Glossary: how MentionBench names the work
MentionBench Consulting LLC sells one offer—AEO/GEO Engagement—people + acceptance criteria + accountability. Inside that Engagement, method has three layers: diagnosis, on-site improvement, and bounded off-site. Success is mention rate and citation rate on a locked prompt panel (baseline → retest after the observation window). We do not promise rank position, share-of-voice leadership, every AI answer, or permanent occupancy.
Updated 2026-08
Method cadence: see Resources · What changed (Aug 2026).
This page is the naming layer: framework definitions with include/exclude (or acceptance), then a short term list. Buyer objections live on FAQ. Path contrast lives on Compare.
Framework naming
AEO/GEO Engagement
AEO/GEO Engagement is MentionBench’s single commercial offer: scoped diagnosis, on-site improvement, and bounded off-site, with a named owner, published acceptance criteria, and locked-prompt baseline → retest of mention and citation rates.
Includes: the three method layers to the agreed written scope; human accountability at milestones; a locked prompt panel with baseline recorded and retest after the observation window.
Excludes: monitoring SaaS, always-on dashboards, unlimited writing or PR, a full-industry always-on agency retainer, and any promise of rank position, SOV #1, every AI answer, or permanent occupancy.
Acceptance: agreed pages, items, and off-site actions meet the published bar; scope does not auto-expand. Retest is accountability for before/after—not a guarantee that rates will rise.
Diagnosis
Diagnosis (Diagnosis landing) is the first method layer: a bounded review of the site and competitors that produces a readiness conclusion, a competitor gap table, and a prioritized fix list—plus a locked prompt panel with baseline recorded.
Includes: coverage of existence/structure/cross-platform consistency signals; a one-time competitive mention-style contrast; an ordered, actionable fix list.
Excludes: monitoring SaaS, always-on dashboards, exposing an internal tool stack, and unfolding a full public scorecard product.
Acceptance: the client can state what to fix first, why, and what “done” looks like; the prompt panel and baseline are written down. Diagnosis may be the bounded first step; later layers expand only by agreement.
On-site improvement
On-site improvement (On-site landing) is the second method layer: scoped changes on the site (structure such as schema/FAQ/entity clarity, cross-platform wording consistency, and content within an agreed article count negotiated in the Engagement).
Includes: implementation inside the agreed page/item list and agreed content count.
Excludes: unlimited writing, unlimited full-site rewrites, and promises of rank position, SOV #1, or appearing in every AI answer.
Acceptance: agreed pages and items meet the bar; work outside the written scope does not auto-start.
Bounded off-site
Bounded off-site (Bounded off-site landing) is the third method layer: third-party corroboration within an agreed action count (for example guest placements, relevant referring sources, directories, bounded media outreach, and verifiable testimonials/case facts aligned with the site).
Includes: an action list (target URL + type) and execution record within the agreed count (receipts or live links).
Excludes: unlimited PR, always-on full-service agency retainers, content factories, always-on monitoring, and promises of rank position, SOV #1, every AI answer, or permanent occupancy.
Acceptance: agreed actions are completed and sources are verifiable.
Assess / Improve / Review (aliases only)
Assess, Improve, and Review are optional narrative aliases that map 1:1 to the three method layers—Assess = Diagnosis, Improve = On-site improvement, Review = Bounded off-site. They are not three separately priced sellable tiers. Baseline → observation-window retest lives under Baseline and retest, not under the Review alias. MentionBench sells one AEO/GEO Engagement. Pricing pages should not hang empty tier cards for these names.
How we measure and accept
Locked prompt panel
A locked prompt panel is the written set of buyer-relevant prompts used for baseline and retest. The same panel is reused after the observation window so mention rate and citation rate stay comparable.
Mention rate
Mention rate is the share of prompts on the locked panel where the brand is named or recommended in the observed AI answers—for the models and surfaces agreed in writing.
Citation rate
Citation rate is the share of prompts on the locked panel where the brand’s content or an agreed property is cited or attributed as a source—not merely named without a trail.
Baseline and retest
Baseline and retest means rates are recorded before agreed work ships, then measured again on the same locked panel after the observation window. Retest is accountability for the before/after; it is not a guarantee that rates will rise.
Observation window
An observation window is the agreed period after scoped work ships during which we watch for directional movement on the locked panel. It is not a promised calendar date when the brand will “appear” in every answer.
Acceptance criteria
Acceptance criteria are the published include/exclude and done-bars for agreed pages, items, and off-site actions. Delivery acceptance is met when those bars are hit; success accountability additionally uses locked-panel mention and citation rates (baseline → retest).
Short industry terms
Answer Engine Optimization (AEO)
Answer Engine Optimization (AEO) is the practice of earning clear, citable presence in AI-generated answers—entity clarity, answer-ready structure, and accountable execution—not a promise of guaranteed placement.
Generative Engine Optimization (GEO)
Generative Engine Optimization (GEO) is another common name for the same kind of work as AEO. Surfaces and labels change; MentionBench uses both terms because buyers search both. The Engagement method is the same.
Mention vs citation
A mention is being named or recommended in an AI answer. A citation is being used or linked as a source. Both can be measured on the locked prompt panel; optimizing for one is not automatically optimizing for the other.
Entity consistency
Entity consistency means the brand’s name, category, and core claims stay aligned across the site and major third-party profiles so models receive one clean identity signal—not three conflicting ones.
Citation readiness
Citation readiness means pages are structured so models can extract and ground claims: clear definitions, answer-first blocks, verifiable statements, and machine-readable structure where agreed—without treating any single tactic (such as schema alone) as a complete Engagement.
Boundaries
Monitoring SaaS vs hire
Monitoring SaaS produces measurement, prompt sets, and tracking outputs. Hiring MentionBench is scoped execution with a named owner and a published acceptance bar—diagnosis, on-site improvement, and bounded off-site—then locked-prompt baseline → retest. Keep a tool if it helps your team; hire when insight without ownership is the bottleneck. See Compare.
People + acceptance + accountability
MentionBench is a consultancy path: people who prioritize and ship, acceptance criteria written in advance, and accountability at milestones and at retest—not another DIY dashboard and not an unlimited content retainer.
What we do not guarantee
No ethical AEO/GEO consultancy can guarantee appearance in any specific AI response. Models and surfaces change. What we own is a rigorous, bounded process and accountability for mention rate and citation rate on the locked prompt panel (baseline → ship to include/exclude → retest after the observation window). We do not promise rank position, SOV #1, every AI answer, or permanent occupancy—and retest is not a guarantee that rates will rise. The same boundary is stated on FAQ and Compare.
Next step
Ready for a bounded first step? Book an AEO/GEO Engagement consult—diagnosis priorities, scoped next work, and clear acceptance from a named consultant.
[email protected] · Home · FAQ · Compare · Self case study