- Operational transparency — returns, shipping, privacy policy, contact details, identity
- First-hand proof — reviews enabled, and how many come from *verified* buyers
- Machine-readable trust — your product graph, Merchant Center parity, and entity corroboration
Entity corroboration, matched to what you sell
sameAs is how a search engine tells your business apart from a similarly named one. Pick your store type and the checklist adapts: a B2B wholesaler is asked for Thomasnet and Kompass, a beauty brand for Leaping Bunny, a hybrid store for Apple Maps and Nextdoor. Around thirty directories, split into the handful worth doing first and the rest, each explaining what it actually buys you and linking straight to where you claim it.
Three things it does that a plain link list does not:
- A credential is not an identity. Leaping Bunny and USDA Organic publish as
hasCertification, which states the claim precisely, rather than being buried in a URL list. - A D-U-N-S number is not a URL. It publishes as the
dunsproperty. InsameAsit would simply be invalid. - A mistyped profile is worse than an empty field. A Facebook URL in the BBB box claims an unrelated page *is* your business, so a URL that does not match its directory is flagged and can never read as a pass.
Editorial round-ups and category pages are deliberately absent from the list. sameAs means a page that unambiguously identifies *you* — linking a magazine’s “best of” article asserts that article is your business, which is the opposite of what you wanted.
Command Center: buyer intent
Connect Google Search Console and every search your store appears for is sorted by how close the shopper is to buying — ready to buy, comparing options, researching, or looking for you by name — with a playbook of the structural signals that decide each group, and specific findings such as purchase-intent queries sitting just off page one.
AI product enrichment
Generate meta descriptions and social descriptions for products from their real attributes, category and pricing — not generic filler. Bulk-fill everything that is missing, using your own Claude, OpenAI or Gemini key.
Uploading a CSV of 1,500 products never blocks: imported products are queued and filled in gradually in the background at a pace you choose, with a daily limit so costs stay predictable.
Missing image alt text is written by looking at the image, not by parsing its filename — so a photo uploaded as image-1235.png still gets a description of what is actually in it. Two tiers, and the trade-off is stated rather than implied: standard is cheap enough to run across a whole catalogue but approximates fine detail and can miss a small logo, while precision resolves material, weave and subtle branding from difficult photography at noticeably higher cost per image. Rather than quote prices that go stale the moment a provider reprices, each tier links to that provider’s live pricing page and tells you how many images the run will actually send.
Content Uniqueness: is your product copy actually yours?
If your descriptions came from the manufacturer, every other shop selling that item says exactly the same thing — and a search or answer engine has no reason to pick you over any of them. This finds the manufacturer’s own text, measures how much of yours is theirs, and shows you the passages you share, word for word.
You confirm each brand’s manufacturer website once, not once per product. A 2,000-SKU catalogue usually has about twenty brands, and every result is then required to come from a confirmed domain, which is what keeps resellers and marketplace listings out of the comparison.
There are two ways to get their copy, and the cheap one comes first. If you resell, you probably already have the manufacturer’s product feed — a CSV with the barcode, part number and their own description. Import it and the work is done in one file read: no AI, no waiting, no cost, and their actual copy rather than an approximation of it. Nothing is written until a dry run has told you exactly what would change, including how many products already carry a source that would be replaced. If you have no feed, the plugin can search the manufacturer’s site for each product instead — that one is a billed AI request per product and takes about a minute each, so it runs in the background, capped per day, with the count and the estimate shown before you start.
Every figure carries its provenance, because a percentage without one invites you to trust the wrong thing. Text read from the manufacturer’s own markup is evidence; a summary of that page is not, and can only understate how much was copied — so the two are labelled differently and a product that could not be read is shown as unread rather than scored as original.
Nothing is rewritten on your behalf. Where copy is duplicated you can ask for a rewrite, and it lands in a review for you to approve or discard, never in the product.
The structured data moves the opposite way to the prose. Every shop selling an item *should* publish the same barcode, the same material, the same dimensions — that is what a shopping agent extracts and what Merchant Center matches on — while nobody should publish the same paragraph. So confirmed matches can fill in identifiers and specifications you are missing, one tick at a time. A value that disagrees with yours is shown side by side and never written, because that means either the match is wrong or your catalogue is, and both are yours to settle. A measurement stated in a unit your store does not use is refused rather than converted, because a silently wrong conversion looks exactly like data.
AI crawler log, and a test for when it is empty
See exactly which AI assistants are reading your store — OpenAI, Anthropic, Google, Perplexity, Apple, Meta, Amazon and more — which products they favour, and what HTTP status your site returned to them.
An empty log is ambiguous on its own, so there is an active test beside it. It requests your homepage as each major AI crawler in turn and reports what your server actually returned, which separates “no assistant has visited yet” from “my firewall is turning them away”. A blocked crawler is named along with the likely cause — Cloudflare Bot Fight Mode, Wordfence, a host rate limit — and the exact user-agent string to allow. Where a pass proves less than it appears, the result says so instead of showing an unqualified green tick: Brave states that its crawler deliberately does not advertise a distinct user agent, so a pass there proves a Brave-identifying request is not blocked and cannot prove an unbranded one gets through.
AI Traffic: crawls, retrievals, and people who arrived from an answer
Three completely different things get bundled together as “AI traffic”, and telling them apart is the entire point. A crawler indexing your catalogue overnight, an engine fetching a product page *while somebody is asking a question*, and a person clicking through from the answer they were given are separate events worth separate amounts. Each is counted and labelled on its own:
- Crawls — indexing and training passes. Worth knowing, worth little on their own
- Live retrievals — an engine read the page in response to someone’s question. This is the signal that matters: your page was in the running to be part of an answer
- Citations — a person arrived on your store from an AI answer
- Excluded — everything the plugin refuses to count, shown to you rather than quietly dropped
Every crawler claim is checked against the operator’s own published address ranges. OpenAI, Anthropic, Perplexity and Google each publish the IP ranges their bots use. A request wearing a GPTBot user-agent that did not come from OpenAI is reported as impersonation, not counted as interest in your store. Where an operator publishes no list the result says unverified rather than guessing, and a download that fails keeps the previous copy — a stale list must never start accusing legitimate crawlers.
Shopping agents that sign their requests are recognised. Browser agents such as ChatGPT’s drive a real browser and send an ordinary browser user-agent, so by name they look like a person. Those that sign each request (Web Bot Auth, RFC 9421) are verified against the keys the agent publishes at its own domain and logged as that agent — including when they add to cart through the WooCommerce Store API. A signature that fails is logged as excluded. Nothing is ever blocked or served differently: this is for your data, not a gate.
It also never counts itself. Running the crawler reachability test, or asking an engine whether it cites you, both cause real requests to arrive at your store. Both are recognised and held out of every total, with the number shown so you can see exactly what was excluded and why.
“Did ChatGPT send me that customer?” — an honest answer
Nobody can tell you that with certainty, and a plugin that says it can is selling you a number it invented. AI answers usually arrive with no referrer at all.
What *can* be measured is whether an engine fetched one of your pages and a person landed on that same page shortly afterwards — and, far more importantly, how likely that was to be coincidence. So every attribution carries a confidence tier and shows its working.
On a quiet page, a retrieval followed 74 seconds later by a visitor is genuinely informative. On a page with steady traffic, the same 74-second gap means nothing, and it is reported as Unattributed instead of being dressed up. Identical evidence, opposite conclusion, because the baseline rate is measured per page rather than assumed. A visitor who does arrive carrying an identifiable AI referrer is the one case that reaches Confirmed — inference alone is capped below that tier deliberately and can never be promoted into it.
Periods are comparable without being misleading: a part-finished week is compared against an equal span immediately before it, not against a whole previous week, so a report you open on a Tuesday does not manufacture a decline. Growth from zero is reported as such rather than as an invented percentage.
AI citations: were you actually used, not just fetched
The traffic view above watches what engines and people actually do on your site. This is the other half of the same question, asked from the opposite direction — and both live on one screen, with the free observational half shown first.
The crawler log proves a bot read your pages. It cannot prove any of them were *used* in an answer. This asks the engines directly — your buyer questions run against the web-grounded models of Perplexity, ChatGPT and Gemini, recording whether your domain came back in the citation list the engine itself returned.
Two signals are recorded and never merged, because they mean different things. Cited means the engine listed your site as a source. Named means your brand appeared in the text, which a model can do from memory without retrieving you at all. Only the first is evidence.
There is no “share of voice” percentage, and that is deliberate. Answer engines are non-deterministic and personalise results, so a figure from a small sample would imply a precision the method cannot deliver. You get the raw count, the sample size and the date — “cited in 3 of 12 checks” — and because every observation is stored, each question carries a trend line showing whether it is getting better or quietly stopped being cited a month ago.
Every check is a billed call on your own key, so the module is off by default, capped per day, never runs on the front end, and the button tells you exactly how many calls it will make before you press it.
What the cited pages have that yours doesn’t
When a question comes back uncited, the report fetches the pages the engine *did* cite and compares them with yours: schema types, content depth, FAQ presence, meta description, named author, heading structure. No AI call, so it costs nothing.
It reports differences, never reasons. “Present on 3 of 4 cited pages, missing on yours” is something we can observe; “this is why you were not cited” is not, and would be guesswork dressed up as analysis. A signal is only raised when most of the cited pages carry it — one competitor doing something is a coincidence.
Each difference links to the screen in this plugin that closes it, so the report ends somewhere useful rather than at a list. Where you already have the feature and it is simply switched off, it says that instead of telling you to build something you own.
Share the audits: email, Slack, Google Docs
One report covering citations, store trust, entity corroboration and content readiness — sent by email, posted to Slack, or created as a Google Doc in your own Drive. Manually, or on a weekly or monthly schedule you set and can see.
All three render from the same data, so the email and the Doc can never quote different numbers. Every section carries its own “what this cannot tell you” note, because a report read in an inbox has none of the caveats that sit beside it on screen.
Nothing sends itself: no send-on-save, no default recipient, and reports go only to destinations you typed in. Slack receives a summary and a link back rather than the whole report, because a Slack block caps at 3,000 characters and the full thing would arrive truncated.
Merchant Center integrity
Compare your catalogue against your live Google and Bing Merchant Center feeds: price and availability mismatches, identifier gaps, and item-level disapproval codes.
Google feed readiness — no Google account needed
WooCommerce has no Merchant Center feed of its own; a feed plugin sends your catalogue to Google. With Google for WooCommerce active, this runs that plugin’s own code on your server to see exactly what it would send for every product and every variation, then checks it against Google’s published rules. Nothing is sent and nothing is changed.
- Never reaches Google — items Google for WooCommerce’s own validator refuses, with the reason (no image, no price, a backorder with no “available from” date)
- Google requires it — brand, GTIN or MPN, size and colour on products sold in several, gender and age group on clothing
- Feed contradicts the store — a feed price or stock status that differs from what the product page charges and sells
- Google recommends it — variation titles that do not name the variation, descriptions sent twice
A store section covers buying inside Google’s AI: Google shows its Buy button in AI Mode and Gemini only for products whose feed sets native_commerce(checkout_eligibility), and says plainly that Google for WooCommerce cannot send it today. Return policy, support contact and shipping cost are checked where the store holds them and marked “not checkable here” where they live in the Merchant Center account.
RAG Engine: your documents, matched to your products
Upload the spec sheets, manuals, safety data sheets and certificates your products come with; they are read on your own server. Product gaps lists, for each product a document is about, the facts in that document the product page never states — measurements, materials, ratings, test results — checked by value rather than wording. Nothing on your pages changes; every suggestion is a Copy button. Off until you switch it on under Modules.
Answer Readiness, per product
A score in the product editor rates how answerable the page is: whether the opening paragraph actually answers the question the title poses, whether specifics such as dimensions, materials and compatibility are present, and whether the copy is structured in passages an assistant can lift verbatim.
Product FAQs from real data
Generate question-and-answer pairs from what the product actually is — stock status, available sizes and colours, shipping destinations and times, and your returns window — so the answers stay true instead of inventing marketing claims.
AI-Ready: llms.txt and agent discovery
Publish /llms.txt describing your store for language models, plus a Content-Signal directive in robots.txt, agent-skills and MCP server-card discovery endpoints, and Link: relations advertising them. Machine clients can find and read your catalogue without scraping HTML.
Knowledge bundle at /okf/ (Open Knowledge Format)
Where llms.txt is the one-page overview, /okf/ is the full reference behind it: your store served as a directory of markdown concept files following Google’s Open Knowledge Format v0.2. An index manifest, your organization with its verified profiles, every product at its own /okf/products/<slug>.md path — brand, price, live availability, attributes and your product FAQs as plain facts — plus product categories, your policy pages and your articles, each with YAML frontmatter an AI system parses without rendering a page. Generated live from your catalogue, cached, linked from llms.txt, and on by default.
A sitemap that understands stock
General SEO plugins treat a product like any other post, so they have no answer for the things merchants actually argue about: what to do with a sold-out product, whether variations belong in the file, and whether hidden items should be advertised.
This one is opt-in and off until you turn it on. Out-of-stock handling is a choice, not a policy imposed on you — keep them (they hold ranking history and often come back), drop them, drop them only after a number of days, or follow the setting you already made in WooCommerce, which is the default so your sitemap agrees with your storefront. Backorder is treated as its own status, because a backorderable product can still be bought.
It never lists something you hid: catalogue-hidden products, anything marked noindex by your SEO plugin, URLs that are not the canonical, query-string duplicates, and drafts and password-protected items all stay out. Whatever your rules exclude is counted and shown back to you with the reason, so a missing product is never a mystery.
Availability comes from the variations, never the parent — WooCommerce’s parent stock status routinely reads “out of stock” while every variation is in stock, and on a catalogue of mostly variable products, trusting it would quietly drop the majority of your URLs.
Content types get their own feeds — products, product categories, tags, brands, pages and posts — each splitting into numbered files as it grows, all listed in a single /sitemap.xml index. That means you can see at a glance how many product URLs you are actually publishing, and Search Console reports coverage per type instead of one undifferentiated total. Each file is built independently, so serving a large catalogue does not re-walk it once per file.
If your store already publishes a sitemap, you are told which plugin is doing it and asked to turn that one off. Two sitemaps disagreeing about a catalogue is worse than one incomplete one. And if you would rather keep theirs, leave this off and the plugin audits it instead — reporting lastmod coverage, query-string URLs, listed URLs that are noindexed or non-canonical, and live status codes, with the fix written for whichever plugin generates it.
Local Store: “can I buy this near me?”
If you sell from a physical shop as well as online, this publishes the chain an assistant has to follow before it can name you: your store as a located Store entity with address, coordinates and opening hours, linked to your Organization — and every in-stock product’s offer pointing back at it with availableAtOrFrom. Without that last link your shop and your catalogue are two unrelated facts and nothing can join them.
- Service area from your real shipping zones — the postal codes you already serve become
areaServed, kept true as you edit your zones. Or set a delivery radius, or type it in. - In-store pickup, curbside and local delivery declared in a form machines read
- Live pickup availability on the product page — “Available for pickup at our [City] store today”, derived from the variations, so a variable product is never wrongly shown as unavailable
- Local questions answered in visible text — where you are, what your hours are, whether this product can be collected — and marked up as FAQ schema *only when they are actually rendered, never markup for content a visitor cannot see*
- Google local inventory feed at a fixed URL, ready to register in Merchant Center as a scheduled fetch — no API connection required
- Location landing pages for the places you genuinely serve, generated as drafts from real store data for you to edit and publish, never bulk-published
- A local completeness score plus a step-by-step trace of exactly where the “buy this near me” chain breaks — the first failing link is your answer
It is off until you switch it on. If another plugin already publishes local business schema, you are told which one by name and asked to pick — two competing entities for one shop is worse than either alone.
Smart Collections: answering “what should I buy?”
Your catalogue answers “what do you sell”. It cannot answer “what should a beginning photographer buy” — and that is the question a shopper asks two steps *before* they know which camera they want. It is also the question an assistant most often answers by recommending someone else.
Most stores already have the answer sitting in their categories: Starter Kits, Pro Setups, Under $500. Smart Collections turns those into published Collections — a CollectionPage with an ItemList, a plain-English summary of who the group is for, and an entry in your llms.txt.
- Finds the categories that already qualify — your category names are read against intent archetypes covering skill level, budget, use case, persona, problem and ecosystem. A category called “Cameras” is a catalogue division and is left alone; one called “Starter Kits” is an answer to a question
- Reads the intent queries you already rank for from Search Console — “for beginners”, “complete setup”, “commercial grade”, “under $X” — so you can see the questions your store is already close to answering
- Summaries built from your own catalogue — product count, real price range and live stock, all read at render time. Nothing is estimated. Have AI phrase it more naturally if you like, using your own key, and edit whatever it writes
- Build a collection for a query nothing matches — one at a time, from products you actually stock, drafted for you to edit before anything is created. Never in bulk
- Or change no URLs at all — put
[aeowc_smart_collection]on a page you already have and the schema follows the visible list
It is off until you switch it on, and switching the module on is not the same as publishing: there is a separate switch for that, and another per collection.
If Yoast, Rank Math or AIOSEO already publishes collection schema on your category archives, you are told which one by name. Hand the surface over and theirs is suppressed on the categories you have mapped — everywhere else stays theirs, untouched.
Reviews and reputation
Collect first-party reviews through a shortcode, publish them as review schema, and get platform recommendations matched to what your store actually sells — with Yelp called out specifically, because it now powers local-business answers inside ChatGPT.
Company identity and entity grounding
Establish who is behind the store: a single Organization node with logo, contact points and sameAs links to the profiles that corroborate you — managed on the Store Trust page, next to the audit that scores them. Add named entities with Wikidata and Wikipedia grounding, and map which pages are *about* them versus which merely mention them.
Social Graph
Open Graph and Twitter/X card tags generated from product data, so a shared product link renders correctly instead of falling back to whatever the theme guesses.
Built to stack with your SEO plugin, not to fight it
Most stores already run Yoast, Rank Math, All in One SEO or SEOPress, and those plugins do plenty this one does not. So the goal here is not to replace them. It is to make sure the page ends up valid — one clear answer per fact — whichever plugins are switched on.
Nothing changes when you install it. Every shared surface starts exactly as your store already was: they keep publishing what they published, we keep publishing what we published, and the plugin tells you where the two overlap and waits for you to decide. It never seizes a surface, and it never quietly stands down from one either.
Where possible it adds to their work instead of duplicating it. If you tell it to leave Product schema with Rank Math, it does not simply go quiet — it writes your commerce fields *into Rank Math’s own product entity*: offers, price, availability, shipping rates, return policy, quantity rules. One product on the page, published by them, carrying the data they do not collect. Their identity, name and description are never overwritten.
One thing it fixes without asking, because it is damage rather than a preference. If another plugin already publishes a breadcrumb trail at the same identifier and yours would differ, two conflicting trails would land on one entity with no way for a crawler to pick. That is worse than either plugin alone, and your store did not have that problem before this one arrived — so ours steps aside automatically. Switch the other plugin off and yours returns.
It will not tell you to remove a plugin it cannot fully replace. The plugin checks, per surface, whether it actually writes everything the other one writes. Today it does not write page titles, canonical URLs or robots directives — so today it recommends removing nothing, and says so plainly rather than selling you a switch that would cost you those tags.
Where two plugins disagree about the same product
Comparing which *fields* each plugin publishes only goes so far. Two plugins can both publish an offer and still tell an answer engine different things inside it.
The most common case is stock. WooCommerce’s parent stock status on a variable product routinely reads “out of stock” while every variation is in stock — so a plugin that trusts it publishes the wrong availability, and on a catalogue of mostly variable products that is most of your catalogue. This plugin derives availability from the variations instead.
Press one button and it fetches your own product pages, reads every product entity on them, and shows you exactly where the two writers disagree — the page, the field, and both values side by side. Nothing is fetched until you ask, and if your server cannot answer its own requests it says that, rather than reporting a clean result it never actually measured.
Plugins you did not choose to install
Some plugins install other plugins during setup. If several arrive on your site within moments of each other, this reports it: what landed, when, and — where the evidence supports it — which already-installed plugin was in a position to have put them there.
It is built on what happened on your site, never on a general claim about a vendor. Where nothing points to an installer, it says the batch was most likely your own work and lists it for completeness. Removal is a button you press, it only ever deactivates (never deletes), it refuses to touch anything outside what it reported, and it warns you first about anything you would lose.
Conflict detection
Scan your pages for competing JSON-LD emitted by other plugins, see which plugin produced each block, and suppress duplicates per type — so search engines are not handed two contradictory answers about the same product.
Roles and access
Grant staff access to the operational screens without handing over manage_options. Credential and configuration pages stay administrator-only.
Accessible admin screens
Every admin screen has been audited against WCAG 2.2 level AA, the target WordPress’s own accessibility coding standard sets. Toggles and icon-only controls carry proper accessible names, sortable columns are real buttons that work from the keyboard and announce their sort direction, controls that act on the page are buttons rather than links that go nowhere, focus rings are visible (including in Windows high-contrast mode), and the pass / warning / fail colours the whole product speaks in were re-tested for contrast against the background they actually sit on, not an assumed white one.
Every score explains itself
No unexplained numbers. Each score carries a “How this score works” panel covering what is measured, how it is calculated, why it matters — and honestly, what it cannot tell you.
Twelve video tutorials
Recorded on a real WooCommerce store. Each has a written guide, screenshots and a transcript. All tutorials
A licence that doesn’t hold your store hostage
A licence buys twelve months of updates and support for one site. It does not buy permission to keep running the software you already installed.
The plugin never disables itself. If the licence lapses, nothing breaks — it keeps publishing your schema exactly as before; you simply stop receiving updates and support until you renew.
Because it isn’t hosted on WordPress.org, updates come from our licence server instead. Once you enter a licence key the plugin joins WordPress’s own routine update check, sending your licence key, your site URL and the plugin name — nothing else. Before a key is entered it makes no contact at all, and it collects no usage telemetry either way.
Licence
GPL-2.0+
Disables on expiry
Never
Usage telemetry
None
Update check sends
Key, site URL, plugin name
Sites per licence
1
Refund window
30 days
On expiry
Keeps working
Privacy and data
The plugin collects no telemetry: no usage tracking and no analytics call of any kind. Its only contact with the vendor is the license check under External Services, made only after you enter a license key.
The only personal data it stores is what a shopper types into the review form: their name, their email address and their review. That data is wired into WordPress’s own privacy tools, so an export or erasure request is handled like any other — an erasure removes the name and email and reports that the review text was kept, so ratings you have already published stay accurate. Deleting the plugin removes the reviews entirely.
The AI crawler log records the requested URL, the crawler’s user-agent and the HTTP status returned. It deliberately stores no IP addresses, so it holds no personal data at all.
The AI Traffic module, when you enable it, records the page requested, the time, the requesting user-agent, the referring site where one was sent, and the network the request came from. The address is truncated to its network before it is stored, so a row identifies a network and a page rather than a device or a person. No cookie is set, no account is created, and nothing follows a visitor between pages. Its front-end script transmits only derived yes/no verdicts — raw screen and window dimensions never leave the browser — so what is stored cannot form a fingerprint. Do Not Track and Global Privacy Control are both honoured. For the same reason there is deliberately no exporter or eraser for this data: WordPress keys both on an email address, and there is nothing here to look one up by.
The AI Citation Monitor stores the questions you wrote, which engine was asked, the date, and the URLs the engine returned as sources — plus a short excerpt of its answer. It holds nothing about your shoppers. History is pruned to a retention window you set.
Optional AI and Merchant Center features send product and page content to the providers listed below, using credentials you supply. Customer personal data is never included.
Report sharing sends only your own audit results, and only to destinations you enter yourself — an email address, your Slack webhook, or your Google Drive. Nothing is sent to the plugin author, and no report is sent automatically unless you set a schedule.
External services
This plugin collects no telemetry. The only vendor contact it can make is the license server below — dormant until you enter your license key, and limited to the three values listed. Every other service is optional, dormant until you supply your own credentials and enable the feature, and is billed to your own account. No store-visitor or customer personal data is ever sent to any of them.
TWT License Server — licensing and updates
Activates your purchase key and delivers plugin updates. Sends your license key, this site’s URL and the plugin name — nothing else — to tampawebtech.com/wp-json/twt-license/v1. Only after you enter a key on Settings → License: on the buttons there, and during WordPress’s routine twice-daily update check. An expired license stops updates only; the plugin keeps working.
Anthropic (Claude) — AI text generation
Used for AI meta descriptions and product enrichment. Sends the product or page text you ask it to work on, plus your Anthropic API key, to api.anthropic.com. Only when you click an AI action or run a bulk job you started. Terms: https://www.anthropic.com/legal/consumer-terms — Privacy: https://www.anthropic.com/legal/privacy
OpenAI — AI text and image generation
Used for AI meta descriptions, product enrichment and Open Graph image generation. Sends your prompt text and your OpenAI API key to api.openai.com. Only on an AI action you trigger. Terms: https://openai.com/policies/terms-of-use — Privacy: https://openai.com/policies/privacy-policy
Google Gemini — AI text generation
Alternative AI provider for the same features. Sends your prompt text and Gemini API key to generativelanguage.googleapis.com. Only on an AI action you trigger. Terms: https://policies.google.com/terms — Privacy: https://policies.google.com/privacy
Perplexity — brand authority lookups and citation checks
Optional. Resolves authoritative brand profile URLs (sameAs) and, if you enable the AI Citation Monitor, runs your own questions against Perplexity’s web-grounded model to see whether your domain is cited. Sends a brand name, product name or the question you wrote, plus your Perplexity API key, to api.perplexity.ai. Only on an action you trigger or the schedule you set. Terms: https://www.perplexity.ai/hub/legal/terms-of-service — Privacy: https://www.perplexity.ai/hub/legal/privacy-policy
Slack — report delivery (optional)
Off unless you enable the Slack channel and paste an incoming webhook URL. Sends a summary of your own report — section headlines, counts and the date — plus a link back to your WordPress admin, to hooks.slack.com. Detail rows stay on your site. Only that host is accepted; any other URL entered in the webhook field is rejected rather than posted to. Sent only when you click Send or on the schedule you set. Terms: https://slack.com/terms-of-service — Privacy: https://slack.com/trust/privacy/privacy-policy
Stripe — agent checkout payments (optional)
Only with the Agentic Checkout module on and the WooCommerce Stripe Payment Gateway enabled, using that gateway’s own keys. When an AI agent completes a checkout, sends the amount, currency, the payment token the agent supplied, and the order and checkout ids to api.stripe.com to charge it. Once per secret key, reads your Stripe account id (when an administrator opens the Agent Ready page). No card data passes through this plugin. Terms: https://stripe.com/legal/ssa — Privacy: https://stripe.com/privacy
YouTube — tutorial videos (only when you press Play)
Screens that have a tutorial show a “?” beside their title. Pressing it opens a panel; nothing is requested from YouTube until you then press Play, which loads that tutorial in a player from www.youtube-nocookie.com (YouTube’s privacy-enhanced mode). From then on YouTube receives what any embedded player sends: your IP address, browser details and the video being watched. No store data is sent. Closing the panel removes the player. Terms: https://www.youtube.com/t/terms — Privacy: https://policies.google.com/privacy
Google Drive — report export to Google Docs (optional)
Off unless you enable the Google Docs channel. Uploads the report you chose to share — your own audit results — to www.googleapis.com, where Google converts it into a Doc in your own Drive. Uses the drive.file permission only: per-file access limited to documents this plugin creates, so it cannot read or modify anything else in your Drive. Because this permission was added after the original Google integration, an existing connection must be disconnected and reconnected before the channel becomes available. Sent only when you click Send or on the schedule you set. Terms: https://policies.google.com/terms — Privacy: https://policies.google.com/privacy
Cited pages you are compared against (no credentials, no data sent)
When you ask what the pages an engine cited have that yours does not, the plugin fetches those public pages and reads their markup. It sends no data beyond a normal web request identifying itself as this plugin, uses no API key, and caches each page so reopening the screen re-fetches nobody.
AI crawler IP-range lists (no credentials, no data sent)
When a request claims to be an AI crawler, the plugin checks whether it really came from that operator. It downloads the address ranges each operator publishes for exactly this purpose — openai.com (gptbot.json, searchbot.json, chatgpt-user.json, adsbot.json), claude.com (crawling/bots.json), www.perplexity.com (perplexitybot.json, perplexity-user.json), and developers.google.com (static/search/apis/ipranges/*.json). Cloudflare’s address ranges, used only to confirm a request genuinely arrived through Cloudflare before trusting the client address it forwards, are bundled with the plugin and never downloaded. Nothing is sent: no API key, no store data, and no information about your site or your visitors — the addresses being checked never leave your server; the published list is downloaded and compared locally. Refreshed on a schedule or when you click, never on a front-end request. If a refresh fails the previous copy is kept and verification reports “unverified” rather than a false result.
Google APIs — Merchant Center, Search Console, Analytics, Knowledge Graph
Used to compare your catalogue against your live Google Merchant Center feed and to read reporting data. Sends OAuth tokens, account/property identifiers and your own product data to accounts.google.com, oauth2.googleapis.com, www.googleapis.com, shoppingcontent.googleapis.com, searchconsole.googleapis.com, analyticsadmin.googleapis.com, analyticsdata.googleapis.com and kgsearch.googleapis.com. Only after you connect the integration, then on sync or when you open a report. Terms: https://policies.google.com/terms — Privacy: https://policies.google.com/privacy
Microsoft / Bing — Merchant Center and Webmaster Tools
Used for Bing Merchant Center feed parity and Bing Webmaster submissions. Sends OAuth tokens and your own product or URL data to login.microsoftonline.com, content.api.bingads.microsoft.com and ssl.bing.com. Only after you connect the integration. Terms: https://www.microsoft.com/en-us/servicesagreement — Privacy: https://www.microsoft.com/en-us/privacy/privacystatement
IndexNow — instant indexing
Optional. Sends your IndexNow key and the public URLs of published or updated content to participating search engines. Only when the module is enabled. Protocol and terms: https://www.indexnow.org/
Changelog
Every change in every release, newest first.
Version 1.0.3
- Changed: Cloudflare’s address ranges (used to verify AI crawlers behind Cloudflare) ship with the plugin instead of being downloaded.
- Changed: AI Ready’s rate limit applies only to the agent API endpoints; it no longer turns crawlers (Googlebot included) away from your pages.
- New: a “?” beside 14 screen titles opens that screen’s video tutorial (loads from YouTube only when you press Play).
- Changed: requires WordPress 6.9 or later (was 6.5).
- New: Agentic Checkout (UCP + ACP) — AI agents can check out on your store through Google’s Universal Commerce Protocol and the Agentic Commerce Protocol. WooCommerce calculates every total; with Stripe set up, agents pay through your own Stripe account and the order lands in WooCommerce. Off until switched on.
- New: find any product on Products & Feeds by name, SKU or ID.
- Changed: AI defaults are now Claude Haiku 5.5, Claude Opus 5.5 and Gemini 3.8 Flash.
- Fixed: OpenAI meta descriptions were all failing.
- Fixed: detergents and creams were classed as services; a blank flat shipping rate published an empty price.
- Fixed: browsers filled the WordPress login into the Merchant Center and API key forms.
- Fixed: two breadcrumb lists on product category pages.
- Fixed: free-shipping coupons described as “$0.00 off”; the on-page coupon note now matches its FAQ markup word for word.
- Fixed: the Crawler Log counted scrapers and impersonators as AI crawlers; AI Citations counted a failed check as “not cited”.
- Fixed: screens that contradicted their own measurements (Dashboard ownership rows, Agent Ready, Store Trust shipping, Local Audit), Edit Schema’s image field, and the Smart Collections category list and Add form.
- New: Google feed readiness — runs Google for WooCommerce’s own code to show exactly what it would send to Merchant Center, checked against Google’s product data and AI checkout rules. No Google account needed; nothing is sent or changed.
- New: Agent Ready — whether AI shopping agents can reach, read and buy from the store, with inventory checks, a UCP/ACP discovery-file check against each protocol’s spec, and the payment routes reported for Muse (Meta AI).
- New: AI agents that sign their requests (Web Bot Auth) are recognised in AI Traffic, Store API carts included. Nothing is blocked.
- New: RAG Engine — upload product documents; Product gaps lists the facts they state that your product pages do not. Off until switched on.
- New: AEO Ultimate for WooCommerce connector for Claude — off by default; every change is a preview you approve, with an exact undo. 49 tools (28 reads, 21 changes), all also available to the official WordPress MCP Adapter through the Abilities API.
- New: Advanced AI Citations — each market’s leading AI engines write their own question for every word order of your real searches, and are asked it; “cited X of Y” per engine, a suggested heading-and-opening layout per page, and before/after re-checks.
- New: AI Citations asks eight engines, as a queued run; shopper personas, follow-up questions and missing shopper vocabulary.
- New: the setup wizard can bring settings across from TWT AEO Ultimate without overwriting anything set here.
- Fixed: Perplexity checks (retired endpoint) and Gemini citation matching.
- Fixed: the AI meta-description job now retries a busy provider instead of stopping.
- New: license key activation and automatic updates (Settings → License). Nothing is sent to the vendor until a key is entered; afterwards only the key, this site’s URL and the plugin name are sent, on your clicks and during WordPress’s routine update check. An expired license stops updates and nothing else.
- Removed: the Dashboard’s “Plugins that were installed together” section. The ownership table (“What this plugin writes, and what something else writes”) is unchanged.
- New: /llms.txt can be handed to an owner on the ownership table — choosing AEO Ultimate stands the free plugin down (2.13+); choosing the free plugin stands this one down, rewrite included, so the URL always serves exactly one file.
- Fixed: the free-plugin stand-down handshake could be silently absent on requests where nothing loaded the bridge class; it is now guaranteed at boot.
- New: Content Uniqueness “Check the web” — exact-phrase searches for distinctive passages of your own product copy, reporting how many other pages carry the same sentence (a floor counted from results actually read, with links). Needs a Brave Search or Serper key; capped per day.
- New: Citation Monitor product questions judged against your real prices and stock — an engine quoting a wrong price or availability is flagged WRONG, with what it said versus what your store says. Deterministic judging, no extra API spend. Every check now also records the model that answered, with per-model citation splits per question.
- New: Pinterest Product Rich Pin tags on product pages — live price, currency, availability and brand in the exact vocabulary Pinterest reads. Range-priced and subscription products deliberately carry no price tag; stands down when Yoast WooCommerce SEO writes the same tags.
- New: Meta Descriptions page — catalogue-wide description templates ({{title}}, {{brand}}, {{category}}, {{price}}, {{store}}) with validation that refuses typos and duplicate-producing templates, a free preview, batch apply that never overwrites without permission, and a duplicate scanner that certifies zero shared descriptions or names every product carrying one.
- New: Command Center buying-intent queries now name the catalogue product they mean (conservatively — an ambiguous query is never linked), flag matched products missing a meta description, and drop rank-tracker noise from the query lists.
- New: catalogue issues census on Products & Feeds — brand-equals-store-name, missing GTINs (parent and variations), thin descriptions, uncategorized products and conditional pricing, each explained with example products linked.
- New: Find image on Content Uniqueness — imports the product photo the manufacturer’s own page declares, for products that have none. Gated (logo tripwires block; Merchant Center size floors warn with the numbers), merchant-confirmed, never replaces a chosen image.
- New: the Manufacturer page’s Wikidata suggestions show the entity’s label, description and aliases, and an entity whose name does not match the brand is never suggested.
- Changed: the AI Input / RAG content signal now recommends and defaults to “Allow as input”.
ai-input=notold answer engines not to use the store’s content in answers — the opposite of what an AEO plugin is for. A previously saved choice is never changed. - Fixed: the setup wizard’s API Keys step only offered Claude and OpenAI. It now carries Gemini and Perplexity too, matching the Settings page — and no longer silently drops a Gemini key on save.
- Fixed: stored product FAQs corrupted em-dashes, accents and emoji (saved as literal “u2014”-style text). Non-ASCII characters now survive save and read intact. Re-save any FAQ that shows the corruption.
- Changed: the dynamic robots.txt is shared-cacheable for 5 minutes instead of forced no-cache. Google treats a failed robots.txt fetch as “Robots.txt unreachable” and aborts the whole crawl, so an origin hiccup could reject indexing requests site-wide; a short CDN cache bridges that. Filter
aeowc_robots_cache_ttlto tune. - Fixed: heading-based FAQ detection on pages and posts required headings that actually read as questions. A feature-card grid could previously be detected — and published — as FAQ schema full of fake questions.
- Fixed two PHP 8.4 deprecation notices:
AEOWC_Offer_Pricing::shared_breaks()andAEOWC_Dynamic_Pricing::status_label()declared typed parameters with anulldefault without marking the type nullable. Both now use explicit nullable types. No behaviour change.
Version 1.0.2
- Google Merchant Center moved to the Merchant API. Google shuts down the Content API for Shopping on 18 August 2026, after which the Products & Feeds comparison would have stopped working. Three fields changed shape and are handled rather than left to fail quietly: barcodes became a list (reading the old field reports your whole catalogue as missing GTINs), item rejections moved to a new field (disapproved products would have disappeared from the screen while Google kept refusing them), and prices arrive in millionths (a $15.00 product compared against 15000000 reports every item as mismatched).
- The Merchant API needs your Google Cloud project registered against your Merchant Center account, one project per account. Content API had no such step and Google answers with a bare 403, so the plugin names that as the cause instead of showing the raw error.
- Connect Google from Settings. The integration had a complete back end and no way to reach it — nowhere to enter credentials, nowhere to pick a Search Console property, nothing listening for Google’s reply — so everything downstream was inactive on every install: Command Center buyer intent, the Search Console panels on URL Health, index decay, Smart Collections intent queries, empty-category search signals and Google Docs report delivery.
- Import a manufacturer’s product feed (Content Uniqueness → Import feed). If you resell, the manufacturer’s CSV already carries the barcode, part number and their own description. Importing it replaces a per-product AI search that costs money and takes about a minute each with a file read that costs nothing — and gives you their actual copy rather than an approximation of it. Scanning is now positioned as the answer for merchants who have no feed.
- Nothing is written until a dry run reports what an import would do: rows matched by barcode, part number or SKU, rows matched through a variation, rows refused and why, and how many products already carry a manufacturer source that would be replaced. Every count comes with named examples. SKU matching is offered but off by default, because your SKU is usually your own internal code rather than the manufacturer’s. A row whose identifier matches two products is reported, never guessed at.
- The same upload can confirm your brands’ manufacturer websites from the feed’s product links, which is what unblocks scanning for products the feed does not cover. Offered as checkboxes, never applied automatically, and a website you already confirmed is never replaced by a file.
- Importing a feed will usually push your “Theirs” percentages up, and the screen says so before the button. Products with nothing to compare against were shown as unmeasured, not as original; importing the manufacturer’s copy is what finally makes existing duplication visible.
- Scan a whole catalogue for manufacturer pages in the background rather than one product at a time. Runs on its own so you can close the page, stops on request, and pauses at a daily limit you set, resuming the next day. Products that cannot succeed are excluded before anything is spent and counted separately by reason, because a product with no brand and a product whose brand has no confirmed website need opposite fixes.
- Fixed: brand website suggestions were wrong for any brand with “&” in its name. The lookup did not encode what it sent, so “Ben & Jerry’s” was searched as “Ben” — whose top result is a person — and the suggested manufacturer website was that person’s employer. Barnes & Noble, Procter & Gamble and Dolce & Gabbana were the same bug. A person can no longer be offered as a brand’s website at all, and the Suggest button now links the entry it used so you can check it.
- Fixed: a manufacturer description you pasted in yourself was refused unless that product’s brand already had a confirmed manufacturer website. Pasting text fetches nothing, so it never needed one — and products with no brand set could never accept a pasted description at all. Pasting a link still requires the brand’s website, because a link does get fetched.
- A brand can now have more than one manufacturer website, for manufacturers that run a separate dealer or spec-sheet portal on its own domain. Existing settings are unchanged.
- Internal compliance pass against Plugin Check: fifteen errors cleared across translator comments, output escaping and database query annotations. No behaviour change.
Version 1.0.1
- Fixed: “Generate Schema for All” silently stopped at 200 products. The bulk action read from a scan capped at 200 rows, so on a larger catalogue it generated schema for the first 200, skipped everything after, and still reported success. It now walks the whole catalogue in batches with visible progress and resumes from where it stopped if a batch fails.
- Fixed: the Products & Feeds product count reported the scan cap rather than the catalogue, so a 949-product store showed “200 products”. Counted directly now. The results table still scans the 200 most recent products — doing a whole catalogue in one page load times out — but the screen now says so and says which figures that limit applies to.
- Fixed: bundled translations were never loaded, so every translated string fell back to English regardless of the translation supplied.
- Plugin classes are autoloaded instead of unconditionally required. Roughly 160 includes were loading every class on every request, including all 28 admin page classes on front-end product views, REST requests and cron.
- Google, Bing and Merchant Center errors go to the plugin’s own log — visible on Diagnostics, redacted and capped — instead of straight to the server error log.
Version 1.0.0
- Initial release for WooCommerce.
- One connected JSON-LD
@graphper product page — Organization/OnlineStore, WebSite, WebPage, Product or ProductGroup, Offer or AggregateOffer, and BreadcrumbList, all cross-referenced by@id. - The full product gallery publishes as
ImageObjectnodes with captions where written; filename-derived titles are suppressed. Open Graph stays a singleog:image. - Entity corroboration: a store-type-aware directory checklist feeding
Organization.sameAs, with certifications routed tohasCertification, D-U-N-S toduns, and a warning when a URL does not match the directory it was entered under. - AI Citation Monitor: runs your questions against web-grounded Perplexity, ChatGPT and Gemini and records whether your domain was cited, storing every observation so each question shows a trend. Cited and named are recorded separately; no share-of-voice percentage is invented. Off by default, capped per day, never runs on the front end.
- AI Traffic module (opt-in) — separates crawls, live retrievals and human arrivals from an AI answer into three counted buckets, plus a fourth for everything refused. Crawler identity is verified against the operator’s own published IP ranges, so impersonation is reported rather than counted, and an operator with no published list reads “unverified” instead of being guessed at. Attribution carries a confidence tier and the coincidence risk behind it, measured from each page’s own baseline — the same time gap is evidence on a quiet page and Unattributed on a busy one — with inference capped below “Confirmed” by design. The plugin’s own crawler tests and citation probes are recognised and excluded from every total. Periods compare against an equal preceding span, so a part-finished week never manufactures a decline. Search, engine filtering, sorting and pagination on every tab. No cookies, no fingerprint; addresses are truncated to their network before storage and Do Not Track and Global Privacy Control are honoured.
- Citation gap analysis: fetches the pages an engine cited and reports what they carry that your page does not — schema types, depth, FAQ, meta description, author, headings. No AI call, no cost. Reports differences, never causes. Each difference links to the screen that closes it, and says so when you already own the feature and merely have it switched off.
- Shareable reports by email, Slack or Google Docs, covering citations, store trust, entity corroboration and content readiness. One renderer per channel from one dataset, every section carrying its own limitations note. Manual or scheduled; nothing sends itself.
- Image alt text is generated by analysing the image rather than its filename, with a stated cost/accuracy choice between a standard and a precision tier and links to live provider pricing instead of figures that go stale.
- Local Store: worked examples for pickup, curbside and delivery instructions, guidance on what a valid region is for both the address and the service area, and a working link out of the location-pages screen when the module or the location is not yet enabled.
- Schema is rebuilt from live WooCommerce data on every render; only merchant edits are stored, so price, stock, variants and shipping never freeze.
- Prices use the tax-inclusive figure shoppers see; all three stock statuses map correctly, including on-backorder; shipping reads live WooCommerce shipping zones; GTIN uses WooCommerce’s native product identifier.
- Variable products emit a ProductGroup with every variation addressable, plus an AggregateOffer price range on the parent.
- Suppresses WooCommerce’s duplicate Product and BreadcrumbList markup.
- Store Trust scorecard — three pillars, ten checks, letter grade, every check explained.
- Local Store module (opt-in) — a located
Storeentity linked to your Organization, withavailableAtOrFromon in-stock offers,areaServedbuilt from your live shipping zones, opening hours, in-store pickup, curbside and local delivery; visible pickup availability and local FAQs on product pages; a Google local inventory feed; draft location landing pages gated on your real service area; and a local completeness score. - Smart Collections module (opt-in) — turns intent-shaped categories (“Starter Kit”, “Pro Setup”, “Under $500”) into published Collections:
CollectionPagewith anItemList, a summary built from real product counts, prices and live stock, and an llms.txt entry. Finds qualifying categories automatically, surfaces the intent queries you already rank for from Search Console, and can build a collection for a query nothing matches — one at a time, drafted from products you stock, editable before anything is created, with an exact Undo. Takes over the empty collection schema other SEO plugins publish on mapped archives. Also available as a shortcode for merchants who want no URL changes at all. - Promotions & Price Rules module (opt-in) — the two conditional prices that can be published honestly: quantity breaks as a
UnitPriceSpecificationwitheligibleQuantity, and member pricing as aMemberProgramwithvalidForMemberTier. Your offer price andpriceValidUntilare never rewritten. Coupon codes go to Google Merchant Center as promotions and to a visible on-page note, marked up only because it is rendered. Reads WooCommerce coupons directly, so Advanced Coupons, Smart Coupons, WebToffee and Power Coupons are covered, and names anything that can change a price without a code. Quantity breaks are withheld unless your store can actually apply them, and every coupon that cannot be used is listed with the reason. - Sitemap generator (opt-in) — off until you turn it on, and when on it takes over
/sitemap.xmland names whichever plugin was publishing one so you can turn that off. Out-of-stock handling is a setting: keep, drop, drop after N days, or follow your WooCommerce catalogue setting (the default). Backorder is treated as its own status. Excludes catalogue-hidden products, noindexed items, non-canonical URLs, query-string duplicates, drafts and password-protected content — and counts every exclusion with its reason so nothing goes missing silently. Availability is derived from variations, never the unreliable parent stock status. Separate feeds per content type — products, categories, tags, brands, pages and posts — each chunking as it grows, all listed in one/sitemap.xmlindex, generated per file so a large catalogue is not re-walked for every request. Pages through the whole catalogue rather than truncating. Leave it off and the plugin audits your existing sitemap instead:lastmodcoverage and staleness, query-string URLs, listed URLs that are noindexed or non-canonical, and sampled status codes, each with a fix written for the plugin that actually generates it. - Crawler visibility test — requests your homepage as each major AI crawler and reports what your server returned, so an empty crawler log can be told apart from a firewall block. Names the blocking layer and the user-agent to allow, and states plainly where a pass proves less than it appears.
- Command Center — buyer-intent classification of Search Console queries.
- Google and Bing Merchant Center feed parity, with item-level disapproval codes.
- AI meta descriptions and product enrichment via your own Claude, OpenAI or Gemini key; CSV imports queue instead of blocking, paced with a daily cap and off by default.
- AI crawler log with per-bot, per-product detail and real HTTP statuses.
- AI-Ready — llms.txt, Content-Signal robots directive, agent-skills and MCP discovery.
- Knowledge bundle at /okf/ — the store as Open Knowledge Format v0.2 markdown concept files (organization, products, categories, policies, articles), linked from llms.txt.
- Answer Readiness score and product FAQ generation in the product editor.
- Reviews with schema output and store-category platform recommendations.
- Company identity, entity grounding with Wikidata/Wikipedia
sameAs, and Social Graph tags. - Schema conflict detection across other SEO plugins, suppressible per type.
- Per-surface ownership for every schema surface two plugins can both write. The default is no change at all — both keep publishing exactly what they published before, the overlap is reported, and nothing moves until you choose. Choosing us suppresses theirs through a filter the other plugin publishes itself, leaving its settings untouched.
- Choosing to keep the other plugin’s product schema does not throw your commerce data away: offers, price, availability, shipping, returns and quantity rules are written into *their* product entity. One product on the page, published by them, carrying what they do not collect. Their identifier, type, name and description are never overwritten.
- Contradictory markup is resolved automatically, and only where it is genuinely contradictory: if another plugin publishes a different breadcrumb trail at the same identifier, ours steps aside rather than leaving one entity with two conflicting trails and no way to choose. Ours returns the moment nothing else claims it. A merely differing type is left alone, because subtypes resolve correctly on their own.
- The plugin will not recommend removing another plugin it cannot fully replace. It checks per surface whether it writes everything the other writes; page titles, canonical URLs and robots directives are not written here, so today that recommendation never appears.
- Value comparison between writers: fetches your own product pages on request and reports where two plugins publish different values for the same field — most often stock status, because WooCommerce’s parent status on variable products is unreliable and this plugin derives availability from the variations. A run that could not reach your pages reports that it could not check, never that nothing was found.
- Unrequested-install detection: reports plugins that arrived in one burst and, where the evidence supports it, which installed plugin was in a position to have put them there. Evidence from your own site only; deactivation is a button, never automatic, never deletes, and warns you what you would lose first.
- Weight and dimensions published as measured quantities with the unit code your store uses.
- A product with no price set publishes no offer, rather than an offer priced at zero.
- Role-based access to operational screens; credentials stay administrator-only.
- HPOS and Cart & Checkout blocks compatibility declared.
- Admin screens audited against WCAG 2.2 level AA — accessible names on every toggle and icon control, keyboard-operable sortable columns that announce their sort direction, real buttons in place of links that navigate nowhere, visible focus rings that survive forced-colors mode, and status colours re-tested for contrast against their actual background.
- Full WordPress privacy-tools integration: personal-data exporter, eraser and suggested privacy-policy text. No telemetry, no license check, no call-home.
Support
You reach a person
Most replies land within the hour during Tampa working hours. Email, Slack or WhatsApp — no ticket system, and free-tier users get the same attention.
Compatibility
Plays nicely
HPOS and Cart & Checkout blocks compatible. Runs alongside Yoast, Rank Math and AIOSEO without duplicating schema.
AI features
Your key, no meter
Bring your own Claude, OpenAI, Gemini, Perplexity, xAI or Mistral key and pay the provider’s raw rate. Nothing is sent until you trigger a feature.
AEO Ultimate for WooCommerce
For sale now. Nothing breaks if you stop.
One site, twelve months of updates and support, on a GPL licence. Full refund within 30 days.