Part one
Readiness and eligibility
Inspect the hotel website and authoritative sources. Record evidence, the finding, the fix, an owner, and the date. Some checks may require help from someone comfortable with structured data or page source.
WORKGROUPS 5 + 6 OUTPUT
A two-part hotel AI visibility check
First inspect what the hotel publishes and controls. Then inspect what AI systems actually say. Keep the two sets of findings separate.
Rule
Findings and fixes — not a comparative score
A self-assessment result is never converted to a score, never compared with a conformance score, and never compared between properties.
How to use the two parts
Part two
Observed behavior
Run a declared prompt set under consistent conditions. Record each AI engine separately. One favorable screenshot is not a measurement, and no engine results are pooled into one number.
PART ONE: Readiness and eligibility checks
Use the ten checks as separate findings. The first five provide a focused browser-based starting point; the remaining checks cover identity, source authority, consistency, and ownership. Each check now includes a plain-language procedure that can be completed without paid tools.
Before you start
A browser, about ninety minutes, and two simple habits
You need a desktop or laptop browser. You do not need a paid tool, plugin, or account. Capture a screenshot for every check, record the exact URL and date, and keep the finding even when the check passes.
Find on page
Press Ctrl + F on Windows or Cmd + F on Mac to search the text of the page you are viewing.
View page source
Right-click a page and choose View page source in Chrome, Edge, or Firefox, or Show page source in Safari after enabling the Develop menu. The code may look intimidating; you only need Find on page to locate one word at a time.
Plan for limited help on checks 02 and 05. These are the two most likely to require a few minutes from the person who maintains the website. Everything else is primarily reading, recording, and comparing.
Retrieval access
Check: Open the hotel's robots.txt file and check whether approved search and retrieval agents are blocked from public pages that contain material hotel content.
How to do this
- Type your website address followed by /robots.txt in the browser address bar—for example, yourhotel.com/robots.txt—and press Enter.
- If you see a plain-text file, continue with the steps below. If you see a not-found page, nothing is being blocked through a robots.txt rule, so this check passes.
- Read the file in pairs: a User-agent line names a crawler, and the Disallow lines underneath apply to that crawler. Disallow: / blocks that crawler from the whole site; Disallow: /admin/ blocks only that folder.
- Use Find on page to look for Disallow: / with nothing else after the slash. For each match, scroll up to the nearest User-agent line and record the crawler name.
- Repeat the check for any separate address a guest reaches—especially a booking subdomain such as book.yourhotel.com/robots.txt, which may be managed by another supplier.
Record every crawler name you found blocked. Do not judge the crawler in this step, and keep training-only crawlers on a separate line of your notes.
Mark: Pass when the main public content is available; Needs fix when blocking is limited to a relevant public path; Fail when a retrieval agent is blocked from the main content.
Fix: Remove the broad block or narrow it to private, account, cart, search, or administrative paths. Record training-crawler choices separately and do not grade them.
Correct lodging type
Check: Validate the homepage's structured data and confirm that the property is identified with the correctly cased lodging type that reflects its actual operating category.
How to do this
- Open the homepage, right-click an empty area, and choose View page source.
- In the source tab, use Find on page and search for @type.
- Step through the matches and look for a lodging category such as Hotel, Resort, BedAndBreakfast, Hostel, Motel, Campground, Apartment, or VacationRental.
- Check capitalization: Hotel is correct; hotel is not.
- Compare the type with what the property actually is. A small guesthouse marked as Hotel is a mismatch worth recording.
Common outcomes: Organization or LocalBusiness without a more specific lodging type is overly generic. No applicable lodging type in the page source is a failed check.
May need help. If the search returns nothing, or you cannot tell which match describes the property, ask the website maintainer to confirm the structured-data type on the homepage.
Mark: Pass for the correct lodging subtype; Needs fix for an overly generic type; Fail when the property is incorrectly typed or has no applicable lodging type.
Fix: Correct the structured-data type and make sure it agrees with the visible page and the property itself.
Eight essential hotel fact groups
Check: Check identity, location, category, contact, stay times, policies, amenities, and price. The workbook defines what counts as complete.
How to do this
- Work through the eight fact groups in the table below. Mark a group complete only when a guest can reach a real published value on the hotel's website.
- Use Find on page on the relevant page, or search the web for site:yourhotel.com followed by the fact you need—for example, site:yourhotel.com "check-in".
- Do not count placeholders such as Contact us for details, varies by season, or on request. A value that appears only after dates are entered in a booking widget does not count either.
| Fact group | What counts as complete |
|---|---|
| Identity | The official property name used on legal and booking documents. |
| Location | Full street address, postal code, city, and country. |
| Category | The lodging type, stated in words on the page. |
| Contact | A telephone number and email address, both presented as text. |
| Stay times | Check-in and check-out with actual clock times. |
| Policies | Cancellation, pets, smoking, and deposit or card hold. |
| Amenities | Wi-Fi, parking, breakfast, and accessibility, each with its real status. |
| Price | An indicative nightly rate or rate range. |
Mark: Pass when seven or eight groups are complete; Needs fix for three to six; Fail for fewer than three.
Fix: Work from missing high-value facts first. Publish clear, specific answers as readable website text. The professional audit separately checks structured and machine-readable data.
16 priority traveler questions in visible text
Check: Confirm that the hotel's own domain answers all 16 practical guest questions listed below as readable page text—not only in images, PDFs, widgets, or third-party listings.
How to do this
- Take the sixteen questions listed directly below in this check and work through them one at a time on the hotel's own website.
- For each answer, apply the select-and-copy test: highlight the answer with the mouse and copy it. If you can, it is readable page text and counts.
- If the answer cannot be selected, it does not count. This excludes answers found only in pictures, scanned PDFs, downloadable brochures, or a booking widget that must be opened first.
- An answer that exists only on a booking site, review site, or a brand's central site does not count for this check; it must be on the property's own domain.
- Keep a tally of the questions answered and record the exact page address for every answer.
Complete question set
Why this check uses 16 questions
Parking availability and parking price are separate guest decisions. Wi-Fi and breakfast each require both provision and price, so all sixteen questions must be checked individually.
- What time can I check in?
- What time do I have to check out?
- Can I cancel, and by when?
- Can I bring my dog?
- Is there parking?
- How much is parking?
- Is it accessible for a wheelchair?
- Is there Wi-Fi, and is it free?
- Is breakfast included, and what does it cost?
- What kinds of room do you have?
- Roughly what does a night cost?
- Will you hold a deposit on my card?
- Can I smoke, and where?
- Where exactly are you?
- What is your phone number?
- What is the one thing that makes you worth choosing?
The answers most often missing are parking price, breakfast price, deposit or card-hold information, and the property's distinguishing feature.
Mark: Pass for at least twelve answers; Needs fix for eight to eleven; Fail for fewer than eight.
Fix: Add clear sentences for the missing policies, facilities, prices, room information, contact details, location, and property differentiator.
Property-specific official path
Check: View the homepage source, find the booking call to action, and confirm that it is a real link leading to a usable page for this property without an unintended robots block.
How to do this
- Open the homepage and find the main Book now button.
- Right-click it and choose Copy link address in Chrome, Edge, or Firefox, or Copy link in Safari.
- If there is no copy option, or the option is unavailable, the button is not a normal link. Record a failed finding and stop this check.
- Open a new tab, paste the copied address, and press Enter.
- Inspect the destination. A page prepared for dates for this specific property passes. A brand or group search that requires finding the hotel again needs a fix. An error, blank page, or different property fails.
- Compare the part of the destination address after the domain name with the robots.txt file from check 01. If a Disallow rule covers that path, record it in both checks.
This check confirms that a property-specific path exists and is reachable. It does not establish that a guest completes a booking.
May need help. If copying the link or comparing its path with robots.txt is unclear, a website maintainer should be able to confirm both quickly.
Mark: Pass for a direct property-specific link; Needs fix for an indirect brand or group search; Fail for an empty, script-only, blocked, incorrect, or unusable destination.
Fix: Use a real link to the correct property page. This checks the path only; it does not claim that a booking or transaction occurred.
Canonical property identity
Check: Compare official name, address, telephone, coordinates, website, and parent or brand relationship across the hotel's authoritative sources.
How to do this
- Create a table with six rows: official name, full address, telephone, coordinates, website address, and parent or brand relationship.
- Add a column for every place the property is described: its own website, Google Business Profile, the brand or group's property page when applicable, and the principal booking-site listings it maintains.
- Copy exactly what each source says, including punctuation and abbreviations, rather than filling in what you already know.
- To collect coordinates, open the property in Google Maps, right-click the marker, and select the latitude and longitude shown at the top of the menu.
- Compare each row. Ignore harmless style differences such as St. versus Street, but record anything that could change a guest's decision or ability to arrive, including a different phone number, street number, or map position.
Mark: Pass when material identity fields agree; Needs fix for a limited mismatch; Fail when identity is materially ambiguous or contradictory.
Fix: Designate a source of truth, correct owner-controlled surfaces, and document third-party corrections that are still propagating.
Entity disambiguation
Check: Check that similarly named properties, restaurants, residences, destinations, and parent brands can be distinguished using stable facts, identifiers, and clear page content.
How to do this
- Search the exact property name in a search engine, then repeat the search in a maps application.
- Record every similar result: another property in the group, a same-name hotel elsewhere, a restaurant or spa, a residence, a street, or a district.
- Open the homepage as if you were a stranger. Confirm that the city, full address, and relationship to a parent brand or sister property make the identity obvious.
- If a restaurant or spa has its own pages, check that the pages identify it as part of the hotel rather than as a separate business.
If another entity could reasonably be mistaken for the property and the website does not resolve the confusion, record that ambiguity as the finding.
Mark: Pass when the hotel is unambiguous; Needs fix when one relationship or identifier is weak; Fail when another entity could reasonably be mistaken for the property.
Fix: Strengthen canonical identity, entity relationships, official descriptions, and stable identifiers.
Official source coverage
Check: For every material guest fact, identify an accessible, current, correctable source controlled by the hotel, brand, or responsible operator.
How to do this
- Start with the fact lists created in checks 03 and 04.
- For each fact, record the exact page address where the property publishes it and the person or role able to change it.
- Treat a source as official when the hotel, brand, or operator controls it. The hotel website, a brand property page, and Google Business Profile can qualify; booking sites, review sites, and travel blogs do not.
- Flag every fact for which the only answer appears on a source the property does not control. Cancellation, pet policy, deposit or card holds, and accessibility commonly reveal this gap.
Mark: Pass when every priority fact has an official source; Needs fix for isolated gaps; Fail when important policies or attributes exist only on third-party sites.
Fix: Publish missing facts on an appropriate official page and assign an owner.
Cross-source consistency
Check: Compare priority facts across the hotel site, business listings, booking content, brand systems, and major third-party profiles.
How to do this
- Select eight to ten decision-relevant facts, such as stay times, cancellation, pets, parking and price, Wi-Fi and price, breakfast and price, accessibility, and phone number.
- Open the hotel website, Google Business Profile, principal booking-site listing, brand page when applicable, and booking engine side by side.
- Read the same fact across every source and record the value shown by each one.
- Mark every row where two sources disagree, then state which value is believed to be correct and why.
An outdated phone number on one listing is a limited inconsistency. Conflicting cancellation windows on the hotel website and booking engine are critical because they lead guests and AI systems to different conclusions.
Mark: Pass when no material contradiction remains; Needs fix for a limited inconsistency; Fail when guests or AI systems receive conflicting critical information.
Fix: Correct the source of record first, then submit and track downstream corrections.
Freshness ownership
Check: Confirm that changing facts—policies, prices, facilities, hours, availability signals, and contact details—have an owner, evidence source, review date, and cadence.
How to do this
- List the facts that change over time: rates, policies, facility hours, seasonal closures, contact details, and availability signals.
- For each fact, record four items: the responsible role, where the true value lives, the last review date, and how often or under what condition it should be reviewed.
- Prefer event triggers to calendar reminders when possible. For example, review the public cancellation wording whenever the cancellation policy changes in the operating system rather than only once per quarter.
- If nobody can name an owner for a fact, the missing ownership is the finding even when the current value happens to be correct.
This check examines how the property is operated, so it is completed from internal ownership records rather than from the public website alone.
Mark: Pass when priority facts have current evidence and ownership; Needs fix for incomplete coverage; Fail when material facts are stale and unowned.
Fix: Assign the responsible role and a review trigger for each volatile fact.
Report the result
- Keep all ten individual findings and their evidence.
- State how many checks pass, clearly labeled as a count — not a score.
- Call out a retrieval-access failure separately and fix it first.
- Record training-crawler choices separately; do not grade them.
- Date the assessment because websites and crawler policies change.
PART TWO: Observed-behavior methodology
AI visibility is non-deterministic. The practical question is not “What rank is the hotel?” but whether it appears, is recommended, is represented accurately, is supported by observable sources, and is described appropriately under declared conditions.
Record five outcomes separately
Presence
For discovery prompts that return a usable hotel list, record whether the property appears. Keep each AI engine, market, language, and prompt family separate.
Factual representation
When the property appears, compare material claims with the hotel's approved facts. In the workbook, record how many facts you checked and how many were correct.
Recommendation
Record whether the hotel is actively recommended or shortlisted rather than merely mentioned. Keep recommendation separate from presence.
Observable sources
When the answer shows sources, record them and mark whether the hotel's own website appears. A source is evidence to inspect, not a complete explanation of why the AI answered that way.
Description and tone
Choose Accurate, Materially incomplete, or Misleading for description quality. Record tone as Positive, Neutral, or Negative. These are observations—not scores.
Separate branded prompts from discovery prompts
Branded
Name the hotel
Use branded prompts to inspect what an AI system knows, how it compares and describes the property, and what review themes it reports.
Discovery
Do not name the hotel
Use discovery prompts to test whether the property surfaces for a category, traveler, occasion, or need without being led to it.
Begin with the six-prompt core
What do you know about [hotel name] in [destination]?
How does [hotel name] in [destination] compare to other [category] hotels there?
What do guests say about [hotel name] in [destination]?
What are the top [category] hotels in [destination]?
Best hotels in [destination] for [guest type].
Which hotels would you recommend in [destination] for [occasion]?
A filled-in discovery prompt
“What are the best boutique hotels in Napa Valley for couples?”
Here, the category is boutique hotels, the destination is Napa Valley, and the guests are couples. Replace these details with ones that fit your property. This is an example of tailoring a guest-type discovery prompt, not a seventh required prompt. Once you choose your wording, keep it the same when you repeat the test.
Add only the optional prompts that fit the property. A practical working set often grows to roughly 15–25 prompts for a smaller property. The prompt set is a framework to tailor — not a universal fixed checklist.
For a separate factual spot-check, use: “What is the [fact] at [hotel name] in [destination]?” Apply it only to approved facts guests actually decide on, then repeat it after corrections or material changes.
Expand by angle only when it reflects a real guest decision
| Prompt class | Angle | What it examines | Placeholders |
|---|---|---|---|
| Branded | Grounding | What does the system know about this hotel? | Hotel name + destination |
| Branded | Comparison | How does it position the hotel against its set? | Hotel name + destination + category |
| Branded | Reviews and reputation | What themes and sentiment does it report? | Hotel name + destination |
| Discovery | Category | Does the hotel surface for its actual category? | Destination + category |
| Discovery | Guest type | Does it surface for a relevant traveler? | Destination + guest type |
| Discovery | Occasion | Does it surface for a relevant trip occasion? | Destination + occasion |
| Discovery | Need or amenity | Does it surface for a real decision criterion? | Destination + need or amenity |
| Discovery | Brand or loyalty | Does it surface when a guest explicitly asks for a brand or program? | Destination + brand or loyalty program |
Brand and loyalty caution: use a brand or loyalty prompt only when the relationship is real and material to the guest. Do not treat a brand-led result as proof of generic discovery visibility.
Fill each placeholder from approved property facts
[hotel name]The official property name used across authoritative sources.
[destination]The market a traveler would naturally use: city, district, resort area, or landmark.
[category]The hotel's genuine operating category or competitive set—not an aspirational label.
[guest type]A traveler the property actually serves, such as families, business travelers, or guests with accessibility needs.
[occasion]A relevant trip purpose, such as a conference, anniversary, ski trip, or weekend stay.
[need or amenity]A material decision criterion supported by approved facts.
[fact]One factual field to verify separately, such as check-in, cancellation, parking, Wi-Fi, pets, or breakfast.
Keep test conditions consistent
Fresh context
Use a fresh chat for every prompt so a prior answer does not influence the next run.
Account state
Test logged out where possible. Otherwise use a dedicated account with memory and personalization off.
Same setup
Keep the AI engine, product mode, market, location, and language the same during a baseline. Record that setup once on the workbook's Self Test Conditions tab.
One controlled prompt
Keep the prompt wording and placeholders unchanged inside a baseline. If wording changes, start a new comparison series.
Which answers count
Mark an answer Usable for this test? = Yes when the AI gives an answer. For discovery prompts, the answer must also list hotels. Rows marked No stay in the log but are left out of the rates.
Baseline volume
Use 10–14 independent runs per prompt and engine across approximately two weeks when a stable baseline is needed.
Simple run log
For each run, record the date and prompt; whether the answer was usable; whether the hotel appeared or was recommended; description quality and tone; any sources shown; fact counts when checked; and a short note or pasted answer.
Repeat cadence
Use a monthly review as the operating default. Use weekly monitoring only during active remediation, after a major launch or material change, or when paid monitoring requires it; keep the method fixed.
Use rates only when the denominator supports them
One response is recorded as a finding. After enough usable independent runs, the workbook calculates per-engine rates from the linked run log. Rows marked not usable remain visible but are left out of the rates. These rates are observations, not a self-assessment score.
| Observation | Formula |
|---|---|
| Discovery presence | Usable discovery runs in which the hotel appears ÷ usable discovery runs |
| Discovery recommendation | Usable discovery runs in which the hotel is recommended ÷ usable discovery runs |
| Hotel-website source | Usable appearances that show sources and include the hotel website ÷ usable appearances that show sources |
| Positive tone | Usable appearances marked Positive ÷ usable appearances with Positive, Neutral, or Negative recorded |
| Fact accuracy | Facts recorded as correct ÷ facts checked |
Prioritize engines deliberately
Priority 1
The AI engines and search assistants most used by the hotel's target guests in its priority markets.
Priority 2
Additional engines important to a specific region, language, guest segment, brand, or distribution context.
Watch list
Emerging surfaces worth observing, without mixing them into an established baseline until the protocol is stable.
What this self-assessment does not do
- It does not create a universal AI visibility rank or composite score.
- It does not compare one property with another property.
- It does not prove why a model produced an answer.
- It does not treat a citation, recommendation, or favorable description as proof of a transaction.
- Transaction execution, payment, booking completion, conversion, revenue attribution, and emerging commerce protocols remain outside this public flow.
Correct, wait, and repeat
- Correct the controlling source, not only the symptom in one answer.
- Record the correction and the expected propagation window.
- Repeat the same prompt under the same declared conditions.
- Report each engine, prompt family, market, and language separately.
Use the self-assessment workbook
XLSX
- What it is
- The workbook contains eight connected tabs: Usage, Readiness & Eligibility, Observed Behavior, Placeholders, Prompts, Self Test Conditions, Dashboard, and ChatGPT.
- When to use it
- First inspect what the hotel publishes and controls. Then inspect what AI systems actually say. Keep the two sets of findings separate.
- What you enter
- Record evidence, the finding, the fix, an owner, and the date. Run a declared prompt set under consistent conditions. Record each AI engine separately.
- What it produces
- The Dashboard is linked by formulas to the per-run ChatGPT log. It keeps readiness findings separate from observed behavior and reports per-engine rates without a composite or comparative score.
A self-assessment result is never converted to a score, never compared with a conformance score, and never compared between properties.
Check Google Free Booking Links as action-readiness evidence
Google Free Booking Links can expose a direct supplier's name, an itinerary rate, and a landing page without charging for the click. Participation commonly depends on a reservation system, CRS, booking engine, franchise, or integration provider supplying accurate rates and landing-page information to Google.
Confirm
Is the property participating?
Ask the booking-engine, CRS, franchise, or integration provider whether Free Booking Links are active for the property and who supplies the rates and landing pages.
Verify
Are the submitted facts current?
Check the displayed itinerary rate, direct landing page, property identity, and applicable policies against the hotel's live approved sources.
Identify
Who holds the Official site treatment?
Record which owner, direct supplier, CRS, or integration partner is treated as the official direct source where that designation appears.
Record
Keep the evidence in the readiness record
Capture the date, market, search conditions, displayed rate, landing page, provider, evidence, owner, and any discrepancy requiring correction.
Keep the boundary clear
Participation, an Official site designation, a displayed rate, or a Free Booking Link click is action-readiness evidence only. It does not prove that an AI answer mentioned or recommended the hotel, and it does not prove transaction completion, conversion, or revenue.