AI visibility testing
Measure whether AI systems identify, describe, cite, recommend, or ignore a brand.
Testing beyond best practicesEdward Sturm Podcast Appearance
The SEO Lessons From Working With 5,000 Publishers
Practical systems from my conversation with Edward Sturm for fixing AI identity confusion, measuring visibility, prioritizing SEO work, and protecting expertise that creates lasting value.
What changes after 20 years in SEO?
I entered search when the work centered on finding demand, creating the right page, and earning the click.
At Refinery29, TMZ, and other major media companies, scaling that work required large editorial, engineering, analytics, and execution teams.
At Raptive, the challenge changed. I helped turn high-touch expertise into repeatable systems serving more than 5,000 independent publishers. That required automation, network-level data, product development, engineering, and support.
Today, many of those systems can be researched, prototyped, tested, and operated securely in the browser by one experienced operator.
The opportunity is to build better growth systems that clarify how machines understand a business, measure what is happening, prioritize the right action, and protect the expertise that creates lasting value.
Watch the full episode
Edward and I discussed entity confusion, the changing economics of SEO, lessons from serving thousands of publishers, and what one experienced operator can now build with AI.
Four ideas from the episode
Entity disambiguation
Help machines distinguish one person, company, or product from another.
Why entity disambiguation mattersConstraint-based SEO
Pick the highest-value action that fits the time, authority, and data available.
How to spend 10 SEO hoursDiscovery vs. discretion
Decide what stays public for trust, and what stays protected for value.
What makes publishers growStarter workflows
Start with the workflow you need
These are working versions of the systems discussed in the episode. Each is useful now. The strongest ones will become deeper guides, worksheets, or tools.
AI Visibility Testing Loop
Starter workflow available
Problem
Randomly asking ChatGPT about your brand is not measurement.
Six steps
Start with real questions your users ask.
Group prompts by persona and intent.
Run the same prompts across multiple models.
Score mentions, accuracy, citations, recommendations, and failures.
Improve the source inputs behind weak results.
Repeat the same test over time.
Starter output
A baseline scorecard showing how each model handles each prompt.
Suggested fields
- Prompt
- Persona
- Intent
- Model
- Mentioned
- Correctly described
- + additional scoring fields
Define the behavior, establish the baseline, change the inputs, and test again.
Entity Disambiguation
Starter workflow available
Problem
Search and AI systems may confuse a person or brand with a similar, misspelled, or more prominent entity.
Five stages
Claim a clear public identity.
Map where the confusion appears.
Publish a canonical distinction page.
Add semantic structure and third-party validation.
Retest and evaluate readiness for public knowledge systems.
Starter output
An entity-confusion worksheet showing which name and context variations succeed or fail across platforms.
Reduce the amount of guessing the machine has to do.
The 10-Hour SEO Action Plan
Starter workflow available
Problem
Most companies do not need another full audit. They need one valuable action they can finish.
New or low-authority site
Choose one audience.
Choose one painful problem.
Validate the demand.
Build a one- or two-page solution path.
Connect it to one measurable business goal.
Starter output
Either one live solution path, or one controlled recovery experiment.
Mature site
Identify the fourth- or fifth-largest year-over-year revenue loss.
Compare the declining page with current winners.
Identify one meaningful difference.
Form one recovery hypothesis.
Apply and measure one bounded intervention.
New sites need focus. Mature sites need recovery discipline.
Discovery vs. Discretion
Starter workflow available
Problem
Publishing everything makes expertise easier to extract. Hiding everything makes the business harder to understand and trust.
Discovery content — make public what helps customers and machines
Identify you
Understand the offer
Trust the expertise
Compare alternatives
Place you in the consideration set
Starter output
A matrix that classifies each asset as Public, Gated, Personalized, or Client-only.
Discovery examples
- Service pages
- Public definitions
- Author biographies
- Selected case studies
- Comparison content
- Research summaries
- Citations and evidence
Discretion — protect what contains
- Proprietary data
- Custom decision logic
- Detailed implementation
- Client-specific diagnosis
- Internal automation
- Scoring formulas
- High-value operating procedures
Publish enough to be understood. Protect the layer where the real value is created.
Release sequence
Building the workflows in public
The podcast introduced the ideas. This page will track which ones become deeper guides, worksheets, experiments, and tools.
Podcast companion
Episode, concepts, four starter workflows, and CTA.
AI Visibility Worksheet
Prompt inventory, scoring rubric, baseline, and retest.
Entity Test Harness
Name variations, source tracking, and validation map.
10-Hour SEO Planner
New-site and recovery paths with measurement.
Discovery vs. Discretion
Classify content as public, gated, or protected.
Current work & positioning
From SEO recommendations to working growth systems
“I’m finally building the growth systems I was writing proposals about 10 years ago.”
At Refinery29 and TMZ, large-scale SEO depended on close coordination across editorial, engineering, product, analytics, executives, and execution teams.
At Raptive, I helped adapt that expertise for more than 5,000 publishers through automation, network-level data, diagnostics, products, and support.
Today, I help founders, publishers, and expert-led businesses build cross-channel growth decision systems across search, social, email, and AI visibility.
Build the system behind your growth
I help businesses turn scattered data, ideas, and recommendations into a reliable system for deciding what matters and putting it into action.
Prefer email? hank@hankazarian.com