AEO/GEO Experimentation Pilot
Find and fix outdated, risky, and misleading content before AI does
AI search tools are already answering questions about your organization. Your content may or may not be visible or cited correctly if it is.
We have spent the last few months researching and developing workflows for AI Engine Optimization (AEO) or Generative Engine Optimization (GEO) for large Canadian Public Sector organizations.
First, we developed the AEO/GEO DIY playbook, and then we continued our research by experimenting with workflows and a technical toolset to provide a repeatable and scalable process.
Our AEO/GEO pilot is a formal experiment that we will be conducting in partnership with up to five organizations. We plan for the pilot to run between January and March 2026. We aim to provide evidence-based recommendations to the pilot participants by the end of the pilot, outlining which practical AEO/GEO tactics are most effective for their specific context, enabling content to be made visible and correctly cited in AI overviews and summaries.
What “ROT” content reduction means in the context of AI search
ROT stands for Redundant, Outdated, and Trivial content.
In a traditional website audit, ROT content mostly hurts usability and SEO.
In AI-driven search, ROT content creates bigger risks:
AI tools may surface outdated instructions as if they are current
Conflicting pages may be merged into a single, incorrect answer
Draft or edge-case content may be presented without context
Users may never click through to see dates, disclaimers, or updates
The challenge
AI-powered search is now the first place many people turn when they need answers.
People ask full questions, and they often get an answer directly from the AI tool without visiting your site at all.
Many of these interactions are now zero-click.
That creates a real problem:
AI tools can surface outdated or conflicting content
Older pages can override newer guidance
Edge cases or draft content can appear as official advice
Users may never see dates, disclaimers, or updates
The impact
When outdated or confusing content is reused by AI:
Incorrect information spreads faster than teams can correct it
Citizens become confused when answers do not match official processes
Call centres and front-line staff absorb the fallout
Manual triage and rewrites become hard to scale
Trust erodes, even when the source content was once correct
This is something that is already happening.
The AEO/GEO pilot approach
This 3-month pilot aims to:
Understand how and when Canadians are using AI search tools to look for public sector content.
Conduct an experiment at scale to test which AEO/GEO tactics are most effective to make content more visible and more accurately cited in AI overviews and summaries while maintaining (or improving) readability and understandability for human users.
Prove out workflows and technical tools that will provide reliable and ongoing monitoring and measurement of citation visibility and accuracy for selected topics and query clusters related to public sector content.
At the end of the pilot, participants will have evidence-based insight into which AEO/GEO strategies will work to improve visibility and accuracy of their most important content.
-
We start by identifying content that is most likely to be reused by AI systems and most likely to cause harm if it is wrong.
This includes:Scan and inventory
Detect redundant, outdated, or trivial content, broken structure, schema gaps, and indexability issues.
Real-question testing
Test content against long-tail, natural-language questions that people ask.
Visibility and accuracy scorecards
Assess how your pages appear in AI Overviews and other AI search tools, including whether they are cited and whether the information is correct.
The result is a clearer picture of how citizens are interacting with your content through AI tools and also how accurately your content is being cited.
-
Once priorities are clear, we help teams correct issues without breaking what already works.
This includes:
Rewrite drafts
Human-reviewable drafts based on plain language, usability, GC style guidance, and AEO/GEO best practices
Schema output
JSON-LD schema draft recommendations that improve machine readability and reduce misinterpretation
Teams stay in control. Nothing is published without a review.
-
AI systems change constantly. Content drifts. Fixes need validation.
The pilot includes:
Ongoing monitoring for the duration of the pilot
Track visibility and accuracy over time to see what improves citation and correctness
Feedback loop
Confirm whether updates increase inclusion, clarity, and correct reuse in AI answers
This evidence feeds into repeatable, department-specific standard operating processes.
What you get at the end of the pilot
A prioritized list of AI-exposed ROT content (candidates for optimization)
Clear explanations of why each issue matters
Practical recommendations to fix, merge, archive, or leave content alone
Human-reviewable drafts and technical guidance
A baseline you can reuse for future updates
Who this is for
Our pilot is ideal for teams responsible for:
Citizen-facing content
Program and policy guidance
Digital service delivery
Web communications
UX, research, and testing
Content governance and oversight
It is especially useful for organizations with large content inventories, evolving programs, or high public visibility.
Why this matters now
AI is already where many people look for answers, and all indications are that this is a growing trend.
If your wording is not visible or current, other sources will fill the gap, and those answers may not reflect official guidance.
Using tested and proven AEO/GEO tactics along with a consistent “visibility” dashboard will help you regain control, based on evidence rather than guesswork.
Want to talk it through?
We have some confirmed participants in the pilot but still have room for a few more.
If you are interested in learning more or exploring whether this pilot is a fit for your team, we would be glad to discuss timing, scope, and pricing.
Please get in touch by using our contact form or by booking a 30-minute discovery call.