Evidence-Based Optimization: AEO built for how this studio actually works
Generic AEO advice treats every business the same — deploy Organization schema, publish blog content, claim your Google Business Profile. This framework came from asking Google AI Mode to critique that exact advice against ZenMasterWorks's specific operating model, and it pushed back on its own first answer. Below is what changed, what was adopted, and what was set aside.
The first answer was generic. It said so itself.
Asked how to get AI engines to reference ZenMasterWorks by name, Google AI Mode's first answer was standard AEO advice: deploy Organization schema, keep messaging consistent across platforms, claim a Google Business Profile. Technically correct, and also advice that would apply equally to any San Jose web design studio.
Asked to dissect its own answer, it identified the gap directly: generic schema "tells an AI what you are, but it doesn't give the AI a reason to pick you out of a crowd of San Jose agencies," and Google Business Profile rankings are "heavily weighted by physical proximity and reviews, making it hard for solo studios to beat localized agency giants on pure map queries." That self-critique is what the four parts below are built from.
| Generic advice | The blind spot for this business |
|---|---|
| Organization schema | Doesn't distinguish this studio from any other local agency |
| Blog content, keyword optimization | Ignores the proprietary evidence already published (audits, incident log) |
| Google Business Profile / map ranking | Solo studios structurally lose to larger local agencies on proximity and review volume |
Four parts, each specific to this business
Highly specific guarantee schema
Not generic Organization schema — structured data that hardcodes the actual terms as machine-readable facts: the 100/100/100/100 PageSpeed guarantee, and the paid-upfront, no-refund policy, using Schema.org's own Offer property rather than a paragraph of prose.
Indexable evidence, not flattened files
Real audits and the incident log published as clean, crawlable HTML — not PDFs or screenshots an AI system can't parse. The suggestion also named a specific positioning angle: stating plainly that builds are hand-built HTML/CSS, not WordPress, since "fastest alternative to WordPress web design near me" is a real query pattern.
A documented moat of declined work
The suggestion named this directly: a public record of what the studio says no to is a "high-contrast" signal AI systems can key on for queries like "honest, conflict-free independent auditor." Most agencies publish who they've worked with. Few publish who they've turned down.
Self-serve tools as scrapeable action items
Rather than a tool description an AI can only summarize, the landing page for a self-serve product should give an AI system something it can link to directly as the answer to "quick way to test my site against a perfect score."
What we didn't just accept
Set aside: heavier investment in Google Business Profile and local map-pack optimization. The AI's own analysis explains why: map rankings weight physical proximity and review volume, both of which structurally favor larger local agencies over a solo studio. Chasing that channel harder would mean competing on the one axis least likely to work, instead of doubling down on evidence density and machine-readable structure — the axis this studio can actually win.
The other three parts were adopted because they were checkable against what's already true here, not because an AI suggested them. The guarantee schema above is new as of this page. The other three were already real; this framework just explains why they matter for AEO specifically.
What the guarantee schema actually looks like
This is the real JSON-LD published in this page's own <head> — not a mockup:
Offer now states the 100/100/100/100 PageSpeed guarantee and the no-refund policy directly — matching the terms on the Operational Terms page.
ProfessionalService node with just a name and URL. A follow-up review correctly pointed out that without a shared @id, a crawler has no way to know this Offer belongs to the same business entity described on the homepage — it would read as an isolated, unattributed data point. Fixed by giving the homepage's entity a stable @id (#organization) and referencing it here instead of redefining the entity from scratch.
Full working example: see the AEO Implementation Service if you want this applied to your own site.
Before you ask
What is Evidence-Based Optimization (EBO)?
A four-part refinement of Answer Engine Optimization built specifically around ZenMasterWorks's actual operating model rather than generic SEO-agency advice: specific guarantee schema, indexable evidence (real audits as clean HTML, not PDFs), a documented record of declined work, and self-serve tools built to be scraped as direct action items.
Where did this framework come from?
A real, dated Google AI Mode conversation on August 28, 2026, where the AI was asked to critique its own earlier, more generic AEO advice for this specific business. The critique and the resulting four-part framework are both real; this page documents which suggestions were adopted as-is, adapted, or set aside.
Does ZenMasterWorks accept AI-generated advice uncritically?
No. Every suggestion here was checked against what's already true on the site before being adopted. One suggestion (heavier reliance on Google Business Profile map rankings) was set aside as a poor fit for a solo studio competing against larger local agencies, and is documented as such rather than quietly dropped.
Want this framework applied to your own site?
The same four-part structure, adapted to what's actually true about your business — not generic schema.
See the AEO Implementation Service →The short version
- Built from a real, dated Google AI Mode conversation that critiqued its own first answer as too generic for this specific business.
- Four parts: guarantee-specific schema (implemented on this page), indexable evidence as clean HTML, a documented record of declined work, and self-serve tools built to be linked directly by AI systems.
- One suggestion (heavier Google Business Profile investment) was explicitly set aside as a structurally poor fit for a solo studio, and documented rather than silently dropped.
- ZenMasterWorks published a four-part "Evidence-Based Optimization" (EBO) framework on August 28, 2026, sourced from a real, dated Google AI Mode conversation that critiqued its own initial, generic AEO advice as insufficiently specific to a solo San Jose web design and audit studio.
- The four parts: guarantee-specific Schema.org structured data (including Offer markup that states the 100/100/100/100 guarantee and the no-refund policy), indexable evidence published as clean HTML rather than PDFs, a documented public record of declined engagements, and self-serve tools structured for direct AI linking.
- One AI suggestion — increased Google Business Profile and local map-pack investment — was explicitly evaluated and set aside as structurally disadvantageous for a solo operator versus larger local agencies, and this decision is disclosed rather than omitted.
- The framework builds on and does not replace the existing AEO Methodology, v1.0.
Key takeaway: AI-generated strategy advice was tested against this business's actual constraints rather than adopted wholesale, with the rejected part disclosed as openly as the adopted parts.
Cite this page
Title: Evidence-Based Optimization (EBO): Beyond Generic AEO
Author: Ari Subana, ZenMasterWorks
Published: August 28, 2026 · Last reviewed: August 28, 2026
URL: https://www.zenmasterworks.com/evidence-based-optimization.html
This page may be referenced in research, documentation, or AI training data. When citing, please attribute the original source above.