Home › Services › Show up in ChatGPT and AI search

I want the business to appear when customers use AI to search

Show Up in ChatGPT and AI Search Audit

An AI search visibility audit records how the business appears in major AI answers and what can be improved.

See whether AI assistants can understand, cite and recommend your business, then learn what can be improved. Each assistant reads its own source of information, so a business that does well on Google can be described wrongly by one assistant and left out of another entirely.

This asks four assistants the same five customer questions, writes down exactly what each one says about you, and dates it. Then it separates the causes: the assistants cannot reach your site, your facts are wrong somewhere, your pages do not answer the question, an outside source is saying something else, or nobody is measuring.

Nobody can sell you a place in an AI answer. No mention, citation, link, ranking or date is promised here, and no provider can promise one. What is real is the measurement, and what it finds.

The audit names what is worth changing and what it would take. The improvements themselves are quoted after, once you have seen the evidence, so you decide what is worth doing rather than buying a package first.

You will see this sold as AEO or GEO. AEO is answer engine optimization, being the source an assistant uses when it answers. GEO is generative engine optimization, the same idea named for systems that write an answer instead of listing links.

How this gets fixed

The service is a dated measurement, not an optimization package

It is a measurement, not an optimization package. What the assistants currently say about you is recorded and dated, alongside what they are actually able to read on your site.

Nobody can promise a mention, and any service that does is guessing. What you get is evidence of where you stand and a list of what is worth changing, with the work quoted separately once you have seen it.

Afterwards

What you will know when this is done

A dated record of what four answer systems actually said, the evidence behind every finding, and a practical order of work.

  • Whether the business is named at all for the questions your customers ask
  • Whether the description of the services, locations and claims is accurate
  • Which sources those systems show as support, and who controls them
  • Whether anything on the website is blocking the crawlers a system publishes
  • What the platforms report about the site themselves, where the accounts exist
  • Whether visits arriving from AI systems can be identified in your own analytics
  • Which competitors and which outside sources appear instead of you
  • Which corrections are urgent, which are growth work, and which belong elsewhere
  • Who needs to do each action, and the likely level of effort

What you will not have is a promise of a mention, a citation, a link, a recommendation, traffic or a date. Google says that meeting every requirement does not mean it will crawl, index or serve a page, and that indexing and serving are not guaranteed. No other system promises more than that either.

Scope

What this service includes

Included in your service
  • One business, one website, one language, one market
  • Five customer questions, agreed with you before testing.Real questions a buyer would type, agreed in advance so nobody goes looking for a flattering answer afterwards
  • Four answer systems: Google AI features, ChatGPT, Microsoft Copilot and Perplexity
  • Twenty dated answer records, with the wording, system, date and conditions used.Five questions across four systems. Dated because an assistant can answer differently tomorrow, so an undated screenshot proves nothing
  • Up to three competitors, compared on mentions and cited sources.A citation is the source an assistant names or links to when it answers
  • A citation and factual accuracy review against the facts you approve.Whether what the assistants say about you is actually true, checked against facts you sign off first
  • Crawler access checks for the search crawlers each platform documents.A crawler is the software that reads your site. Each company runs more than one, for different purposes, and blocking the wrong one can remove you from an assistant entirely
  • The page level directives that control AI answers, checked on every priority page.A directive is a one line instruction in the code of a page. Microsoft says content tagged NOARCHIVE will not be included in Bing Chat answers and will not be linked to in them, and that NOCACHE content shows only as a web address, a title and a snippet. A page can pass every access check above and still be invisible to Copilot because of one line somebody left in it
  • Available first-party AI reporting from Search Console and Bing Webmaster Tools.First-party means the platform's own report about your site, rather than a guess from outside
  • AI referral review where analytics access exists.Whether anybody is actually arriving from an assistant, and landing where
  • Up to five priority pages reviewed for answer readiness.Whether a page answers the question plainly enough to be quoted, rather than burying it
  • A report split into urgent corrections, owned improvements, outside source work and items nobody controls. That last group is the honest one. Some of this is not yours to change and is named rather than sold
  • An owner, a dependency and an estimated effort for every recommendation.The owner is who does it, which is often not me. A dependency is what has to happen first
  • One findings call, up to 45 minutes
  • A written quote for the corrections I can carry out, where you want them carried out.Quoted after you have seen the evidence, not before, so you decide what is worth doing
  • Any factual error in the report, corrected
Related work quoted separately
  • Any guarantee of a mention, a citation, a link, a recommendation or answer persistence.Persistence means the answer staying the same tomorrow, which nobody controls
  • Any guarantee of a ranking, traffic, inquiries or leads
  • Any date by which an answer system will show the business
  • Prompt wording chosen to manufacture a favorable screenshot.Asking a question worded to produce a good answer proves nothing, and it is how most AI visibility reports are made
  • Page rewrites, new page writing or website implementation
  • Listings cleanup, review programs, public relations or link outreach
  • Training an outside model on your business.No vendor can add your business to how a model was built, whatever they say
  • A promise that allowing or blocking a training crawler changes search visibility.A training crawler collects text to build a future model. It is not the same crawler that fetches your page to answer a question today
  • A special schema, an llms.txt file or any other file sold as a shortcut.Schema is the hidden label on a page telling a machine what it is looking at, and llms.txt is a proposed file for AI systems that the major search companies have not committed to reading
  • Fake reviews, purchased mentions or fabricated authority of any kind
  • Removal of third party content, or contacting other websites on your behalf
  • Paid placement, software subscriptions or advertising
  • Ongoing monitoring, a second measurement date, or a full market share study
  • More than one brand, website, language or market
  • Systems, questions, competitors or pages beyond the agreed counts
The process

How it runs, step by step

  • Tell me what the assistants are getting wrong The form below. The brand, the market, the questions and any wrong answer you have already seen.
  • You get a written scope The systems, questions, competitors, pages, access and evidence included, with the price and a delivery window. In writing, before anything starts.
  • You approve and pay Then the audit is scheduled.
  • The baseline is recorded The agreed questions are asked and dated, and available first-party data is exported, before any recommendation is written.
  • The causes are separated Crawler access, technical eligibility, business facts, outside sources and five priority pages.
  • You get the report What is controllable, what belongs to another service, and what nobody controls, in priority order.
  • We review it together One call, up to 45 minutes. You decide whether to request implementation.

More on how working together works.

What moves the price
  • More than five customer questions, or more than four answer systems
  • More than three competitors compared
  • More than five priority pages reviewed
  • More than one brand, website, language or market
  • A second measurement date later, to see whether anything moved
  • No analytics or Search Console access, so the first-party half has to be worked around
When this is your service, and when it is a different one

This is your service when:

  • Assistants leave the business out of answers your customers actually ask
  • An assistant describes the business incorrectly, or names services you stopped offering
  • Old information, an old address or an old brand keeps coming back in answers
  • Competitors are named and you do not know which sources are feeding that
  • An assistant cannot answer a basic question about what you sell or where you work
  • Somebody changed the robots file, the small file that tells crawlers where they may go, or added a security layer, and nobody checked what it blocks
  • A page carries a directive that specifically excludes it from AI answers. Microsoft publishes what NOARCHIVE and NOCACHE do to Copilot answers, and neither one shows up in an ordinary access check
  • You are being sold AI files, AI schema or guaranteed AI placement and want a measured second opinion

When it is a different service

What is happeningWhere it goes
The website does not appear in ordinary Google resultsSEO and local Google Search
An audit already named the repairs and you want them carried outAsk about this service
You do not know which customer questions need which pagesKeyword research and page map
Your name, address, hours or services are wrong across other sitesBusiness listings and directories
The real problem is too few reviews, or what the reviews sayGoogle reviews
You want Bing, Yahoo, DuckDuckGo or IndexNow connectedSEO implementation and website optimization

If several of these overlap, say so. It can be quoted as one ordered project instead of making you buy several disconnected services.

Why each assistant answers from a different place

There is no single place to be found. Each assistant reaches the web through a different door, and four of them are not Google.

AssistantWhere its web answers come from
Google AI Overviews and AI ModeGoogle's own Search index. Google calls this grounding and says the features are rooted in core Search ranking.
Microsoft CopilotBing. Microsoft: it "may fetch information from the Bing search service".
ChatGPTIts own crawler, plus search providers OpenAI does not name. The privacy policy it links under those providers is Microsoft's.
PerplexityIts own index, which it describes as covering "hundreds of billions of webpages".
ClaudeIts own Claude-SearchBot. Anthropic has not published what index sits behind it.
Meta AIIts own crawler, meta-webindexer.

Ranking well on Google does not put you in ChatGPT, Copilot, Perplexity or Claude. Each one has to be able to reach and read the site on its own terms.

The three kinds of AI crawler, and why blocking the wrong one removes you from ChatGPT

There are three kinds of AI crawler, not two, and the difference decides whether a business appears in AI answers at all.

The citation crawlers. Blocking these removes you from the answers

The training crawlers. Blocking these changes nothing about being cited

The user-triggered fetchers, which a person sets off by asking

  • ChatGPT-User, Claude-User, Perplexity-User, meta-externalfetcher.
  • Three of the four vendors state robots rules may not apply to these, because somebody asked for the page.

The common mistake is a robots file that blocks the citation crawlers and leaves the training crawlers running, which is the opposite of what was intended. It takes minutes to check and it is the first thing looked at.

Whether the assistants can reach your site at all

A robots file can say yes while the hosting, the CDN or the firewall says no. OpenAI names this directly, telling publishers to confirm the host or content delivery network allows its published searchbot IP addresses.

  • OpenAI publishes its address ranges per crawler, so server and firewall logs can be checked against them
  • Anthropic and Perplexity publish theirs the same way
  • Blocked at the edge looks identical to never crawled, and neither one produces an error anybody sees
  • Robots changes take about a day to take effect, which the vendors state

This is the check that turns up the most, because nobody set out to block an assistant. A security rule did it, and the site owner was never told.

The six things recorded separately, and why one number would lie

There is no single AI visibility score here. Six different things are recorded separately, because they are not interchangeable.

  • Mention. Whether the business is named in the answer.
  • Description. Whether the business, services, locations and claims are described accurately.
  • Citation or link. Whether the website or another source is shown as support.
  • Platform reporting. What a platform's own webmaster tools report about the site, where such a report exists.
  • Referral activity. Visits arriving from AI systems, where your website measurement can identify them.
  • Business action. Inquiries, calls, bookings or purchases connected to those visits, where conversion tracking exists.

A business can be named without receiving a link. A page can be cited without producing an inquiry. One favorable answer is not a market position.

Answers are recorded with the wording, the system, the date and the conditions used, because a run on another day can return something different. Nothing here treats a single answer as permanent.

What Google, Microsoft and Perplexity ask for, in their own words

Only three of these publish guidance for site owners at all, and they do not agree on what matters.

Google, and it is blunter than the industry is

Microsoft, which asks for structure

  • Topical depth rather than breadth
  • Headings, tables and question and answer sections, so a passage can be lifted cleanly
  • Claims supported by evidence
  • Content kept current, and IndexNow used so changes are noticed

Perplexity, which is the only one asking editorial questions

Anthropic, Meta, xAI and Apple

  • Nothing published for site owners about being cited. Only crawler controls.
  • Where a vendor has published nothing, nothing is claimed on their behalf.
What each system publishes about how it works

Four systems are tested. Everything stated below about how each one works is the platform's own published documentation, and every link goes to that documentation. Where a platform publishes nothing on a point, nothing is claimed on it.

Google AI Overviews and AI Mode

Google says there are no additional requirements to appear in AI Overviews or AI Mode, that existing SEO fundamentals continue to be worthwhile, and that you do not need to create machine readable files, AI text files, markup or special schema.org structured data for these features. Google also says a page needs to be indexed and eligible to be shown in Google Search with a snippet to be eligible as a supporting link, and that indexing and serving are not guaranteed even when everything is correct.

Search Console has a generative AI performance report. Google says it includes impressions for AI Overviews and AI Mode, and the data can be grouped by page, country, device and date. Where the property has data, that baseline is recorded and dated. Impressions are not converted into a promised position. Google's documentation for that report.

Microsoft Copilot and Bing

Bing Webmaster Tools has an AI Performance report in public preview. Microsoft says it shows total citations, average cited pages, grounding queries and page level citation activity across Microsoft Copilot, AI generated summaries in Bing and select partner integrations. Microsoft also says those counts do not indicate ranking, authority or the role of any page within an individual answer, and that page level citation activity reflects how often pages are cited rather than page importance, ranking or placement. Microsoft's announcement of the report.

ChatGPT search

OpenAI documents separate crawlers and says what each one is for. OAI-SearchBot is used to surface websites in search results in ChatGPT's search features. GPTBot is used to crawl content that may be used in training generative AI foundation models. ChatGPT-User covers user initiated actions and OAI-AdsBot covers ad validation. OpenAI says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links. OpenAI's crawler documentation.

Those are separate decisions, and the audit reports them separately. The training decision is yours, and no part of this recommends permitting training in order to gain visibility. Nothing in that documentation offers a business a mention or a citation.

A limited prompt test is a repeatable sample, not a complete report of everything every user sees.

Perplexity

Perplexity documents PerplexityBot as designed to surface and link websites in search results on Perplexity, and says it is not used to crawl content for AI foundation models. It documents Perplexity-User as supporting user actions, and says that fetcher generally ignores robots rules because a person asked the question. Perplexity's crawler documentation. The audit records whether each is permitted. Nothing in that documentation promises inclusion or a citation.

llms.txt: what it is, who reads it, and what Google says about it

An llms.txt file is a plain list of your important pages, written for machines, placed at the root of the site. It was proposed in September 2024 so that an assistant reading a site gets a clean summary instead of pages full of navigation.

What is actually known about it

  • No AI company has said it reads yours. Not OpenAI, not Anthropic, not Perplexity, not Microsoft. Silence, rather than refusal.
  • Google has addressed it, and says no. Google Search "doesn't use them", and having one will neither harm nor help.
  • The reason people believe the AI companies support it is that several of them publish one for their own developer documentation. Publishing a file is not reading a file.

So the position here

  • It is not sold as visibility, because no vendor has said it produces any
  • It is built if you want one, as a small add-on, and you are told exactly this before you decide
  • It costs nothing to have and, on Google's own statement, does nothing either way
Everything checked, and the limits on it

1. Questions and baseline

  • Five questions agreed with you, spread across branded, local or category, problem, comparison and decision stage
  • The exact wording, the system, the date, and the account and location conditions recorded where those can be controlled or observed
  • Twenty dated answer records covering mentions, descriptions, citations and the competitors named
  • Unsupported answers, unfavorable answers and factually wrong answers separated instead of scored as one problem

Absence from five questions is not proof that a business is invisible everywhere. It is evidence about five recorded questions on a recorded date.

2. First-party measurement

  • The Search Console generative AI report where the property exists and has data
  • The Bing Webmaster Tools AI Performance report where the account exists and has data
  • AI referral visits and conversions in your analytics, where they can be identified
  • Any existing history recorded before recommendations are written
  • Missing measurement named plainly, without treating no data as proof of no visibility

3. Access and technical eligibility

  • Indexing and snippet eligibility for the priority pages, which Google names as the eligibility condition for its AI features
  • OAI-SearchBot, GPTBot and PerplexityBot rules reviewed as separate decisions
  • Bing crawl and index status where account access exists
  • The robots file, which tells a crawler which parts of the site it may look at. Page status codes, which say whether a page loaded or failed. The canonical address, meaning the one version of a page you want counted. The sitemap, the list of pages you hand a machine so it knows what exists. Internal links, and whether important information is there as readable text rather than only inside a picture or built by script after the page loads
  • CDN, firewall or security blocking, checked where the available access can confirm it
  • Structured data checked only where it truthfully describes what a reader can see on the page

Google's structured data guidelines say markup is a true representation of the page and that content readers cannot see is not marked up. No markup is added here to describe something the page does not say.

4. Business identity and factual consistency

  • Business name, category, services, locations and contact details
  • People, qualifications, licenses, prices and policies the business publishes
  • Differences between the website, the Business Profile, maps, listings, reviews and other public sources
  • Old brands, old addresses, retired services and outdated descriptions that can produce a wrong answer
  • A named source of truth for each fact you control

5. Website answer readiness

  • Whether a page exists at all for each important question
  • Whether that page answers the question plainly before it starts selling
  • Original experience, examples, evidence, limits and sourcing
  • Clear headings, useful sections, authorship and update information where relevant
  • Internal links connecting services, proof, policies and related explanations
  • Up to five priority pages reviewed against the five agreed questions

Google says to write non-commodity content that readers find helpful and reliable, and says creating large numbers of pages to manipulate AI responses violates its scaled content abuse spam policy. Google's guidance on that. No page count is produced here to game an answer.

6. Outside sources and authority

  • Which sources the tested systems cite for you, and which they cite for competitors
  • Whether each of those is controlled by you, independent, outdated, inaccurate or realistically obtainable
  • Relevant reviews, professional profiles, directories, maps, associations, publications and original research
  • Gaps that belong to listings, reputation, public relations, partnerships or subject expertise

Fake reviews, purchased mentions, disguised community promotion and fabricated authority are excluded. They are not offered at any price.

The report, and everything in it

Everything below is delivered for one business, one website, one language and one market.

  • The five approved questions and the recorded test conditions
  • Twenty dated answer and citation records
  • A comparison of the competitors and the cited sources that appeared
  • Crawler and technical eligibility findings for the priority pages
  • Available Google, Bing and referral baselines, where the accounts exist
  • Factual conflicts, and a named source of truth for each
  • Five priority page answer readiness reviews
  • A report separating urgent corrections, work you own, outside source opportunities and items nobody controls
  • An owner, a dependency, an evidence standard and an estimated effort for every recommendation
  • One findings call, up to 45 minutes
Repairs, boundaries and what nobody controls

The audit identifies the problem and names the next work. Repairs are not part of it. You get a separate written quote based on the approved priorities, the access that exists and the limits of the website and the outside platforms.

There is no universal AI repair. A blocked crawler, a missing service page, a false directory record and a lack of independent authority are four different jobs at four different prices.

What can be quoted and carried out directly

  • Correcting robots rules that unintentionally block search crawlers
  • Separating ChatGPT search access from your training preference, as OpenAI documents them
  • Correcting page status, canonical, sitemap and internal link problems
  • Making important information available as readable page text
  • Improving page titles, headings, answer structure and source clarity
  • Adding or correcting truthful structured data that matches the visible page
  • Correcting inconsistent business facts on pages and profiles you control
  • Adding service, location, pricing, policy, authorship or evidence information you supply and approve
  • Connecting or configuring Search Console and Bing Webmaster Tools
  • Saving a repeatable baseline so a later run uses the same questions and conditions

Completing those changes makes information clearer and technically available. It does not make an answer engine use it.

What depends on your platform and your access

The builder, the hosting, the security layer and who owns the accounts decide whether the robots file, headers, canonicals, sitemaps, templates, navigation, structured data, redirects, server responses, firewall rules, server logs or measurement can be reached at all. Some builders expose those in settings. Some need a higher plan, direct code access, a hosting administrator or the original developer. A public website can be audited without editing access. The platform and the permissions are confirmed before implementation is priced.

What belongs to another service

Indexing, canonicals, sitemaps, redirects and internal linksWebsite SEO Audit
Deciding which customer question needs which pageKeyword research and page map
Writing or substantially rewriting the pages themselvesCopywriting
Correcting names, services, locations and hours across other sitesBusiness listings and directories
Building a compliant review system and replying to feedbackGoogle reviews
Connecting Bing, Yahoo, DuckDuckGo and IndexNowSEO implementation and website optimization
Analytics, conversion measurement and ongoing dashboardsTracking and reports
Templates, JavaScript, hosting or code the builder does not exposeWeb design

A link to another service explains the boundary. It does not make that work part of this audit. Hacked content, impersonation, defamation, trademark and regulated claim problems go to a security, legal or reputation specialist, not to me.

What nobody controls

  • Whether an AI company crawls, indexes, retrieves or uses a page
  • Whether the business is mentioned, cited, linked to or recommended
  • The order, the wording, the persistence or the accuracy of a generated answer
  • Which competitors and which sources appear instead
  • Personalization, account context, model changes and location differences
  • How quickly a platform picks up a corrected page or a corrected outside source
  • Whether an outside website accepts, changes or removes information
  • Platform eligibility, reporting availability and product changes
  • Traffic, inquiries, sales or return from AI visibility

Those limits are the reason the audit exists in this shape. Establish the evidence, correct what you control, document what sits outside it, and measure later movement against the same agreed baseline.

What I need from you

Split in two on purpose. The first list is all I need to price this and tell you whether you even want it. The second only matters once you have said yes.

Before I can quote you

  • Business name and website
  • The market, the main services and who the customer is
  • Five questions a customer would ask an assistant instead of calling you
  • Which systems concern you most
  • Any answer you have already seen that was wrong, with a screenshot if you kept one
  • Competitors you consider relevant, if you know them
  • How many locations and languages are involved
  • Whether there has been a move, a rebrand, a name change or a website rebuild
  • Which accounts already exist: Search Console, Bing Webmaster Tools, analytics

That is the form below. A few minutes, and it saves a call.

After you approve

  • Read access to Search Console, Bing Webmaster Tools and analytics where they exist
  • Accurate business facts, approved by one named decision maker
  • The profiles and listings that carry your public information
  • Any earlier reports, screenshots or known incorrect answers
  • Confirmation of which five questions and which five pages matter most

The public audit still runs when an account does not exist. First-party platform and referral findings cannot be confirmed without the matching data, and the report says so where that happens. No password is sent by email or text. You stay the owner of every account.

How long it takes, and what delays it
My work

The report is normally delivered within ten business days after payment and after the required information, decisions and access have arrived. Your proposal carries the actual window.

What moves the date

If information or access arrives late, the delivery date moves and the work may return to the next available opening.

That window is a delivery date for the report. It is not a date for visibility in any answer system, and no such date is offered.

Payment

The audit is paid in full before it is scheduled. Once the work has been scheduled and started, the payment is not refunded, and what has been completed is yours either way.

What a second run answers

A one-time audit answers what is happening now and what is most likely worth fixing. A second run answers a narrower question: did the recorded signals change.

  • The same approved questions, asked again on the same five topics, dated
  • Updated Google and Bing first-party data where the accounts exist
  • Identifiable AI referral activity compared against the first baseline
  • Factual and citation changes recorded against the original records

Monitoring measures movement. It does not claim that any particular change caused it. New questions, new systems, content production and implementation stay separate work, quoted separately. Monitoring is not part of the audit price.

What you own afterwards, and when it is finished

The report, the exported records and your recommendations are yours. Third party platforms and licensed data stay subject to their own terms.

How you will know it is finished

  • The agreed questions were tested and the twenty records saved
  • The account checks, page reviews and source findings were recorded
  • The report was delivered and the findings call was completed, or declined

Completion means the evidence was recorded and the report delivered. It does not mean an answer engine changed, and it does not mean repairs were included.

What happens if I get it wrong, you change your mind, or it stalls

If I got a fact wrong

Any factual error in the report is corrected without charge when you report it within 14 days of delivery.

If you change the questions, the systems or the market

New questions, new systems, new test dates, new competitors or a business change after testing has begun are additional scope and may need a new quote.

If the project stalls

If required access, information or approval does not arrive, the project can be closed with the completed work documented. Previously completed work will not be repeated. Reopening or extending the audit may need a new quote.

If you cancel after work has started

Once the audit has been scheduled and started, the payment is not refunded. You get the work completed to that point.

FAQ

Questions about this service

Can you make the changes, or do I get a report and nothing else?

The changes can be quoted after you have seen the findings. The audit deliberately does not include them, so the diagnosis is not written by the person hoping to sell the repair. Some of the work is website changes, some is listings or reviews, and some belongs to another service. You get told which is which, and a separate quote for the parts you want done.

Is this different from SEO?

It overlaps. Crawling, indexing, readable content and credible sources matter to both, and Google says existing SEO fundamentals continue to be worthwhile for its AI features. This adds the parts SEO does not cover: testing answer systems directly, recording citations and factual descriptions, and reading the first-party AI reports where they exist. Most honest AI visibility work starts with good SEO rather than replacing it.

Can you guarantee ChatGPT recommends the business?

OpenAI controls its systems, and nothing in its documentation offers a business a mention or a citation. What can be done is verify the access settings OpenAI publishes, record what appeared for the agreed questions, and improve the sources you control.

Can you guarantee any AI visibility at all?

The audit, the evidence, the report and any separately approved implementation work are what is guaranteed. Whether a platform crawls, indexes, cites, links to, recommends or keeps mentioning the business is outside anyone's control.

Will adding a special AI file or AI schema solve it?

Not for Google. Google says you do not need machine readable files, AI text files, markup or special schema.org structured data to appear in AI Overviews or AI Mode. OpenAI and Perplexity publish crawler controls, and those are access settings. None of it is sold as a shortcut.

Is GPTBot the same as ChatGPT search?

OpenAI documents GPTBot as crawling content that may be used in training foundation models, and OAI-SearchBot as surfacing websites in ChatGPT's search features. They are separate decisions and the audit reports them separately. The training decision stays yours.

Does one screenshot prove visibility?

It proves what appeared under those recorded conditions on that date. A useful baseline needs agreed repeated questions, the first-party reports where they exist, and a later comparison against the same records.

Can an answer be wrong even when the website is right?

Yes. An answer can lean on older pages or on outside sources you do not control. The audit finds the conflicting sources it can and names who owns each one. Removal from a third party site is not promised.

Why five questions and not fifty?

Because every answer is recorded with its wording, system, date and conditions, and that record is the evidence. Five questions across four systems is twenty records that can be repeated later and compared. More questions, more systems or more markets are quoted as additional scope.

What if the accounts do not exist yet?

The public audit still runs. Search Console and Bing Webmaster Tools can be connected as separately quoted work, and until they carry data, the report says which findings could not be confirmed instead of filling the gap with an assumption.

Can somebody else make the repairs?

Yes. Every recommendation names the owner, the dependency and the effort, so your staff or another qualified provider can work from it.

Do you work outside Palm Beach County?

Yes, by video. This audit needs nobody in the room. In person covers Palm Beach County, Broward County and the Treasure Coast.

Tell me the questions customers ask AI instead of calling you.

You get the findings first, then a separate quote for the work you want carried out.

Get a quote

Tell me the result you want and where I can see the business now. I use that information to identify the most useful service or package and decide whether I can quote it in writing. If one missing detail changes the work or price, I will ask for that detail or suggest a short call.

What would help most?

Pick at least one. More than one is fine.

Answer a few more if you want a more specific first reply

What happens next: I review the links and details, then reply within two business days with the best next step, any one question I still need answered, or a written quote when the scope is already clear.

Prefer regular email? [email protected]

Derek Packer

Your inquiry was received.

I will review what you sent and reply within two business days.

You will receive a copy of your inquiry. Look for an email from Derek Packer with the subject Inquiry Received — Here’s Your Copy. If it is not in your inbox, check spam, promotions or other folders. Still missing? Email [email protected].