The vocabulary of car dealer search, defined precisely.

Generative engine optimization is the practice of making a website's information easy for generative AI systems to retrieve, interpret and use as source material when they assemble an answer.

GEO does not replace SEO; it depends on it. The work is crawlable pages, unambiguous entity information, original explanations worth quoting, and corroboration on sources a model already trusts. In automotive the abbreviation is routinely confused with geographic targeting — in this glossary GEO always means generative engine optimization.

Answer engine optimization is the practice of structuring content so a specific question is answered directly, completely and early enough on the page that both people and machines can extract the answer without reading the whole document.

In practice that means one page per real question rather than one page per keyword variation, a direct response in the opening lines, then evidence, exceptions and next steps beneath it. AEO is about content structure, not about a single search feature, so it survived Google narrowing FAQ rich results.

AI Overviews are Google's AI-generated summaries shown above traditional search results, synthesized from multiple web sources with links to those sources.

Eligibility follows normal indexing: a page must be crawlable and indexable to be used. Google documents that standard robots and snippet controls (nofollow, nosnippet, max-snippet, data-nosnippet) govern how content may appear in these features.

An answer engine is any system that responds to a query with a synthesized answer instead of a ranked list of links — including ChatGPT, Google AI Overviews, Perplexity, Copilot and Claude.

The practical difference from a search engine is that visibility is not a rank position but inclusion in a generated response, and often a citation link. That makes measurement prompt-based rather than keyword-based.

Retrieval-augmented generation is an architecture where a language model retrieves relevant documents from an external source at query time and grounds its answer in those documents rather than relying only on training data.

It matters for marketing because most AI search products are RAG systems. If a page cannot be retrieved — blocked, unrendered, unclear or unattributable — no amount of model capability will surface it.

Grounding is the practice of tying a model's generated statements to identifiable source documents so the claim can be checked rather than taken on trust.

A grounded system prefers sources that state facts plainly and consistently. Pages that hedge, bury the claim, or contradict the site's own structured data are harder to ground and therefore less likely to be used.

A hallucination is a confident, fluent statement produced by a language model that is not supported by any real source.

For a dealership the business risk is a model stating wrong hours, wrong departments, wrong inventory policy or a wrong claim about the store. The defence is the same as the GEO work: publish the correct facts clearly, keep them consistent across every profile, and make them easy to retrieve.

An AI crawler is an automated agent that fetches web pages on behalf of an AI product, either to build a search index, to answer a live user query, or to collect training data.

They are distinct and separately controllable. OAI-SearchBot and PerplexityBot build search indexes; ChatGPT-User and Perplexity-User fetch a page because a user asked; GPTBot and Google-Extended relate to training and grounding. Blocking a training crawler does not block a search crawler, and vice versa — the distinction is worth getting right before writing a robots rule.

llms.txt is a proposed plain-text file at a site's root that summarizes the site and lists its most important URLs in a form a language model can read cheaply.

It is a convention, not a standard, and no major search engine has committed to using it as a ranking or retrieval input — Google has publicly said it does not use it. It costs almost nothing to publish and can help agents that do read it, but it is a supplement to crawlability and sitemaps, never a substitute.

A markdown mirror is a plain-text Markdown version of an HTML page, served at a predictable URL and declared as an alternate representation, so an AI agent can read the content without parsing markup, scripts or navigation.

It removes the boilerplate-to-content ratio problem: an agent that fetches the Markdown gets only substance. It is cheap to generate from the same content model that renders the HTML, which also guarantees the two never disagree.

Prompt-set monitoring is the practice of tracking a fixed list of representative questions across AI answer systems over time, recording whether a business appears and which sources are cited.

It exists because AI answers are non-deterministic: the same question can produce different sources on different days, so a single screenshot proves nothing. A stable prompt set, re-run on a schedule, turns AI visibility into something measurable.

Share of model is the proportion of AI-generated answers about a topic or market in which a given brand is mentioned or cited.

It is the AI-era analogue of share of voice. It is only meaningful against a defined prompt set and a defined set of systems, measured repeatedly — the absolute number matters far less than its direction over time.

In AI search a citation is the source link an answer engine attaches to a generated statement, indicating which document the claim was drawn from.

In local SEO the same word means something else entirely — a mention of a business's name, address and phone number on a third-party directory. Both senses matter to a dealership, which is why they are separated here.

Answer-first writing puts the complete, direct answer to the page's core question in the opening lines, before context, background or narrative.

It serves extraction — a snippet, a voice response or a model quoting a passage all take the top of the relevant block — and it serves readers, who decide within seconds whether the page addresses their question.

Speakable is a schema.org property that identifies which parts of a page are suitable to be read aloud by a voice assistant, using CSS selectors or XPath.

Beyond voice, it is a useful explicit hint about which block on the page is the answer — a machine-readable pointer to the passage the publisher considers self-contained.

Car dealer SEO is the work of improving a dealership website's technical foundation, inventory indexation, local signals, content and internal linking so it appears for the vehicle, model, service, parts and local searches that produce real business.

It differs from general SEO because of the operating environment: inventory that turns over daily, faceted URLs that multiply endlessly, OEM and platform constraints, multiple revenue departments under one roof, and heavy third-party script load.

Search intent is the underlying goal behind a query — what the searcher is actually trying to accomplish — usually classed as informational, navigational, commercial or transactional.

For a dealership the practical use is assigning each page one intent and one job. Someone comparing two trims, someone booking a recall repair and someone looking up a specific VIN need different pages, not different keyword insertions on the same template.

A topic cluster is a group of related pages covering one subject area, internally linked around a central page that anchors the topic.

It signals topical depth rather than isolated keyword coverage, and it distributes internal link equity to the pages that most need it. The failure mode is clustering near-duplicates, which produces competition rather than depth.

A pillar page is the comprehensive central page for a topic, covering the subject broadly and linking out to the deeper pages that handle each sub-topic in detail.

Keyword cannibalization is when several pages on the same site target the same query closely enough that search engines struggle to choose between them, splitting signals instead of concentrating them.

It is endemic on dealer sites, where templates generate many near-identical model or location pages. The fix is consolidation and clearer differentiation of purpose, not more optimization of each competing page.

Thin content is a page with little or no independent value to a reader — typically templated text with a swapped variable, scraped or spun copy, or an auto-generated combination page.

The classic dealership example is a city page identical to forty others except the place name. These rarely rank and, in volume, drag down how a site's overall quality is assessed.

A doorway page is a page created mainly to rank for a query variation and funnel visitors somewhere else, rather than to be a useful destination in its own right.

Google names large groups of near-duplicate location or keyword pages as a specific spam pattern. The distinction from a legitimate local page is whether the page contains genuinely market-specific substance.

E-E-A-T is the framework in Google's Search Quality Rater Guidelines used to assess whether content demonstrates first-hand experience, subject expertise, recognized authority and overall trustworthiness.

It is not a ranking score and cannot be optimized directly. It describes qualities raters look for, which practitioners approximate through clear authorship, documented editorial standards, accurate citations and verifiable corroboration.

Editorial standards are a published statement of how an organization researches, writes, reviews, sources and corrects its content.

Publishing them makes accountability checkable rather than asserted, and they can be referenced from structured data via publishingPrinciples and correctionsPolicy so the commitment is machine-readable.

Content freshness is how recently a page's substance was updated, and how visibly that update is signalled to readers and crawlers.

It matters most for queries where the correct answer changes — platform behaviour, search features, AI systems. A changed date with unchanged content is not freshness; it is a signal search engines have learned to discount.

Last reviewed is the date a page's content was verified as still accurate, which is distinct from the date it was last edited.

Expressing both — dateModified for edits, lastReviewed for verification — tells a reader and a parser that a page has been checked rather than merely touched.

A featured snippet is an extracted answer shown at the top of a search results page, taken directly from an indexed page and attributed to it.

Snippets are extracted, not submitted: the practical lever is writing a concise, self-contained answer immediately beneath a heading phrased the way the question is asked.

Content negotiation is the HTTP mechanism where a client states which formats it can accept and the server returns the most suitable representation of the same resource.

Relevant to AI search because an agent can request text/markdown or application/json for a URL that serves HTML to a browser, provided the server advertises and honours the alternates.

Crawl budget is the amount of crawling a search engine is willing and able to do on a site in a given period, determined by demand for the site's content and by how much load the server can absorb.

It becomes a live constraint on dealer sites specifically because faceted inventory URLs can consume most of the crawl on combinations no one searches for, leaving genuinely valuable pages discovered slowly or not at all.

Index bloat is the accumulation of large numbers of low-value URLs in a search engine's index, diluting a site's overall quality signals and consuming crawl capacity.

On dealership platforms the usual sources are filter combinations, sort parameters, session or campaign parameters, expired vehicle pages and paginated archives with no distinct value.

Faceted navigation is a filtering interface that lets visitors narrow a listing by attributes — make, model, year, price, trim, colour, mileage — with each combination typically generating its own URL.

Because facets combine multiplicatively, a modest inventory can generate an effectively unbounded URL space. Google's guidance is to decide deliberately which facet combinations deserve to be crawlable landing pages and to prevent the rest from being crawled or indexed.

A canonical URL is the version of a page a site designates as the authoritative one when the same or very similar content is reachable at multiple addresses.

It is a strong hint, not a directive: search engines can and do select a different canonical when signals such as internal links, sitemaps and redirects point elsewhere. Consistency across all of those beats a lone rel=canonical tag.

Duplicate content is substantively identical content available at more than one URL, whether within a site or across sites.

On dealer sites the usual causes are syndicated OEM model descriptions, inventory feeds shared across rooftops, and parameterized URLs. It is rarely a penalty; it is usually a dilution problem, where signals split across versions.

A soft 404 is a page that tells a user the content is unavailable while returning an HTTP 200 success status, leaving crawlers to infer the page still exists.

It is the classic dealership failure: a sold vehicle's URL keeps returning 200 with a 'this vehicle is no longer available' message. At the scale of a full inventory turnover this teaches search engines that a large share of the site is empty.

A redirect sends a request for one URL to a different URL, most commonly as a permanent 301 or a temporary 302.

In dealership migrations, redirect mapping is where organic traffic is preserved or lost. Redirecting every retired URL to the homepage is treated as a soft 404; the value is in mapping to the closest genuinely equivalent destination.

A site migration is any change to a website's domain, URL structure, platform or rendering approach significant enough to affect how search engines discover and evaluate its pages.

Dealership migrations are unusually risky because inventory URLs, department pages and third-party integrations all change at once, often on a vendor's timetable. The mitigations are a complete pre-launch URL inventory, a validated redirect map and post-launch index monitoring.

Structured data is machine-readable markup, usually JSON-LD following the schema.org vocabulary, that states explicitly what a page is about and how its entities relate.

It is a clarity layer, not an authority substitute: it should describe what the visible page actually contains. Markup describing reviews, credentials or locations the page cannot support is a structured-data spam violation and can get a site's markup ignored entirely.

An entity is a distinct thing a search or AI system can identify and reason about — a business, a person, a product, a place, a concept — as opposed to a string of characters.

Entity resolution is the step where a system decides that a website, a Google Business Profile, a LinkedIn page and a directory listing all describe the same organization. Consistent naming and external corroboration are what make that merge succeed.

sameAs is a schema.org property listing external URLs that unambiguously refer to the same entity, such as a company's LinkedIn page, Crunchbase profile or Wikidata item.

It is the most direct mechanism a site has for asserting external corroboration. An organization node with no sameAs values is asking a knowledge graph to take its self-description entirely on trust.

A knowledge graph is a database of entities and the relationships between them, used by search and AI systems to answer questions about things rather than match strings.

Core Web Vitals are Google's field-data metrics for page experience: Largest Contentful Paint for loading, Interaction to Next Paint for responsiveness, and Cumulative Layout Shift for visual stability.

On dealer sites the dominant cause of poor scores is third-party script load — chat widgets, trade valuation tools, tag managers, inventory scripts — rather than the base template.

Largest Contentful Paint measures how long it takes for the largest visible content element in the viewport to render, as a proxy for when a page feels loaded.

Interaction to Next Paint measures how quickly a page visibly responds to user interactions across the whole visit, replacing First Input Delay as the responsiveness metric.

Cumulative Layout Shift measures how much visible content moves unexpectedly during a page's lifetime, capturing the experience of a button jumping away as you reach for it.

A third-party script is JavaScript loaded from a domain other than the site's own — chat, analytics, tag management, trade-in tools, retargeting pixels and inventory widgets.

Dealership sites commonly carry dozens. Each adds network requests and main-thread work the site does not control, which is why performance work usually starts with a script audit rather than image compression.

robots.txt is a file at a site's root that tells compliant crawlers which paths they may request.

It controls crawling, not indexing: a URL disallowed in robots.txt can still appear in results if other pages link to it, because the crawler is never allowed to read the noindex tag that would exclude it. To keep a page out of the index, allow the crawl and use a robots meta tag or X-Robots-Tag header.

noindex is a directive, delivered in a robots meta tag or an X-Robots-Tag HTTP header, instructing search engines not to include a page in their index.

It only works if the crawler can fetch the page, which is why combining noindex with a robots.txt disallow on the same URL is self-defeating.

An XML sitemap is a file listing the URLs a site wants search engines to know about, optionally with last-modified dates, to aid discovery.

It aids discovery; it does not guarantee indexing. Its practical value on large dealer sites is diagnostic — comparing submitted against indexed URLs in Search Console exposes indexation problems that page-by-page inspection would not surface.

IndexNow is a protocol, supported by Bing, Yandex, Seznam and Naver, that lets a site push a notification the moment a URL is added, updated or deleted instead of waiting to be recrawled.

Because ChatGPT Search and Copilot draw on Bing's index, faster Bing discovery has a direct downstream effect on AI-search freshness. Google does not participate.

Passage indexing is a search engine's ability to identify and rank a specific section of a long page for a query, rather than judging only the page as a whole.

It rewards clear, descriptive headings and self-contained sections — the same structure that lets an answer engine quote one block of a page without dragging in surrounding context.

Internal linking is the practice of connecting a site's own pages with descriptive links, shaping how crawlers discover pages and how importance is distributed across the site.

It is the most under-used lever on dealership sites, where high-value department and content pages are often reachable only through a mega-menu or a script-driven component a crawler treats as low signal.

An orphan page is a page with no internal links pointing to it, discoverable only through a sitemap or an external link.

Anchor text is the visible, clickable text of a link, which both readers and search engines use as a description of the destination.

Descriptive anchors outperform generic ones. A site where most internal links read 'learn more' has discarded a free, sitewide relevance signal.

hreflang is an annotation declaring which language and regional audience a page is intended for, so search engines serve the right variant.

On a single-locale site a self-referencing hreflang plus x-default is still useful: it states explicitly that this is the only and default version rather than leaving the pairing to inference.

A vehicle detail page is the page for one specific vehicle in a dealership's inventory, identified by VIN and carrying its photos, price, specification, history and enquiry actions.

VDPs are the highest-intent pages on a dealer site and also the most volatile — created and retired continuously as stock turns. How a sold VDP is handled is one of the consequential technical SEO decisions on a dealership website.

On a dealership website, a search results page is the filtered inventory listing that shows multiple vehicles matching a shopper's criteria, linking through to individual vehicle detail pages.

SRPs are where faceted navigation lives, so they are simultaneously the main discovery path into inventory and the main source of crawl waste. A small number of SRPs with durable search demand deserve to be indexable landing pages; the long tail of combinations does not.

A vehicle identification number is the unique 17-character code assigned to an individual vehicle, used to identify it across inventory, history, recall and registration systems.

For SEO it functions as the stable key behind a VDP, which makes it the natural basis for URL structure, deduplication and redirect mapping during a migration.

Google Business Profile is the free listing that controls how a business appears in Google Search and Maps, covering name, category, location, hours, attributes, photos, posts and reviews.

For dealerships the structural question is departments: sales, service, parts and collision can be eligible for their own listings under specific Google rules, which changes how each department competes in local results.

A department listing is a separate Google Business Profile for a distinct department of a larger business — such as a dealership's service or parts department — where Google's guidelines permit one.

Eligibility depends on the department being distinct in customer-facing terms, typically with its own entrance, hours and phone line. Creating ineligible department listings risks suspension of the profiles involved.

The local pack is the map-and-listings block shown for queries with local intent, typically presenting three nearby businesses above the organic results.

Google describes local ranking as a combination of relevance, distance and prominence. Distance is largely fixed; relevance and prominence are where the work is.

A local citation is a mention of a business's name, address and phone number on a third-party site such as a directory, marketplace or review platform.

Citations corroborate that the business exists as described. Inconsistent details across citations are a common, quietly damaging problem for dealer groups that have rebranded or moved.

NAP consistency is the practice of publishing a business's name, address and phone number in exactly the same form everywhere it appears online.

Inconsistency forces entity-resolution systems to guess whether two records describe one business or two. For AI answers this is a direct accuracy risk — a model may confidently return a superseded address.

A service area is the geographic region a business serves, as opposed to the location it operates from.

For a dealership the two differ meaningfully: the store has one address but draws customers from a much wider trade area, which is what local landing pages should reflect — with genuinely market-specific content, not templated city swaps.

Attribution is the assignment of credit for a conversion to the marketing touchpoints that preceded it.

In automotive the same shopper is routinely counted by the OEM tier-three vendor, the website provider, the CRM and the paid partner, so summed vendor reports can exceed actual sales. Reconciling that overlap is usually worth more than optimizing any single channel.

Incrementality is the additional business a marketing activity generates beyond what would have happened anyway.

It is the question branded paid search rarely gets asked: how much of that converted demand would have arrived organically. Answering it usually requires a holdout or geo test rather than a platform report.

Branded search is a query containing a business's own name or a close variant, as distinct from non-brand queries describing a need.

The split matters because branded volume mostly reflects demand created elsewhere, while non-brand visibility is what genuinely expands the funnel. Reporting that blends them can make a flat program look like growth.

Google Search Console is Google's free service reporting how a site performs in Search, including queries, impressions, clicks, indexing status and structured-data validity.

It is the only source of first-party query data Google provides, and its Page Indexing report is the fastest way to diagnose dealer-site indexation problems at scale.

Conversion rate is the share of visitors who complete a defined action, such as submitting a lead form, starting a trade valuation or booking a service appointment.

It is only interpretable when segmented by intent and page type — a rise driven by low-intent traffic can mask a decline on the pages that matter.

Automotive marketing is conventionally split into three tiers: Tier 1 is national manufacturer advertising, Tier 2 is regional dealer-association advertising, and Tier 3 is the individual dealership's own marketing.

The tiers frequently compete for the same queries with the same brand terms, which is why a store's paid and organic performance cannot be judged without knowing what is running above it.

Co-op advertising is manufacturer funding that reimburses a dealership for qualifying advertising, subject to the OEM's creative, media and compliance rules.

The rules shape what a dealership can practically run and with which vendors, so co-op constraints belong in any honest channel or vendor assessment.

Programmatic advertising is the automated buying and placement of digital ad inventory through real-time auctions and demand-side platforms.

First-party data is information an organization collects directly from its own customers and systems — CRM records, service history, website behaviour, transactions.

It is the durable asset as third-party cookies and cross-site identifiers decline, and it is the raw material that makes a custom AI product useful rather than generic.

Identity resolution is the process of determining that records from different systems refer to the same person, producing a single unified profile.

In a dealership the same customer commonly exists several times over — a service record, a sales lead, a website session, a marketplace enquiry — which distorts both reporting and targeting until the records are reconciled.

A custom AI product is software built around a specific organization's workflows, data and decisions, rather than a general-purpose tool adapted after the fact.

The defensible part is rarely the model — it is the proprietary data, the evaluation that catches failures before customers do, and the fit to how the work actually happens.

Human in the loop is a design pattern where a person reviews, approves or corrects an AI system's output before it takes effect.

It is the appropriate default anywhere output touches price, brand voice, compliance or a customer commitment — which covers most dealership marketing applications worth building.

Prompt engineering is the practice of designing the instructions and context given to a language model so it produces reliable, appropriate output for a task.

In production it matters far less than the surrounding system — retrieval quality, evaluation and review usually determine whether an AI workflow is trustworthy.