AI search & visibility
The vocabulary of the new front door. If your customers are asking AI assistants questions, these are the words that describe what's happening.
AEO (Answer Engine Optimisation)
Getting your business named inside answers AI assistants give to your customers.
Where SEO tries to rank you on a results page (the ten blue links), AEO tries to get you cited when an AI assistant generates a single direct answer. The two work together but have different mechanics: SEO ranks pages, AEO names entities.
Why it matters. When a customer asks ChatGPT "best London accountant for SaaS startups," the assistant doesn't show ten options — it usually names one or two. AEO is what determines whether that name is yours.
GEO (Generative Engine Optimisation)
A near-synonym for AEO, used interchangeably in most contexts.
"GEO" puts emphasis on the fact that AI engines generate answers from multiple sources, rather than retrieving a fixed page. Some practitioners prefer the term because it captures the multi-source synthesis better than "AEO." Functionally, the two refer to the same work.
Why it matters. If you hear someone use either acronym, they're talking about the same outcome: getting cited in AI-generated answers.
SEO (Search Engine Optimisation)
Optimising a website to rank higher in traditional search results.
The discipline most marketers know: keyword research, link building, page-speed work, on-page content, technical site health. SEO is still important — Google still drives meaningful traffic — but it optimises for clicks on results, not citations inside answers.
Why it matters. SEO and AEO are complementary. A solid SEO foundation usually helps AEO. But doing only SEO leaves you invisible inside the new AI-driven half of search.
Citation
When an AI assistant names your business by name in an answer.
The currency of AEO. Unlike search positions, citations are binary: an AI assistant either names you or it doesn't. There's no equivalent of "page 2" — being unnamed in an AI answer is the same as not existing for that query.
Why it matters. Citations compound. AI engines re-cite brands they've cited before, because confident citations make for cleaner answers. The brands cited today become the defaults of tomorrow.
Entity / Entity authority
How confident an AI engine is that you're a real, identifiable business worth recommending.
An "entity" in this context means the AI engine's internal model of who you are: name, category, location, services, and the strength of the signals connecting them. Entity authority is the cumulative weight of those signals.
Why it matters. A business with weak entity authority gets confused with similarly-named competitors, or gets left out of answers entirely. Schema, citations, and consistent third-party mentions all build entity authority.
Topical authority
The depth and breadth of signals showing your business is a credible expert in a topic.
Where entity authority is "the AI engine knows who you are," topical authority is "the AI engine believes you know your subject." Built through detailed content, third-party citations, and consistent semantic context across your site and the wider web.
Why it matters. Two businesses can have equal entity authority — both clearly identified — but the one with higher topical authority gets cited for category questions, not just brand-name questions.
AI Visibility Check
A free 7-day audit Rex Commerce® runs for waitlist subscribers.
Covers current citation share across the six major AI assistants for your top 10 category prompts, competitor citation gap, schema audit, feed compatibility check, and a fit-for-Rex-Commerce assessment. Runs after waitlist signup, with no obligation to proceed.
Why it matters. It's the cheapest way to find out whether AI search is currently helping or hurting your business — measured, not guessed. Join the waitlist to get yours.
AI assistants & agents
The systems doing the asking, answering, and increasingly the buying. Different from search engines, different from each other.
AI assistant
A consumer-facing app powered by a large language model that answers questions and increasingly takes actions.
Examples: ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google), Perplexity, Microsoft Copilot, Apple Intelligence. They differ in personality, training data, and which surfaces they appear on, but they all answer category questions and increasingly mediate purchases.
Why it matters. Each assistant has its own ranking signals, its own preferred data formats, and its own merchant programmes. Rex Commerce® tracks visibility across all six.
AI agent
An AI system that takes actions on a user's behalf — browsing sites, comparing products, completing checkouts.
The difference between an assistant and an agent is autonomy. An assistant answers; an agent does. Agents browse, click, fill forms, and check out without the user manually approving each step. ChatGPT's Operator and Anthropic's Computer Use are early examples.
Why it matters. The number of buying decisions made by agents on behalf of humans is set to grow. If your site doesn't speak the languages agents understand (schema, ACP, UCP), they can't recommend or buy from you.
Agentic commerce
Buying and selling that happens through AI agents acting for customers, not the customers themselves clicking through traditional ecommerce flows.
It's the umbrella term covering everything that has to be in place for an AI agent to discover, compare, recommend, and complete a purchase from your business. Rex Commerce® is named for it.
Why it matters. Agentic commerce is to ecommerce what ecommerce was to mail-order catalogues — same fundamental transaction, completely different mechanics. The merchants ready for it now will be the defaults when it goes mainstream.
AI-referrer (traffic)
Visitors who arrive on your site after clicking through a link inside an AI assistant's answer.
Behaves differently from search-referred traffic: typically arrives later in the buyer journey (the AI has done the comparison work), tolerates fewer hand-holding steps, expects machine-readable confirmation of facts, and frequently doesn't load tracking pixels.
Why it matters. AI-referred visitors convert at higher rates than search-referred visitors when the landing page matches their expectation — and at lower rates when it doesn't. AI-referrer CRO is a specialised discipline for this.
Open standards & protocols
The technical specifications that make AI search and agentic commerce actually work. You don't need to know the details — that's our job — but knowing the names helps you ask the right questions.
Schema.org
A shared vocabulary for describing what's on your website in a way every AI engine understands.
Founded by Google, Microsoft, Yahoo, and Yandex in 2011 and still maintained collaboratively. Schema.org defines hundreds of "types" — Product, Service, Organization, FAQPage, Person, and so on — each with named properties.
Why it matters. If you don't use Schema.org vocabulary on your site, AI engines have to guess what your business is. Guesses are wrong more often than facts.
JSON-LD
The format AI assistants prefer to read structured data in.
Stands for "JavaScript Object Notation for Linked Data." A small block of code embedded in your page's source — invisible to human visitors, fully readable by AI engines and search crawlers. Recommended by Google over older formats (microdata, RDFa).
Why it matters. JSON-LD is how AI engines extract facts from your site without parsing the visual layout. Bad or missing JSON-LD is one of the most common reasons businesses are invisible to AI assistants. json-ld.org ↗
Structured data
Information on your website that's labelled in a machine-readable way.
The umbrella term for what you get when you combine Schema.org vocabulary with JSON-LD format. "We added structured data to your product pages" means "we used JSON-LD to mark up products with Schema.org Product type."
Why it matters. Structured data is the difference between an AI engine reading "London accountant Sarah Smith" as a string of words versus reading it as: Person.name = Sarah Smith, Person.jobTitle = Accountant, Person.address.locality = London. Facts beat fuzzy-string matching.
OpenAI ACP (Agentic Commerce Protocol)
OpenAI's open standard that lets ChatGPT and other AI agents browse and buy from merchants on behalf of users.
ACP defines the data formats and endpoints a merchant has to expose for an AI agent to read the catalogue, place an order, and confirm fulfilment — without bespoke integration per assistant.
Why it matters. If you're not ACP-compatible, ChatGPT cannot buy from you. As ChatGPT's shopping behaviour grows (around 50 million shopping queries per day), this matters more every quarter.
Google UCP (Universal Commerce Protocol)
Google's specification for AI shopping agents to discover, read, and recommend a merchant's product catalogue.
The Google-side counterpart to ACP. Designed to be portable across multiple assistants, not just Google's own. Includes the UCP capability profile mechanism for declaring what a merchant supports.
Why it matters. Google still mediates the largest share of buyer intent on the open web. A UCP-compliant feed is how you stay in Google's AI shopping recommendations.
Stripe SPT (Shared Payment Token)
Stripe's specification that lets an AI agent securely complete a checkout for a user.
Without SPT, every AI agent would need its own bespoke payment integration with every merchant — unworkable at scale. SPT lets the agent present a one-time tokenised payment that the merchant can charge through Stripe without seeing the user's card.
Why it matters. SPT is what makes agentic checkout actually safe and scalable. Without it, agentic commerce stops at "compare and recommend" and never reaches "complete the sale."
UCP capability profile
A small file at /.well-known/ucp on your domain that tells AI shopping agents what your business supports.
Declares pricing currencies, shipping zones, return windows, fulfilment SLAs, and category coverage. AI agents read it to decide whether to recommend you for a given customer query.
Why it matters. A missing or stale capability profile means AI agents have to guess your business model. Guesses are conservative — they default to not recommending you.
Product feed
A structured file (or live API endpoint) listing your products with prices, descriptions, images, and stock.
The data source AI agents pull from when making product recommendations. ACP and UCP both define how the feed should be structured. A feed that updates in real-time as your inventory changes is essential for high-volume catalogues.
Why it matters. Stale feed data is worse than no feed at all — recommending an out-of-stock product wastes the agent's customer's time, so AI engines deprioritise sources with stale feeds.
Merchant programme
An AI assistant operator's onboarding process for merchants who want their catalogue read.
Examples: OpenAI's ACP merchant programme, Google's UCP coalition. Submission requires a compliant feed, a capability profile, and verification of your business identity.
Why it matters. Until you're admitted to a merchant programme, the corresponding AI assistant cannot buy from you — only recommend you. Programme admission is a one-time effort with compounding returns.
Conversion & measurement
Once AI traffic reaches your site, how you turn it into revenue and prove the work paid back. Different rules apply.
Server-side attribution
Tracking which sales come from which sources by recording the data on your own server.
Instead of relying on third-party cookies or browser pixels — which are blocked by browsers, AI agents, and increasingly users — server-side attribution captures the relationship between visitor and purchase using first-party data on your own infrastructure.
Why it matters. A meaningful share of AI-referred traffic doesn't load tracking pixels at all. If you're only measuring with browser pixels, AI traffic looks like phantom direct visits — and you'll under-credit the channel that's actually working.
Third-party pixel
A tracking script loaded from another company's domain to record conversions.
Examples: Meta Pixel, Google Ads conversion tag, LinkedIn Insight Tag. These pixels were the standard of the 2010s and 2020s but are increasingly bypassed by AI agents, ad blockers, browser privacy modes, and users opting out via cookie banners.
Why it matters. Pixels still work for some channels, but cannot be your only attribution layer. Pair them with server-side attribution for full coverage.
CRO (Conversion Rate Optimisation)
Redesigning landing pages and journeys to turn more visitors into buyers.
The discipline of measuring what works, A/B testing alternatives, and shipping the winners. AI-referrer CRO is a specialised subset focused on visitors arriving from AI assistants — different intent profile, different friction tolerance.
Why it matters. A 10% conversion-rate lift compounds with everything else: better citations, better feed, better landing experience. Rex Commerce® treats CRO as the third leg, after AEO and discovery.
A/B testing
Showing two versions of a page to different visitors and measuring which converts better.
The standard method for evidence-based optimisation. Requires enough traffic to reach statistical significance — usually a few hundred conversions per variant, depending on the metric you're optimising.
Why it matters. Without A/B testing, "improvements" are guesses. With it, you build a compounding library of validated wins specific to your business.
Catalogue freshness
How quickly catalogue changes (prices, stock, new products) propagate into AI assistant recommendations.
The retainer phase of a Rex Commerce® engagement keeps the feed and capability profile in sync with your live inventory in near-real-time. Stale data degrades AI engine trust over weeks.
Why it matters. AI engines penalise sources whose feeds drift from reality. A retainer that maintains catalogue freshness protects the visibility you've built.
Got the language. Now find out where you stand.
Free AI Visibility Check on signup. 48-hour response. Any UK business with a website and a product or service to promote.