Glossary — search, AI discovery, and your audit kit

Field language

GEO (Generative Engine Optimization)

The practice of making your public site clear enough that search engines and AI assistants can find and accurately describe your business. GeoSummit scores on-page GEO-related Digital Visibility Score readiness — not whether any assistant will cite you. See Methodology — Is this really AI? and Sources. For RAG (a different architecture): RAG.

AEO (Answer Engine Optimization)

Closely related to GEO: structuring facts, questions, and answers so engines can quote clear responses (hours, services, pricing ranges you already publish, contact paths). Still a readiness discipline — not a promise of featured answers or AI Overview placement.

RAG (Retrieval-Augmented Generation)

A technical architecture where a large language model retrieves document chunks to compose answers. Customer audits are not RAG scoring. GeoSummit uses a documented, rule-based pipeline with sector-weighted checks and templated exports. We do not send your Customer Data to third-party generative models for scoring or report generation. Is this really AI?

Digital Visibility Score readiness

Your overall number is a weighted sum of on-page categories: how clearly machines can identify and describe your business from your homepage and limited supporting public pages (identity, contact consistency, answerable copy, trust cues, crawler access, and related hygiene — prioritized by industry). It is not a ranking forecast, traffic promise, Maps pack position, lead guarantee, AI-citation probability, or predictive citation claim. After you publish fixes, re-run: score movement means your Digital Visibility Score improved. Free remains homepage-centric — not a multipage pack score. What the score means · SMB honesty bullets · What the score is not

E-E-A-T

Experience, Expertise, Authoritativeness, and Trust — language Google uses for content quality expectations. GeoSummit looks for on-page trust and credibility signals appropriate to your industry (clear identity, experience cues, consistent facts). Higher bands prefer a verifiable citation path — a named source, <cite>, or a real link near the author/source cue — not bare “author” words or date stamps alone. That is checklist hygiene — not a Google ranking certificate, “E-E-A-T score” sold by third parties, or proof any assistant will cite you. Related references live on Sources.

NAICS / industry sector

GeoSummit maps businesses to the U.S. Census NAICS 2022 taxonomy (twenty parent sectors) plus specialized SMB subcode packs where available. Industry rules change weights, vocabulary, and schema defaults — a law firm is not scored like a restaurant. Methodology — industry rules · Census / NAICS on Sources

Sector-weighted scoring

Category priorities differ by industry pack. Professional services may emphasize expertise and trust cues; retail may weight product and offer clarity differently. Weights are proprietary and version-stamped; this glossary describes the idea, not a pasteable formula. Construct overview

Site signals

Schema (JSON-LD)

Machine-readable markup on your site (often in a <script type="application/ld+json"> block) that describes your business type, services, FAQ, and contact fields for search and AI tools. GeoSummit checks whether that markup is present, sector-appropriate, and consistent — not whether Google will rank you. Vocabulary: Schema.org. Bibliography: Sources.

Structured data

Broader term for machine-readable facts on a page (including JSON-LD schema). Supports rich results and clearer entity understanding when accurate. Structured data is not a ranking guarantee; hollow or mismatched markup can hurt trust more than it helps.

Brand profile links (Entity Authority / sameAs)

On-page links from your site or schema sameAs to profiles you own (for example Google Business Profile, major social or directory hosts). GeoSummit’s Entity Authority row measures that on-page link hygiene so machines can confirm which brand the page is about. Empty or thin sameAs usually lands near a mid-band (~55) honest absence — we never invent profile URLs. This is not a Knowledge Graph panel score, off-site authority graph, or proof that two social icons equal entity authority. SMB honesty

NAP

Name, Address, Phone — the contact facts customers and crawlers expect to match across your footer, contact page, and schema. Never invent NAP; only publish numbers and addresses you own. On-site NAP is tracked separately from optional Maps comparison — see dual-track.

robots.txt

A small public file that tells crawlers what they may fetch. If it blocks the paths we need, you may get a limited package instead of a full score. Fixing robots is often a CMS or host setting — not a marketing slogan.

Direct answers & question headings

Clear prose and FAQ-style H2/H3 phrasing that machines can quote (who you are, what you offer, where you serve, how to contact you). Higher bands need extractable Q&A — a usable answer next to the question — not question-shaped headings alone, hero marketing blurbs, or hollow one-line FAQ theater. Part of answerability on the checklist — not proof you will appear in AI Overviews or that any assistant will cite you. Visible conversational Q&A can score complete while optional FAQPage JSON-LD markup remains a separate residual — markup-only gap ≠ missing FAQ content. Hollow/spammy Q&A can hit the answer-quality limit. SMB honesty

Customer reviews (on-page)

On-page review signals on your public site (visible review content and related markup when present). Stronger credit prefers a real review path — third-party platform context or a product-path review markup trail — not only self-published “5-star” marketing copy. Empty or absent reviews usually land near a mid-band (~55) as honest absence; we never invent reviews or AggregateRating to green the score. Still Digital Visibility Score readiness — not Maps pack or reputation-agency scores. SMB honesty

Indexability & crawler access

Whether public pages allow search engines and similar crawlers to fetch and use content (robots rules, noindex signals, thin/parked shells). If we cannot read the site, we do not invent a full scorecard — see limited package. Confirmed noindex or full crawl blocks also apply hard overall ceilings so blocked discovery cannot look fully ready. SMB honesty

Homepage-centric crawl

The live audit focuses on your primary public URL first, with optional thin same-host recovery — not every URL on a large site and not a free multipage pack or free multipage composite score. A denser multipage modeling or self-audit run is not the same as free Top-3 results. That keeps free (and typical paid single-site) paths honest and fast; deep multi-page enterprise crawls are out of current free-beta scope. Owner summary: SMB honesty · Limitations

Core Web Vitals (and what we check)

Google’s lab/field page-experience metrics (LCP, INP, CLS, and related). GeoSummit’s public scorecard emphasizes on-page content and entity clarity plus basic mobile/page hygiene — it is not a substitute for Search Console or Lighthouse lab CWV reports. Do not treat our checklist as a CWV certification.

Scoring honesty

Fail-closed score

Your official overall treats missing category scores as zero (fail-closed), not as a perfect 100. Incomplete evidence cannot inflate overalls. That keeps incomplete crawls from looking greener than they are. Related: honest absence mid-bands (for example empty reviews or sameAs) are not invented green scores; multi-location is out of scope on a true single-site. SMB honesty · How we stay honest

Limited package

When bot protection, robots blocks, SPA shells, or thin public pages prevent a full read, you get limited guidance instead of a pretend full scorecard. That is intentional honesty — not a soft paywall trick. After free beta, paid audits are designed so you only pay when we can read what you submitted.

Dual-track (Maps / directories)

Some signals (Maps readiness, external directories) are shown for honesty and progress language but are not silently rolled into your overall until evidence and operator gates say so. Free beta does not live-scrape Google Business Profile via Places API; optional owner-provided Maps links support comparison only.

Answer-quality limit (soft-cap)

If the audit detects hollow or spammy Q&A patterns, related answer categories may be limited (for example at most 70/100) so stuffed FAQs cannot inflate your overall. This is a checklist anti-gaming rule — not a Google or AI “penalty notice,” and not a predictive citation penalty. Layout/trust limits for stuffed lists or thin author markers stay off in production until a later methodology version.

Soft checklist adjustments

Small, capped changes to a few category scores when service pricing education is missing or a strong online booking surface is detected. Still Digital Visibility Score readiness only — not rank or leads.

Export / QA gate

Automated checks before a full kit ships: contact consistency, sector-appropriate schema, valid JSON-LD paste blocks, CMS placement consistency, and bundle parity. Critical failures can block export so you do not receive a broken pack. Methodology version is stamped on every deliverable. QA gates

Your pack

Fix difficulty (plain English)

Owner-facing packs use plain labels. Internal pack files may still store short codes (A–D) for agents and QA — you rarely need them.

• Easy technical fix (internal: A) — quick wins your AI can propose now.
• Moderate fix (internal: B) — CMS-constrained or medium risk (plugin, plan, or support ticket path).
• Deeper improvement (internal: C) — business facts only you can supply (hours, phone, regulated claims).
• Advanced / later (internal: D) — specialized, lower priority, or human-only when safety preconditions are missing.

Free Top-3

On the free path you see up to three priority technical gaps on the page when the crawl succeeds. Free is homepage-centric (primary URL + optional thin recovery) — not a multipage pack. Optional email may send a small pack later — not the full hub ZIP. Digital Visibility Score readiness only — not leads, rankings, or AI citations.

Easy Pack

Optional mid pack: every remaining quick technical fix for one URL, step by step — if you DIY after free. Does not re-sell free Top-3 as paid filler. List pricing appears on the site; checkout timing follows free-beta policy. See product format

Full package

Complete professional hub: scorecard, ordered fix guide, download pack, and AI brief (list price after free beta). Diagnosis is automated; you (or your developer) publish changes. Pricing

AI brief

Paste-ready brief for ChatGPT, Claude, Gemini, Grok, or a similar tool — included in the Full package. GeoSummit does not auto-publish to your CMS; you (or your developer) still publish.

AI implement kit

Your hub’s implementation_pack.json + implementation_agent.md with paste-ready steps. Implement assist means code for your AI — not a human services SKU on the free hub.

Prove it (re-audit)

After you publish fixes: mark done in owner_apply_checklist.html, re-run the audit (deep link may prefill your baseline Digital Visibility Score), and record Digital Visibility Score change on outcome_card.html. Score movement means your Digital Visibility Score improved — not leads or rankings.

bundle_manifest.json

Root file in a customer bundle that ties PDF, HTML, CSV, and step anchors to one pipeline run (bundle_version, files, html_anchors). Supports traceability and re-runs — not a marketing claim. Reproducibility

Competitor benchmark

When a market area is available, the pipeline can audit up to five nearby industry peers with the same rules and include comparison context in the PDF. Requires geocoding success for map context; not a national average ranking. Pipeline detail

CMS recipe

Fix steps matched to the site builder detected on your live site (for example WordPress, Squarespace, Wix, Shopify). Export QA checks that recipes do not bleed across builders. You still apply and publish the change.

Knowledge catalog (AI-era shelves)

Knowledge catalog

Owner view of eight public-fact “shelves” on your website (identity, offers, coverage, pricing, how to engage, proof, answers, machine access). Checklist coverage of what search and AI can read — not a knowledge-graph product, agent API, or ranking guarantee. Free may show a one-line teaser; Full includes the shelf rollup in the pack. Methodology

Pricing education (transparency)

Public cost ranges, what drives price up or down, and estimate disclaimers — using figures you already quote. Helps AI answer “how much does it cost?” We never invent prices for your business. Free may surface a related priority gap when ranked high; Full shows the level and next steps.

Booking surface

Whether visitors can find a working book, schedule, or request-quote path on your site (real URL, known scheduler embed, form, or phone) — not a dead button. Not the same as agentic AI booking.

Full bibliography and instructional-design references (presentation of owner results — not scoring formulas): Sources. Construct validity and non-claims: Methodology.