AI readiness and answer discoverability

What is answer discoverability?

Answer discoverability is whether an AI engine can find your brand, understand what you offer, and represent you accurately when a customer asks it a question. It has two layers: the engine has to get you right when someone asks about you by name, and it has to name you when the question is about your category and nobody gave it your name.

Those are different tests, and a brand can pass some and fail others: whether the engine can retrieve the information at all, whether it identifies the right entity, whether it understands the information accurately, whether it cites or recommends the source, and whether what it has is current.

His published formulation is that GEO and AEO are not new disciplines. They are new front doors into the same business of being found and understood. The fundamentals did not change. The surfacing did. The tradeoff is real: an engine can make a brand visible without sending it a visit, and deciding what that visibility is worth, then measuring it on its own terms, is part of the work.

What does AI readiness require?

An agent is now an audience. A person asks a question in their own words, and a model goes looking on their behalf. It reads what it can reach, trusts what is consistent, and repeats what it can verify. If the underlying information is fragmented, stale, or locked in formats the agent cannot parse, the answer it gives is built on whatever it could find instead.

Readiness is the work that prevents that: content and data that are governed, structured, and machine-readable, with the stable facts separated from the volatile ones and access set deliberately rather than by accident. When an agent gets a brand wrong, the cause is usually underneath the content, in the data it was assembled from. Not a technology failure. A data failure, with a new audience watching.

How I work

Fundamentals first. His published position is that SEO, GEO, and AEO are just UX done right: clarity over clutter, focus over volume, brand reach over raw traffic. The three disciplines share one goal. Make the experience so useful that people, and now the agents asking for them, can find it, understand it, and trust it.

Map intent to pathways. At A Place for Mom, his published line was that they didn’t chase keywords. They mapped user intent to pathways through the product. The same discipline governs answer readiness: the questions a customer actually asks, in their own words, mapped to content that answers them completely.

Run the readiness sequence. The sequence is the one he published in Data First, AI Second, Now With Agents As The Proof: govern the graph, resolve identity, separate the stable facts from the volatile offer state, expose structured content, set intentional agent access, and measure agent journeys separately from human ones. It is a published framework, stated here as he published it, not a client delivery record.

Run the Monday test. His published diagnostic: take the questions your customers asked last week, in their own words, and ask them of the engines your customers use. Where the answer is wrong, missing, or built on a competitor’s facts, you have found a readiness gap. Not a prompt problem. A readiness problem.

Audit in categories, then fix. AI search today is a Wild West next to the old single-player SEO world, where one engine set the standard and best practice. With engines proliferating, fundamentals still have to be right, and experimenting across multiple strategies matters more, not less. And not all citations are created equal: a serious scoreboard scores a recommendation, a factual mention, a comparator or uncited mention, a wrong or stale representation, and an absence as different things, because they are. The audit below is this beat, run on his own presence.

The proof, in three separate parts

Sotheby’s: the organic search record. Conductor ran organic search intelligence; Algolia ran on-site search and recommendations, and the two are never the same job. As the measurement of that tenure reported it: organic traffic up 167%, search click-through up 48.7%, consignment search placement doubled, consignment ad spend fell 50%. This is the foundation the answer-engine era rewards: a single governed data layer under findable content. The method behind it is his published one: build something genuinely useful, make it easy to find and easy to use, and the search results follow the usefulness. Read the case study.

A Place for Mom: the diagnosis. The settled diagnosis was a platform and governance problem: schema and structure inconsistent from page to page, so quality varied and search engines read the inconsistency. It was not a content problem. He led the rebuild of the content platform and its governance, and SEO market share rose from 26% to 68%. Read the case study.

Lindblad Expeditions: search tooling in the stack. Conductor was part of the stack at Lindblad. Its presence is recorded here without borrowing the replatform’s results: the Lindblad outcomes were booking-engine outcomes, and this page does not re-badge them as search proof. Read the case study.

The frameworks, in public. SEO, GEO, and AEO Are Just UX Done Right (September 2025) sets the one-discipline position. Data First, AI Second, Now With Agents As The Proof (November 2026) sets the readiness sequence and the Monday test. Own Your Harness supports both: agent governance is the control layer the readiness work depends on.

The audit: method, findings, actions. In November 2026 he designed and ran an audit of his own presence: six questions, asked across five AI surfaces, with every appearance scored in the categories above. The sessions were signed in as him, which personalizes results, so the findings are stated as what they are: a method-level read, not a clean public baseline. A signed-out re-run is scheduled for December 14, 2026.

The findings: the identity layer held. Asked about him by name, the engines had the current role, the employer, and the career history right. The category layer was the gap. Asked about the category, with no name given, the answers named analysts and vendors; a practitioner appeared only occasionally. That gap is not fixed by polishing a profile. It is fixed by citable proof on the open web.

The actions: two fixes were executed the same day. A removal was requested for a data-broker profile that still listed a former employer as his current one; the request is filed and the deletion is being verified. And on the engine with the stalest record of his career, a corrected baseline was established and verified in a fresh session.

The point

The parts stay separate on purpose. The search record is history, the frameworks are published positions, and the audit is a method, run once so far, with a signed-out re-run scheduled for December 14, 2026.