Answer Engine Optimization Agency: We Fix Discoverability Below the Waterline

Answer engine optimization is decided below the surface of the page

Answer engine optimization is the practice of making an organization the source that AI systems extract, cite, and name when a buyer asks a question. AI engines see only the surface of your content system: the answer they can lift. Whether they lift yours is decided underneath, in how the content is structured, what platform produces it, and what operation keeps it current. Our complete guide to AEO covers the discipline; this page covers what we do about it.

We are a headless CMS agency that does AEO, and that order matters. Most AEO offers score attributes of pages. We model the system that produces discoverability, because that is where the fixes hold. The doctrine in one line: measure at the surface, fix at depth.

Every publish event serves two audiences now. People judge the rendered page. AI systems read the whole stack underneath it. McKinsey found that in many cases a brand's own site is only 5 to 10% of the sources AI search references (McKinsey, October 2025), so the work also covers how you are described everywhere else.

Ranking and citation are two systems, and the second one is taking the buyer

68.01% of US Google searches ended without a click in the first four months of 2026 (SparkToro, June 2026). Over the same period Google did not lose its audience: heavy Google use on US desktop devices rose from 84% in Q1 2023 to 87% in Q1 2025 (SparkToro and Datos, August 2025). The answer moved onto the results page and into AI engines, and the click stayed there with it.

So a page can hold position one and still be absent from the answer a buyer reads. Only 12% of the links ChatGPT, Gemini, and Copilot cite appear in Google's top 10 (Ahrefs, August 2025). Gartner now describes answer engine visibility tools as "a baseline martech necessity" and reports 9% of buyers discovering solutions through GenAI against 8% through search (Gartner, March 2026).

Surface fixes alone do not move citation; structure and depth do

Adding schema markup to pages that were already being cited produced no measurable change in AI Mode or ChatGPT citations and a 4.6% decline relative to matched controls in AI Overviews, in Ahrefs' intervention test of 1,885 pages against 4,000 controls (Ahrefs, May 11, 2026). Schema is hygiene we apply at the template level, and it is not the lever.

What the citation data does reward is where the answer sits and how it is framed. 44.2% of LLM citations come from the first 30% of a page, and 78.4% of citations tied to questions come from headings, across 18,012 verified citations (Kevin Indig, Search Engine Land, February 2026). Length does not enter into it: Ahrefs found a 0.04 Spearman correlation between word count and AI Overview citation across 174,048 pages (Ahrefs, December 2025).

Those are content and structure decisions. They hold across a library of thousands of pages only when the content model, the templates, and the publishing operation produce them by default. That is why the platform and the operation are inside the scope of AEO, and why we treat AEO as a CMS selection criterion.

The Waterline Framework: five layers, five owners, five sets of metrics

The Waterline Framework is our model for AI discoverability at enterprise scale. Five layers run from the surface to the seabed. The deeper the layer, the slower it changes and the more it decides. Each layer has a named owner and metrics that owner reports, so accountability runs the full depth rather than stopping at the content team.

The rule at layer one is the one that saves budgets. Symptoms appear at the surface. Causes never live there. We treat remediation aimed at layer one alone as theatre, with the Ahrefs schema test as the measured case for that view.

The Sounding is where every engagement starts

A Sounding is our depth audit. Over three to four weeks we measure the surface and inspect every layer beneath it, and we hand back a one page depth chart: where you stand at each layer, the owner, the metric, and the first remediation in priority order.

1. Surface measurement. We agree a priority query set with you, 20 to 40 commercially important questions, and test each in ChatGPT and Gemini three times. A query counts as won only when you are named in three of three samples on both engines. This is the baseline everything else is measured against.
2. Content and structure inspection. We crawl the library, sample the top commercial pages, and score answer placement, heading framing, authorship, source density, schema coverage, and field population.
3. Platform and operations inspection. We read the content model and templates, and we map how a page moves from brief to publish and who owns accuracy after it ships.
4. Depth chart. One page, five layers, owner and metric per layer, remediation sequenced by depth. Fixes at the platform layer come first when they exist, because every layer above inherits them.

After the Sounding, remediation runs as a program. Structure and platform fixes ship through our headless CMS practice. Content and operations fixes ship through our content operations practice. The same priority query set is re tested every 30 days.

What CMOs, CDOs, CIOs, and COOs each do first

CMO. Write down the twenty questions a buyer asks before shortlisting you and test them in ChatGPT and Gemini this week. Count the queries where you are named. That count is your layer one number, and 40.6% of AEO practitioners report they have no reliable way to produce it (Minuttia, June 2026).

CDO. Audit entity consistency. Compare how your organization, products, and leaders are described on your site, on partner directories, on LinkedIn, and in the last twelve months of press. AI systems draw from all of them, and the McKinsey 5 to 10% figure says most of the sources are not yours.

CIO. Count how many AI ready signals your platform produces by default and how many require a developer per page. Template level schema, author entities in the content model, and clean HTML delivery are the three to check first. If all three are manual, the platform is on the AEO critical path.

COO. Name the owner of each Waterline layer on the org chart. Five layers, five names, five metrics reported monthly. In our experience the layer with no owner is the first to decay.

Three ways the next 18 months play out

Scenario 1, surface only. The team adds schema, an llms.txt file, and FAQ blocks to existing pages on a legacy platform. Watch for: no change on the priority query set after two 30 day re tests, and the Ahrefs schema result reproduced on your own library.

Scenario 2, depth first. The team runs a Sounding, fixes structure and platform where the depth chart says to, and rebuilds the top 20 to 40 commercial pages answer first with named authors. Watch for: the first priority queries moving from gap to named within two quarters, and citation share on the set climbing after that.

Scenario 3, wait for the market to settle. The team holds until AI search behaviour stabilizes. Zero click has already reached 68%, GenAI discovery has already passed search for software buyers at 9% to 8%, and Gartner already calls the tooling a baseline necessity. Watch for: competitors named on your category queries.

We measure at the surface and report monthly

Lack of measurement and attribution is the top challenge for 40.6% of AEO practitioners (Minuttia, 599 respondents, June 2026), so measurement is the first thing we build. Our AEO measurement framework defines the numbers: brand mention rate, citation share, query coverage, accuracy rate, and differentiator mention rate on your priority query set, plus the AI referral channel in GA4 and Search Console impressions and position.

The reason to hold the bar this high is what those visitors do. On Ahrefs' own site, AI search referrals were 0.5% of visitors and 12.1% of new signups, a conversion rate 23 times organic search (Ahrefs, June 2025). That is a single site's data and we treat it as an indicator, not a forecast. It is also the only kind of visitor a buyer becomes after an AI engine has already made the shortlist for them.

Who we do this for

We deliver AEO and content operations for Westman Communications, a telecommunications cooperative, and we have run the digital presence of First Canadian Place for Brookfield Properties for more than twenty years. Our headless practice has shipped production work for Mitsubishi Electric Canada, Oxford Properties, and Financial Horizons Group on Contentful, Storyblok, Agility CMS, and Magnolia.

We are a Contentful Silver tier partner, a Storyblok Silver Partner, and an Agility CMS Implementation Partner. Dotfusion Technology, Inc. has been a Certified B Corporation since May 2021 with a B Impact score of 89.4 (B Lab). We have been building for large organizations from Toronto for 27 years.

Method note

Analysis as of September 17, 2026. Every external figure on this page was fetched from its source on that date and is linked inline. The SparkToro zero click figure uses Similarweb clickstream data for January to April 2026; the heavy Google use figures use the Datos US desktop panel and exclude mobile. The Ahrefs schema test covered pages already being cited and measured change against matched controls. The Indig figures are observational across 18,012 citations from 1.2 million AI answers. Gartner Market Guide figures are quoted from a vendor hosted reprint because the original is paywalled. The Ahrefs 23 times conversion figure is from a single site. Our own panel method samples ChatGPT and Gemini three times per query; it does not cover Perplexity.

Frequently asked questions

What does an answer engine optimization agency do?

An answer engine optimization agency makes an organization the source AI systems extract, cite, and name when buyers ask questions. Our work covers the content, the structure, the platform, and the operation underneath the page, measured by citation share on a priority query set tested in ChatGPT and Gemini.

Is AEO different from SEO?

Yes. Only 12% of links cited by ChatGPT, Gemini, and Copilot appear in Google's top 10 results (Ahrefs, 2025). SEO earns a ranking on a results page; AEO earns a citation inside an answer. Clean technical SEO remains a prerequisite and is included in our work.

Does schema markup get us cited?

Not on its own. Ahrefs' 2026 controlled test found no measurable lift in AI Mode or ChatGPT citations from adding schema. We apply schema at the template level as hygiene and put the effort into answer placement, heading framing, authorship, entity consistency, and the platform that produces those by default.

What is a Sounding?

A Sounding is our three to four week depth audit. It baselines your citation share on 20 to 40 priority queries, inspects all five Waterline layers, and returns a one page depth chart with an owner, a metric, and a sequenced remediation for each layer.

How do you measure AEO results?

Brand mention rate, citation share, query coverage, and accuracy rate on your priority query set, re tested every 30 days in ChatGPT and Gemini, plus the AI referral channel in GA4 and Search Console impressions and position. A query is won when you are named in three of three samples on both engines.

Sources

All pages opened September 17, 2026.

• SparkToro, In 2026 less than one third of Google searches still send a click, June 9, 2026: sparktoro.com
• SparkToro and Datos, AI tool use and traditional search, August 26, 2025: sparktoro.com
• Ahrefs, AI Overviews reduce clicks, April 17, 2025: ahrefs.com
• Ahrefs, AI search overlap with Google, August 11, 2025: ahrefs.com
• Ahrefs, Schema and AI citations intervention test, May 11, 2026: ahrefs.com
• Ahrefs, Short vs long content in AI Overviews, December 3, 2025: ahrefs.com
• Ahrefs, AI search traffic conversions, June 16, 2025: ahrefs.com
• Search Engine Land, Kevin Indig, ChatGPT citations content study, February 18, 2026: searchengineland.com
• Gartner, Market Guide for Answer Engine Visibility Tools, March 9, 2026, vendor hosted reprint: siteimprove.com
• McKinsey, The new front door to the internet, October 16, 2025: mckinsey.com
• Minuttia, State of AEO Report, June 19, 2026: minuttia.com
• B Lab, Dotfusion Technology, Inc.: bcorporation.net
• Partner standing: Contentful, Storyblok, Agility CMS

This page was conceived and written by Chris Bryce, CEO and Partner at Dotfusion, with our stack of AI research agents doing the heavy lifting on sources and data. If you want to talk about AEO, content operations, or enterprise web strategy, talk to us. We love this stuff.