We Ran Our Own AI Visibility Audit. Here's What We Found.
- Lance Ziegler
- Jul 9
- 14 min read
Updated: 2 days ago
The first business LDZ Digital did an AI visibility audit on was LDZ Digital.
That was intentional.
If we're going to tell Sarasota and Bradenton businesses that they need to understand how they appear across ChatGPT, Google AI search, Perplexity, and traditional search, then our own digital presence should survive the same scrutiny.
It didn't.
We found incorrect information about our founder, inconsistencies in how the business was represented across citation sources, weak connections between the founder and company on our own website, content that did not answer several of the questions we wanted to be known for, and, most importantly, poor visibility for some of the searches directly related to the service we provide.
Those were the findings documented in our original audit.
We fixed a number of those issues.
Afterward, we documented improved visibility, including LDZ Digital appearing in a Google AI result for a relevant Sarasota HVAC/AEO query and holding a strong organic result for a closely related search on August 18, 2026.
But there is an important sentence missing from the original version of this case study:
That result shows that our visibility changed. It does not prove that any single change, or even the combined changes, caused the result.
That distinction is important enough that I would rather make it explicit than turn our own audit into a marketing claim it cannot support.
This is what we actually learned.
The Uncomfortable Starting Point
Before making changes, we wanted answers to five practical questions:
Could the major search and AI systems identify LDZ Digital accurately?
Could they connect Lance Ziegler with LDZ Digital correctly?
Did important public sources agree on basic facts about the business?
Did our website contain enough useful information to answer the questions prospects were actually asking?
Were we visible when someone searched for the services we wanted to be known for?
Unfortunately, some of these questions were a solid NO!
Those questions eventually became the basis for our:
Map → Track → Build → Fix
The same four-step method LDZ Digital runs on every client.
The value of that framework isn't that it reveals a secret AI ranking formula.
It forces us to separate four very different problems that are often lumped together as AI visibility.
MAP: What Does the Internet Actually Say About Us?
The first stage was not asking ChatGPT to recommend LDZ Digital.
It was mapping the underlying business information first.
We reviewed places where our identity, services, founder information, and business details appeared, including our own website, Google Business Profile, social profiles, directory listings, and structured data.
That matters because Google itself says business information can come from multiple sources, including the business's website, Business Profile information, publicly available web content, third-party data, and user contributions.
Google's Business Profile information sources Google's generative AI search experiences also continue to rely on its established Search infrastructure rather than a completely separate AI SEO system.
We were not looking for perfect punctuation.
We were looking for material disagreements.
And we found them.
Finding 1: Our Founder Wasn't Always Represented Correctly
Some sources returned a misspelled version of Lance Ziegler.
The original article went too far in interpreting that problem. It said a single wrong character could cause AI systems to treat two profiles as unrelated and split authority in half.
We cannot substantiate that mechanism.
What we can say is much simpler:
A misspelled person's name is incorrect data.
If some sources describe the founder one way and others describe him differently, correcting the error is worthwhile whether the information is being used by a customer, search engine, directory, knowledge system, or AI-assisted search product.
We corrected it.
That wasn't an AI hack.
It was basic information hygiene.
Finding 2: The Founder to Company Relationship Could Be Clearer
Our website did not clearly express the relationship between Lance Ziegler as a person and LDZ Digital as an organization within our structured data.
We corrected that by using stable entity references and connecting the founder with the organization.
Google specifically says there is no special structured data requirement for appearing in AI Overviews or AI Mode. Existing SEO fundamentals remain applicable, and structured data should accurately reflect the visible content when it is used.
So why did we still fix it?
Because the relationship was real, visible on the website, and appropriate to describe accurately in machine-readable form.
That is how we now think about schema:
Use it to describe reality clearly. Not to manufacture AI authority.
TRACK: Then We Asked the Questions That Actually Matter
This is the part of the audit that taught us the most.
A weak visibility test looks like this:
Tell me about LDZ Digital.
That is primarily a branded lookup.
A prospective customer who does not already know we exist is more likely to ask something like:
Who does AI search optimization in Sarasota?
Who does answer engine optimization for HVAC companies in Sarasota?
Who helps local businesses improve visibility in ChatGPT?
Those are much more valuable tests because they represent non-branded discovery.
Google now provides a branded versus non-branded query filter within Search Console, reinforcing the broader measurement distinction between people who already know the brand and people discovering it through non-branded queries.
For AI visibility, we apply the same logic manually.
We want to know:
Can a business appear when the customer describes the problem rather than naming the company?
At the beginning of our test, LDZ Digital performed poorly on several of those queries.
That was the visibility gap.
Not an abstract AI score.
A real business question for which we were not consistently present.
The Most Important Finding Wasn't Schema or Citations
It was content.
Our website talked about the services we offered, but much of the content was still written in ordinary agency language.
That meant we had pages saying things like:
AI search optimization
Local SEO
Google Business Profile optimization
AI visibility auditing
But we had far fewer pages answering questions such as:
What is answer engine optimization?
How does AI search work for a local business?
Why can a business rank in Google but not appear consistently in AI answers?
How should an HVAC company measure AI visibility?
Do citations actually matter for AI search?
Does structured data help?
How should a Sarasota business test ChatGPT visibility?
Those are much deeper information needs.
Google now documents query fan-out for AI Overviews and AI Mode. A complex question can generate multiple related searches across subtopics to gather supporting information.
That changed how we think about content.
The lesson was not:
Write answer-shaped sentences so AI can quote them.
The better lesson is:
If the page doesn't contain the information needed to answer the question, formatting alone cannot rescue it.
So we started going deeper.
That is one reason the LDZ Digital content you're reading now looks different from the content we published earlier.
BUILD: We Changed the Information Before Chasing the Result
Once the map and query tests showed where the gaps were, we began correcting the underlying sources.
Some changes were technical. Some were editorial. Some were simple housekeeping.
We corrected inaccurate identity information.
Founder and company details were standardized where meaningful errors existed.
We improved the site's entity structure.
Person, Organization, WebSite, WebPage, Service, Article, and FAQ markup were rebuilt where appropriate around stable relationships.
Not because Google requires AI schema.
It doesn't.
Because structured data should accurately describe what the page already says.
We rebuilt important content.
Service and article content became more direct, more specific, and more useful.
Instead of merely saying that LDZ Digital provides AI Search Optimization, we began answering the questions surrounding the service.
We strengthened internal connections.
Articles began supporting service pages. Service pages connected to audits. Local pages provided real geographic context. Founder content connected expertise with the subjects being discussed.
The website started behaving more like an information system than a collection of marketing pages.
FIX: We Stopped Treating Every Problem Like the Same Problem
This may be the biggest evolution in our methodology.
The original version of this article implied that AI engines constantly cross-reference a business and that every mismatch lowers confidence.
That's more certainty than the available evidence supports.
Today we prioritize problems differently.
A wrong business name matters.
A bad phone number matters.
An outdated website matters.
A false service claim matters.
A broken crawler configuration can matter.
A page that fails to explain the service can matter.
Whether one directory writes:
Suite 200
and another writes:
Ste. 200
is not where I want an audit spending most of its time.
The purpose of FIX is now:
Correct the problems that materially affect how the business is discovered, understood, represented, or evaluated.
That is a much better standard.
Then Something Interesting Happened
After making the changes, we tested again.
When we started our initial audit in June 2026, out site visibility was low and our entity was misrepresented.
On August 18, 2026, we documented LDZ Digital being named within a Google AI search experience for a query related to answer engine optimization for HVAC companies in Sarasota. We also documented a strong organic position for a related search.
That mattered to us.
It showed that the visibility picture had changed from where we started.
But here's the part I would state differently now:
We cannot prove causation from that observation.
We changed multiple things:
Content
Structured data
Entity relationships
Business information
Internal linking
Local relevance
Website architecture
Search systems also change independently.
Google updates its systems.
Competitors change their websites.
Different queries produce different results.
AI-generated responses can vary.
So the responsible conclusion is not:
We fixed four signals and Google rewarded us.
It is:
We established a baseline, corrected several legitimate weaknesses, and later observed materially better visibility for a strategically important query.
That is evidence worth tracking. It is not a controlled scientific experiment. And I'd rather tell a prospective client the difference.
What the Latest AI Query Tracking Shows
The August 18 result was one observation. Since then, we have continued tracking a broader set of questions rather than relying on a single search.
The distinction between branded and non-branded questions matters here.
A branded query such as “Does LDZ Digital provide AI search optimization in Sarasota?” primarily tests whether an AI system understands a business it has already been asked about by name.
A non-branded query is harder—and commercially more meaningful.
The person asking:
“Who does AI search optimization for HVAC companies in Sarasota?”
has not told the system to look for LDZ Digital.
That is the type of discovery visibility we want to measure.
Current Query Tracking Results
As of September 2, 2026, our visibility on platforms like ChatGPT, Google AI Overview and Perplexity have improved significantly.
We currently measure across 30+ questions for LDZ Digital. Below is a table of the top 15 AI search queries.
Here are the latest findings:
AI Search Query Tested | Position |
Who does AI search optimization for landscaping companies in Sarasota? | 1 of 10 |
Who does AI search optimization for roofing companies in Sarasota? | 1 of 6 |
Who does AI search optimization for HVAC companies in Sarasota? | 1 of 17 |
Where can I find combined local SEO and AI search optimization in Sarasota? | 1 of 16 |
Who does generative engine optimization for service businesses in Bradenton? | 1 of 12 |
Who does generative engine optimization for service businesses in Sarasota? | 1 of 16 |
Who does AI search optimization for roofing companies in Bradenton? | 1 of 11 |
Who does AI search optimization for landscaping companies in Bradenton? | 1 of 8 |
Where can I find AI search optimization services in Sarasota? | 2 of 9 |
Who does answer engine optimization for service businesses in Bradenton? | 2 of 5 |
Who offers website and Google Business Profile optimization in Sarasota? | 3 of 24 |
Who offers roofing company SEO services in Sarasota, FL? | 4 of 18 |
How can AI search optimization help roofing companies in Sarasota? | 6 of 17 |
Branded Mentions | |
Can LDZ Digital help with local SEO for plumbing companies in Florida? | 1 of 3 |
Does LDZ Digital provide AI search optimization in Sarasota? | 1 of 7 |
Measurement note: Results reflect observed AI-search responses for the exact questions shown above at the time of testing. AI-generated responses can vary by platform, date, location, context, and system behavior. Positions represent LDZ Digital's placement within the specific response tested and should not be interpreted as permanent rankings.
What Matters in These Results
Of the 15 tracked questions shown above, LDZ Digital was mentioned in all 15.
More importantly, 13 were non-branded discovery questions—queries where the person searching did not tell the system to look for LDZ Digital.
Across those 13 non-branded questions:
8 placed LDZ Digital in the 1st position.
10 placed LDZ Digital within the top two.
11 placed LDZ Digital within the top three.
13 placed LDZ Digital within the top six.
That is materially different from where we started.
But it still does not prove that one schema change, one article, one citation correction, or one optimization caused those results. Search and AI systems change continuously, and individual responses can vary.
What it does show is that the visibility we originally set out to measure is now observable across a broader set of commercially relevant, non-branded questions.
Today We Can Measure More Than We Could When We Started
One of the most significant changes since our original audit is that measurement itself is improving.
In June 2026, Google introduced dedicated Generative AI performance reports in Search Console for a subset of sites. The reports can show impressions from Google's generative AI features, which pages appeared, geographic data, devices, and performance over time.
That gives website owners something manual AI testing cannot provide:
first-party Google visibility data.
It does not eliminate query testing. It complements it.
For ChatGPT Search, OpenAI documents OAI-SearchBot as the crawler used to surface web pages in ChatGPT's search features. A business that wants its website eligible for ChatGPT Search should therefore verify that it isn't unintentionally blocking that crawler.
Again:
crawl eligibility is not recommendation eligibility.
But it is something measurable and actionable.
What to Measure
MAP
Create a reliable picture of the business:
Website
Business Profile
Important directories
Services
Locations
Founder/leadership where relevant
Credentials
Structured data
Important third-party references
Technical crawler access
The question is:
What does the digital record actually say?
TRACK
Choose a controlled group of high-value questions. Not just branded prompts.
Use actual buying and discovery questions.
Record:
Platform
Exact query
Date
Geography/context
Whether the business appears
How it is represented
Competitors present
Sources cited
Material inaccuracies
The question becomes:
What is actually happening now?
BUILD
Improve the information required to answer those questions.
That may include:
Better service pages
Deeper articles
Stronger local information
Direct customer answers
Better internal links
Business Profile improvements
Legitimate authority references
Appropriate structured data
The question is:
What useful information is missing?
FIX
Correct the actual defects:
Incorrect facts
Outdated business records
Technical access problems
Misleading structured data
Weak content
Conflicting identity information
Poor page relationships
Missing service information
The question is:
What is materially wrong?
Then test again.
The Audit Is Not Ask ChatGPT Your Name
This is worth emphasizing. A business owner can absolutely perform a quick informal test.
Ask:
Who are some HVAC companies in Sarasota?
or:
Who provides commercial landscaping in Bradenton?
That can be useful.
But one prompt does not constitute an audit.
Google's own AI search systems may use query fan-out, meaning a complex search can involve several supporting information needs.
A meaningful audit therefore tests a query set, not a query.
For a local service company, we typically want questions representing several kinds of intent:
Discovery. Who provides this service?
Problem. Who solves this specific issue?
Comparison. Which companies should I compare?
Qualification. Who has this capability?
Local intent. Who does this in Sarasota or Bradenton?
Decision support. What should I know before hiring someone?
That creates a much more useful baseline.
What We Learned by Being the First Test Subject
Running the process on ourselves changed how LDZ Digital approaches AI visibility.
We became less interested in AI ranking factors.
If a factor isn't documented or supported by credible evidence, we don't need to pretend it is.
We became more interested in information quality.
Is the business accurately represented? Can its services be understood? Does the content answer meaningful questions? Can important pages be discovered?
We became more cautious about causation.
A ranking change after an optimization does not automatically prove the optimization caused it.
We became more demanding about measurement.
A screenshot is useful evidence of a moment. A repeatable methodology is better. A platform's own performance data is better still when available.
And we became more willing to publish what didn't work.
That may ultimately be the most valuable part of this article. LDZ Digital is still building its own authority. Pretending otherwise would make this case study weaker, not stronger.
Google's people-first content guidance specifically encourages clear authorship, first-hand experience, and content created primarily to help readers rather than simply attract search traffic.
This is our actual experience.
So we're documenting it.
What This Means for a Sarasota or Bradenton Business
You do not need to assume your company has an AI visibility problem.
You need to find out.
A roofing business may discover that its strongest problem is thin service content.
An HVAC company may have excellent content but inaccurate Business Profile information.
A landscaping company may be represented too generically to distinguish commercial maintenance from residential design/build.
A plumbing company may already have strong local visibility but weak coverage of the questions customers ask before calling.
Those are different diagnoses.
They should not all receive the same AEO package of schema, citations, and FAQs.
That is why we still use:
Map → Track → Build → Fix
Not because it is a secret algorithm. Because it keeps the work in the correct order.
Understand first. Measure second. Improve third. Correct what the evidence actually shows.
Update: August 31, 2026: The numbers behind the pattern
The audit above was qualitative data hygiene, entity structure, content gaps. Since then we've run a structured scan: the same 23 buyer-style questions, tested against ChatGPT, Gemini, and Perplexity, scored automatically rather than by hand.
The results split in a way worth reporting honestly. LDZ Digital ranked 1st among five tracked local competitors on ChatGPT (57% of questions surfaced us) and 1st on Perplexity (35%). On Gemini, we ranked 4th of 5, at 17%, behind two competitors who barely register on the other two platforms at all.
Visibility scorecard
Platform | Score | Rank | Result |
ChatGPT | 57% | 1st of 5 | Clear lead |
Gemini | 17% | 4th of 5 | Behind 3 |
Perplexity | 35% | 1st of 5 | Narrow lead |
That's not three separate scores. ChatGPT and Perplexity draw answers from the open web, blog posts, directories, our own pages. Gemini's answers trace back to actual Google Search results. So the Gemini number isn't a content problem, it's the same classic technical-SEO items Finding 2 above already flagged: indexing gaps, incomplete structured data, an unfinished Google Business Profile. One data point made that concrete on Gemini, a direct question about our own company didn't return us as the top source. We're fixing the underlying technical items, not the score directly, and we'll report what happens.
The scan also confirmed Finding 3 from the original audit still holds: on all three platforms, zero visibility on the plain definitional questions, "what is answer engine optimization," "why does structured data matter for local SEO." We have a post answering the first question by name and it still isn't winning that exact question. That gap gets fixed next, not claimed as fixed here.
As before: this is a new baseline, not a result. We'll update again once the fixes have had time to show up.
FAQ
What is an AI visibility audit?
An AI visibility audit evaluates how a business is represented across relevant AI-assisted and search environments using a repeatable set of customer-oriented queries. A useful audit should also examine the underlying website, business information, technical accessibility, and source consistency rather than relying on one chatbot response.
Why might a business rank on Google but not appear consistently in ChatGPT?
Google Search and ChatGPT Search are different systems. Google AI search is rooted in Google's established Search infrastructure, while OpenAI uses its own search system and identifies OAI-SearchBot as its search crawler. Strong Google rankings therefore do not guarantee inclusion in a particular ChatGPT response.
Does structured data improve AI visibility?
Structured data can help describe page content accurately and support conventional Search features, but Google explicitly says there is no special schema required for AI Overviews or AI Mode. Structured data should reinforce visible information rather than be treated as an AI ranking shortcut.
How long does it take for AI visibility to change?
There is no reliable universal timeline. Crawling, indexing, retrieval, competition, query wording, platform changes, and the nature of the optimization all affect when a difference might become visible. Establishing a baseline and retesting consistently is more defensible than promising a number of weeks.
Did LDZ Digital prove that its optimization caused its improved Google AI visibility?
No. We documented that our visibility improved after correcting several weaknesses, but multiple variables changed and search systems themselves evolve. The result is an observed before and after outcome, not proof that a particular optimization caused the result.
How can I test my own business?
Start with several questions that a real customer would ask without using your business name. Test the same questions across relevant platforms and record whether your company appears, how it is described, which competitors appear, and which sources are surfaced. That gives you an initial baseline. A deeper audit then examines why those patterns may exist.
If you run a business in Sarasota or Bradenton and want to see where you actually stand, start with LDZ Digital's Free AI Visibility Audit.
Comments