AI HOSPITALITY ALLIANCE · WORKGROUP 7 · IMPLEMENTATION GUIDE

AI Readiness Guide for Hotels

Understand how your hospitality operation works today — and what it needs to strengthen before it can absorb AI.

Includes free, reusable tools for each repeatable part of the workflow.

1. Operational readiness ↓2. Digital readiness →

SegmentSMB/independentEnterprise

AudienceOwners, GMs & department headsRevenue, marketing, IT & web teams

Operating levelProperty levelCorporate level

AI readiness

Overview

This free, neutral self-assessment tells a hospitality business where it stands and which part of the operation to strengthen first. It is for both SMB/independent and enterprise hospitality businesses. At property level, owners, GMs and department heads describe how work runs. At corporate level, commercial, revenue, marketing, IT and web teams answer for shared systems, policies and websites. Independents may own these decisions locally; brand-managed and franchised properties should involve the corporate team or provider that controls them.

It is a diagnosis, not a plan: it does not design workflows or estimate their value. It is a questionnaire, not a scan: it does not crawl a website or connect to your systems.

AI readiness needs to be assessed on two fronts: your property-level operation and your digital presence. They measure different things and are never averaged into one readiness number.

This page covers stage 1 - the first of the two assessments: how your operation actually works, and whether it could absorb AI at all. It explains how to run the session, who needs to be in it, how long it takes, and how to answer so the result is worth having.

When you're done with this part of the assessment, use the header to move to stage 2 - digital readiness.

Step 01

Prepare the operational session

54 questions about the inside of your business — how information moves between departments, what is written down, what depends on one person knowing something, and what gets measured. Most of them never mention AI, which is deliberate: asking an operator to rate their AI maturity produces an opinion, while asking what happens when a room is not ready produces evidence.

The companion assessment covers your digital and website readiness — whether AI systems can find, verify and cite your business. It is scored separately, on a different scale, and the two results are never combined into one number.

What you are committing to

About three hours of a facilitated session, plus the time to pull the right people into a room.

Allow time before the session to agree how evidence will be handled, and afterwards to review the scores. The core evidence comes from the conversation itself.

How the six scoring lenses work

Each operational question points to one scoring lens in the workbook. These codes describe what the answer is testing; they are not six extra sections to complete.

Code	What it measures
STR	Strategy and leadership: direction, vision, priorities and leadership ownership of change.
GOV	Governance and decision rights: policies, approval routes, compliance, risk and clarity over who decides.
SYS	Systems and data: the tools in place, how integrated they are, and whether data flows without re-keying.
PRO	Process maturity: how consistently a way of working is defined, followed and improved.
PPL	People and capability: skills, confidence, time and role clarity among the people who use and supervise the work.
MEA	Measurement and insight: whether performance is tracked, understood and used to drive decisions.

MEA is also the confidence check across the result: strong scores in other areas are harder to rely on when the business cannot show what it measures or how it uses that evidence.

The single biggest thing you can do to make this assessment worth having is to stop treating it as a form and run it as a facilitated conversation with the people who do the work.

The questions were written to be asked aloud, with follow-ups. Typed into a box alone, “how does housekeeping tell the front desk a room is ready” comes back as “verbally”, which cannot be placed against a five-band scale by anyone. Asked in a room, the same question produces the radio call, the PMS lag, the twice-a-week failure and the duty manager who fixes it — which is a score, and a finding.

It also surfaces the disagreements, and those are among the most useful things the assessment produces. When the housekeeping manager and the front office manager describe the same handover differently, that gap is the finding. A form collects two answers and averages the difference away.

How to run it

Book one session, or two shorter ones, and bring the people listed in the section table below into the room or the call. Not everyone needs to be there for all of it — the table below shows which sections need whom, so people can join for their part and leave.
Work through the questions out loud in order. Let people interrupt each other — the cross-talk is where the detail lives.
Nominate one person to capture the evidence and, for each question, who actually controls the thing being described: this property, corporate, an outsourced provider, or a mix of them.
Allow about three hours for this half. It is a long session, and it is roughly the only time you will spend on this.

XLSXReusable download

Operational readiness workbook

What it is: A structured workbook containing the 44 property-level questions, 10 revenue questions, scoring rubrics and reporting crosswalk.
When to use it: Use it to run and score the operational half of the readiness assessment.
What you enter or decide: Record what happens today, the supporting evidence, who controls each capability and the lower supported score band.
What it produces: Separate Operations and Revenue percentages calculated from accepted scores, evidence coverage and flagged input issues. A person interprets the findings.

This is a self-assessment workbook, not an automated scan or certification.

Download XLSX ↓

Step 02

Bring the right people

The workbook contains eighteen property-level themes and ten additional revenue and commercial questions. Nobody needs to sit through all of them; use this table to schedule each person for the parts they can evidence.

Section or revenue question	Qs	Who needs to be in the room
Group Alignment & Decision Rights	3	Owner / MD / GM; Owner / MD / GM plus all department heads
Strategic Vision & Leadership	3	Owner / Finance lead; Owner / MD / GM
Technology Stack & Integrations	3	Department heads, each in turn; IT lead or the person who owns systems
Data Infrastructure & Governance	3	Front office / Reservations; Owner / MD / GM or data owner; Reservations / Marketing / Front office
Guest Experience & CX Technology	3	Front office manager; Front office manager / Duty manager
Rooms Division & Front Office	3	Front office manager; Housekeeping manager plus front office manager; Maintenance / Engineering lead
Revenue Management & Commercial	2	Revenue lead / Commercial lead
Sales, Marketing & Digital	2	Marketing lead; Sales / Events lead
AI Visibility, Search & Booking Channels	2	Owner / Marketing lead; Revenue / Marketing lead
Food & Beverage Operations	2	F&B manager; F&B manager plus Head chef
Wellness, Leisure & Signature Experiences	2	Department heads, each in turn; Spa / Leisure / Activities lead
HR, Workforce & Training	3	Department heads, each in turn; HR lead / GM; HR lead plus department heads
Finance & Reporting	2	Finance lead; Finance lead plus GM
Operational Automation	3	Department heads, each in turn
Cybersecurity & Compliance	2	IT lead / HR lead; Owner / MD / GM
Sustainability & ESG	2	Facilities lead / GM; GM / Finance lead
Innovation & Change Readiness	2	Department heads, each in turn; GM plus department heads
Leadership & Executive Effectiveness	2	Owner / MD / GM; Owner / MD / GM plus department heads
Agent-transactability	1	IT / Distribution lead
AI-ready data	1	IT / Distribution lead
AI unit economics	1	Owner / Finance / Commercial
Buy-well and lock-in	1	IT lead / Owner
Human oversight of AI pricing	1	Revenue / Commercial lead
Guest-facing AI disclosure	1	Marketing / GM
Pricing sophistication	1	Revenue / Commercial lead
Pricing outcome discipline	1	Revenue / Commercial lead
Market & pace inputs	1	Revenue / Commercial lead
Total revenue beyond the room	1	Revenue / Commercial lead

The one that catches people out

Several questions are asked of every department head in turn rather than once for the property — which task in the department is done identically every time, where the team has built its own workaround, and what should come off their plate first.

If you have eight department heads, that is eight answers, not one. Warn people before the session so the most valuable part is not rushed.

Who attends which parts

This second view groups the same 54 questions by the department or function that needs to answer them. Questions appear in every relevant row when more than one function is named, so the list overlaps by design.

Department or function	Who to invite	What they cover	Question IDs
Leadership and ownership	Owner, MD, GM and the relevant decision owners	Decision rights, strategy, investment, governance, change readiness, executive effectiveness, AI unit economics, lock-in and guest-facing AI disclosure.	OPS-01–06, OPS-12, OPS-23, OPS-29, OPS-33, OPS-38–41, OPS-43–44; REV-03–04, REV-06
All department heads	Each department head for their own operation; bring the group together where the question says so	System friction, repeated work, workarounds, role-dependent knowledge, rotas, automation opportunities, digital capability and change.	OPS-03, OPS-07–08, OPS-28, OPS-30–31, OPS-34–36, OPS-41–43
IT, systems, data and distribution	IT lead, systems owner, data owner and distribution lead as applicable	System integration, data responsibility, access control, end-to-end transactability, rate and room identity, export rights and open interfaces.	OPS-09, OPS-12, OPS-37; REV-01–02, REV-04
Front office, reservations and rooms	Front office, reservations, duty management and housekeeping leads	Guest data capture, the enquiry-to-arrival journey, message handling, service recovery, room-status handover and daily measures.	OPS-10–11, OPS-13–16, OPS-18
Revenue and commercial	Revenue, commercial and distribution leads	Rate setting, forward view, AI-assisted booking attribution, unit economics, pricing oversight, pricing sophistication, outcome discipline, market inputs and total revenue.	OPS-19–20, OPS-24; REV-03, REV-05, REV-07–10
Marketing, sales and events	Marketing lead and sales / events lead	Guest lists and segmentation, event enquiries, AI visibility accountability, booking attribution and disclosure of automated guest interactions.	OPS-10, OPS-21–24; REV-06
Finance	Finance lead, with the GM or commercial owner where named	Technology investment, invoice processing, management reporting, sustainability reporting and AI unit economics.	OPS-05, OPS-32–33, OPS-40; REV-03
HR and people	HR lead and the relevant department heads	Onboarding, capability with existing tools and removal of access when someone leaves.	OPS-29, OPS-31, OPS-37
Food and beverage	F&B manager and Head chef	Allergy and dietary-information handoff, covers forecasting and staffing.	OPS-25–26
Facilities, maintenance and engineering	Maintenance / Engineering lead and Facilities lead	Maintenance reporting and recurring faults; energy, water and waste measurement.	OPS-17, OPS-39
Spa, leisure and activities	Spa / Leisure / Activities lead	Experience booking and its connection to the room booking.	OPS-27

A coordinator assigns sections to the relevant colleagues. Record each department head’s answer separately where a question calls for every department head, and retain disagreements rather than averaging them away.

Step 03

Capture what actually happens

This is the part that determines whether the result is worth anything.

Record and transcribe it

Record and transcribe the session when appropriate and authorised. If recording is inappropriate, nominate an evidence recorder to capture detailed, question-linked notes and confirm them with participants. The manual workbook route does not require an AI upload; the scoring prompt requires a safely prepared transcript.

The reason is simple. Nobody can capture what fourteen people say in three hours while also facilitating, and the difference between a scoreable answer and an unscoreable one is usually a detail somebody mentioned in passing. The transcript keeps it.

The transcript then becomes the input to scoring. Whether a person scores it or a language model does, working from what was actually said — rather than from notes someone summarised under time pressure — is what makes the evidence real and the score arguable.

Tell people they are being recorded

Say it at the start of the session, and say what the recording is for and who will see it. In many places you need everyone's agreement before you record them, and in some you need it explicitly.

It is also the right thing to do irrespective of the law. People answer these questions honestly when they understand the recording exists to capture their operation accurately, not to catch them out. Say that too.

Answer with specifics. Name the system, say how often, describe what actually happens, quote the number if you have one. General descriptions cannot be scored, by a person or by a machine — and a question that cannot be scored is one you answered for nothing.

Not scoreable

“How does housekeeping tell the front desk a room is ready?”

“Verbally.”

Scoreable

“How does housekeeping tell the front desk a room is ready?”

“Radio to the front desk, then the supervisor updates the PMS at the end of the round. It's wrong maybe twice a week, usually because the update lags the radio call, and the duty manager sorts it out.”

The second answer takes fifteen seconds longer and contains everything needed to place that part of the operation accurately. The first contains nothing.

A few rules the scoring follows

What happens today, not what is planned. “We're planning to” is scored as the current state.
The weakest part of the answer sets the score. If something works in one department and not another, the assessment reflects the gap rather than the best case.
Adjectives are not evidence. “Robust”, “seamless” and “fully integrated” carry no information. Named systems, stated frequencies and described steps do.
Where an answer sits between two levels, use the lower band unless stronger evidence supports the higher one. The assessment is deliberately conservative, so a strong result means something.

Record who controls it

For every question, note whether the thing being described is controlled by this property, decided at corporate level, run by an outsourced provider, or some mix. It takes a second per question and it changes what the result is worth.

A low score on something you control is a task. The same score on something your brand or your agency controls is a conversation with them, and no amount of effort at property level will move it. Without that distinction the report hands you a list of things to fix, some of which you cannot fix.

If you cannot answer something

Skip it. Unanswered questions are removed from the total rather than counted as zero. You are never marked down for a question nobody at your property could reasonably answer, and the assessment will never quietly turn a gap in evidence into a confident number.

If a question does not apply — no spa, no F&B, no events team — skip it for the same reason. Do not guess to fill a blank; a guessed answer damages your result in a way a skipped one does not.

Protect the evidence

Before recording or sharing evidence

Use only the evidence needed for this assessment. Recording and AI-assisted scoring are choices, not a requirement to share personal or confidential information.

Explain the intended use first. Before the session, explain whether it will be recorded or transcribed, whether an AI service will be used, who will receive the material and how long it will be kept. Follow your organisation's recording rules and obtain any required agreement. Permission to record does not automatically authorise sharing with an AI provider.
Prepare a safe copy before uploading. Remove guest identifiers, unnecessary staff identifiers, sensitive personal information, passwords, access tokens and unnecessary confidential business details. Keep the facts needed to judge the workflow. Replace names with roles where useful, but remember that a role or incident may still identify someone. Do not send a raw transcript to an unapproved AI service just to have it redact the data.
Check the service and account. Use a service and account approved by your organisation. Check who can access uploads, whether they may be used for training, retention and deletion terms, and any connected tools or onward sharing. If personal data remains, confirm the lawful basis, provider arrangements and any international-transfer requirements with the person responsible for privacy before sharing.
Limit access and retention. Keep recordings, transcripts, workbooks and reports in approved storage with access limited to the people who need them. Set a retention period, then delete working copies and provider uploads when no longer needed, subject to applicable obligations and the provider's actual retention terms.
Use the manual route when sharing is inappropriate. Keep the assessment in approved local storage and have a person apply the workbook rubric. If recording is inappropriate, nominate an evidence recorder and confirm detailed, question-linked notes with participants. Leave any answer without enough evidence unscored. No upload to an AI service is required for manual scoring.

Where your information goes

This guide has no transcript-upload form. If you follow an AI-tool link or use another language model, any files you submit are handled by that service under its account settings and terms. These instructions are not a guarantee about a provider's practices or a substitute for your organisation's privacy review.

Guidance: ICO guidance on transparency in AI ↗; European Commission guidance on data-protection principles ↗.

Readiness has two axes: property versus corporate responsibility, and controlled versus outsourced execution. Capture Property, Corporate, Outsourced or Mixed for each answer; use Not established when the evidence does not identify control.

All 54 operational and revenue questions

1. Group Alignment & Decision Rights · 3 questions

OPS-01

If you wanted to change a system that touches guest data, who has to say yes, and how long does that normally take?

Asked of: Owner / MD / GM

Reports to: D5 Governance and risk · Both · Rubric GOV

OPS-02

Which technology decisions are made at this property and which are made above it, and is that written down anywhere or just understood?

Asked of: Owner / MD / GM

Reports to: D5 Governance and risk · Both · Rubric GOV

OPS-03

Where does the business draw a hard line on automation, meaning which guest moments must stay human no matter what the technology can do?

Asked of: Owner / MD / GM plus all department heads

Reports to: D1 Strategy and leadership · Operations · Rubric STR

2. Strategic Vision & Leadership · 3 questions

OPS-04

Is there a written plan for how technology should change the way this business works over the next two years, and is it linked to any numbers?

Asked of: Owner / MD / GM

Reports to: D1 Strategy and leadership · Both · Rubric STR

OPS-05

Is there money set aside for technology and improvement work, separate from keeping the current systems running?

Asked of: Owner / Finance lead

Reports to: D5 Governance and risk · Both · Rubric GOV

OPS-06

When you have bought technology before, how did you know afterwards whether it was worth the money?

Asked of: Owner / MD / GM

Reports to: Outcome evidence, all dimensions · Both · Rubric MEA

3. Technology Stack & Integrations · 3 questions

OPS-07

Walk me through every system your team logs into on a normal shift, and tell me which two are the most annoying to use together.

Asked of: Department heads, each in turn

Reports to: D3 Systems and integration · Operations · Rubric SYS

OPS-08

Where does someone type the same information into a second system by hand, and roughly how often?

Asked of: Department heads, each in turn

Reports to: D3 Systems and integration · Both · Rubric SYS

OPS-09

When you last added a new system, did it connect to what you already had, or does it sit on its own?

Asked of: IT lead or the person who owns systems

Reports to: D3 Systems and integration · Both · Rubric SYS

4. Data Infrastructure & Governance · 3 questions

OPS-10

If I asked for a list of every guest who has stayed twice in the last year and told you their dietary requirements, where would you get it and how long would it take?

Asked of: Reservations / Marketing / Front office

Reports to: D2 Data foundations · Both · Rubric SYS

OPS-11

When a guest tells you something about themselves, where does it get written down, and who sees it next time they visit?

Asked of: Front office / Reservations

Reports to: D2 Data foundations · Both · Rubric PRO

OPS-12

Who is responsible for making sure guest and employee data is held correctly and deleted when it should be, and is that written down?

Asked of: Owner / MD / GM or data owner

Reports to: D5 Governance and risk · Both · Rubric GOV

5. Guest Experience & CX Technology · 3 questions

OPS-13

Take me through what happens from the moment a guest enquires to the moment they arrive in their room, step by step, including who touches it and where it can go wrong.

Asked of: Front office manager

Reports to: D6 Operations and process · Both · Rubric PRO

OPS-14

How does a guest message you, how many different places do those messages land, and who is watching each one?

Asked of: Front office manager

Reports to: D3 Systems and integration · Both · Rubric SYS

OPS-15

How do you find out that a guest was unhappy, and what happens between finding out and doing something about it?

Asked of: Front office manager / Duty manager

Reports to: D6 Operations and process · Operations · Rubric PRO

6. Rooms Division & Front Office · 3 questions

OPS-16

How does housekeeping tell the front desk a room is ready, how often is that wrong, and what happens when it is?

Asked of: Housekeeping manager plus front office manager

Reports to: D6 Operations and process · Operations · Rubric PRO

OPS-17

How does a maintenance issue get reported, and could you tell me which fault has come back most often this year?

Asked of: Maintenance / Engineering lead

Reports to: Outcome evidence, all dimensions · Operations · Rubric MEA

OPS-18

What does the front office team measure daily, and where do those numbers come from?

Asked of: Front office manager

Reports to: Outcome evidence, all dimensions · Operations · Rubric MEA

7. Revenue Management & Commercial · 2 questions

OPS-19

Who sets rates, what information do they look at first, and how much of that is pulled together by hand?

Asked of: Revenue lead / Commercial lead

Reports to: D6 Operations and process · Revenue · Rubric PRO

OPS-20

How far ahead can you see what business is coming in, and how confident are you in that view?

Asked of: Revenue lead / Commercial lead

Reports to: Outcome evidence, all dimensions · Revenue · Rubric MEA

8. Sales, Marketing & Digital · 2 questions

OPS-21

When an enquiry comes in for an event or group booking, tell me every step from arrival to signed contract and where it slows down.

Asked of: Sales / Events lead

Reports to: D6 Operations and process · Revenue · Rubric PRO

OPS-22

How do you decide what to send to which guests, and is that a list someone builds by hand each time?

Asked of: Marketing lead

Reports to: D2 Data foundations · Revenue · Rubric SYS

9. AI Visibility, Search & Booking Channels · 2 questions

OPS-23

Who is accountable for how this business appears when someone asks an AI assistant for a recommendation, and is it in anyone's job description?

Asked of: Owner / Marketing lead

Reports to: D5 Governance and risk · Revenue · Rubric GOV

OPS-24

If a guest books after an AI assistant recommended you, would you be able to tell?

Asked of: Revenue / Marketing lead

Reports to: Outcome evidence, all dimensions · Revenue · Rubric MEA

10. Food & Beverage Operations · 2 questions

OPS-25

How does a dietary requirement or allergy given at booking reach the person actually cooking or serving?

Asked of: F&B manager plus Head chef

Reports to: D6 Operations and process · Operations · Rubric PRO

OPS-26

How do you work out how many covers to expect and how many staff to put on, and what do you base that on?

Asked of: F&B manager

Reports to: Outcome evidence, all dimensions · Operations · Rubric MEA

11. Wellness, Leisure & Signature Experiences · 2 questions

OPS-27

How are bookings for treatments, activities or experiences taken, and does that talk to the room booking at all?

Asked of: Spa / Leisure / Activities lead

Reports to: D3 Systems and integration · Both · Rubric SYS

OPS-28

Which parts of the experience you are known for depend on one particular person knowing what to do?

Asked of: Department heads, each in turn

Reports to: D4 People and capability · Operations · Rubric PPL

12. HR, Workforce & Training · 3 questions

OPS-29

Take me through what happens from a new starter accepting the job to their first shift, and how much of that is paper or email.

Asked of: HR lead / GM

Reports to: D6 Operations and process · Operations · Rubric PRO

OPS-30

How are rotas built, how long does it take, and what happens when someone calls in sick at 6am?

Asked of: Department heads, each in turn

Reports to: D6 Operations and process · Operations · Rubric PRO

OPS-31

How confident is your team with the digital tools they already have, and how would you know if someone was struggling?

Asked of: HR lead plus department heads

Reports to: D4 People and capability · Operations · Rubric PPL

13. Finance & Reporting · 2 questions

OPS-32

Talk me through what happens between an invoice arriving and it being paid, and how many people touch it.

Asked of: Finance lead

Reports to: D6 Operations and process · Operations · Rubric PRO

OPS-33

How is the monthly management pack put together, how long does it take, and what happens in the meeting where it is discussed?

Asked of: Finance lead plus GM

Reports to: Outcome evidence, all dimensions · Both · Rubric MEA

14. Operational Automation · 3 questions

OPS-34

If I gave your team back four hours a week, which task would you take off them first, and why that one?

Asked of: Department heads, each in turn

Reports to: D6 Operations and process · Operations · Rubric PRO

OPS-35

Which task in your department is done the same way every single time, with no judgement involved?

Asked of: Department heads, each in turn

Reports to: D6 Operations and process · Operations · Rubric PRO

OPS-36

Where has the team already found their own workaround, a spreadsheet or a group chat, because the system does not do what they need?

Asked of: Department heads, each in turn

Reports to: D3 Systems and integration · Operations · Rubric SYS

15. Cybersecurity & Compliance · 2 questions

OPS-37

When someone leaves, how quickly do their logins get switched off, and who checks?

Asked of: IT lead / HR lead

Reports to: D5 Governance and risk · Operations · Rubric GOV

OPS-38

Is there anything written down that tells staff which AI or online tools they may and may not put business or guest information into?

Asked of: Owner / MD / GM

Reports to: D5 Governance and risk · Both · Rubric GOV

16. Sustainability & ESG · 2 questions

OPS-39

What do you currently measure on energy, water or waste, where does that number come from, and who looks at it?

Asked of: Facilities lead / GM

Reports to: Outcome evidence, all dimensions · Operations · Rubric MEA

OPS-40

Are you asked to report on sustainability by anyone outside the business, and how much work is it to produce that?

Asked of: GM / Finance lead

Reports to: D6 Operations and process · Operations · Rubric PRO

17. Innovation & Change Readiness · 2 questions

OPS-41

Tell me about the last significant change you rolled out here. What worked, what did not, and what would you do differently?

Asked of: GM plus department heads

Reports to: D4 People and capability · Both · Rubric PPL

OPS-42

If someone on the team has an idea for doing something better, what happens to it?

Asked of: Department heads, each in turn

Reports to: D6 Operations and process · Operations · Rubric PRO

18. Leadership & Executive Effectiveness · 2 questions

OPS-43

How much of a typical week do you and your managers spend gathering or formatting information, rather than deciding anything with it?

Asked of: Owner / MD / GM plus department heads

Reports to: Outcome evidence, all dimensions · Both · Rubric MEA

OPS-44

Which decision do you make regularly where you know the information you are working from is incomplete?

Asked of: Owner / MD / GM

Reports to: Outcome evidence, all dimensions · Both · Rubric MEA

Revenue and commercial readiness · 10 questions

REV-01

Has anyone tested whether a booking can be completed end to end without a person touching it, and what happened?

Asked of: IT / Distribution lead

Reports to: D3 Systems and integration · Revenue · Rubric SYS

REV-02

Is your rate and room identity consistent across channels (CRS, OTAs, brand.com, Google), and fresh enough in near-real-time for a system or agent to act on without stale-inventory errors?

Asked of: IT / Distribution lead

Reports to: D2 Data foundations / D3 · Revenue · Rubric SYS

REV-03

For the last tool you bought that you pay for per use or per booking, did you know that unit cost before you signed, and did it hold once it went live?

Asked of: Owner / Finance / Commercial

Reports to: D1 Strategy and leadership · Revenue · Rubric MEA

Uses the MEA rubric and reports to D1 Strategy and leadership.

REV-04

When you buy an AI tool, can you judge its claims and leave it if needed — data-export rights, open interfaces, no hard lock-in?

Asked of: IT lead / Owner

Reports to: D5 Governance and risk · Revenue · Rubric GOV

REV-05

Can the revenue team tell when an AI pricing or offer recommendation is wrong, and override it — rather than accepting it because the tool produced it?

Asked of: Revenue / Commercial lead

Reports to: D4 People and capability · Revenue · Rubric PPL

REV-06

When a guest interacts with an automated system, do they know it's automated, and who decides what it is allowed to say on the business's behalf?

Asked of: Marketing / GM

Reports to: D5 Governance and risk · Both · Rubric GOV

REV-07

Think of the last real demand swing you priced for; what did you actually vary: rate by segment, channel, length of stay, lead time — or did one rate move for everyone?

Asked of: Revenue / Commercial lead

Reports to: D6 Operations and process · Revenue · Rubric PRO

REV-08

The last deliberate rate or strategy change you made; how did you tell whether it worked, and what did you compare it against?

Asked of: Revenue / Commercial lead

Reports to: Outcome evidence, all dimensions · Revenue · Rubric MEA

REV-09

When you price, do you look at your comp set and market demand alongside your own on-the-books pace; and where does each input come from?

Asked of: Revenue / Commercial lead

Reports to: D1 Strategy and leadership · Revenue · Rubric STR

REV-10

Is revenue managed past the room rate: ancillary, upsell, total RevPAR, or is room rate the main lever? Point to the last upsell or ancillary push and what it did.

Asked of: Revenue / Commercial lead

Reports to: D1 Strategy and leadership · Revenue · Rubric STR

Additional workflows to discuss if relevant

Optional and unscored. These discussion prompts are not assessment questions. They have no score, band or effect on question counts, coverage or results. Keep their observations separate from the scored question table; do not invent question IDs or let an unscored finding change a scored priority. Skip workflows the business does not have.

Package inclusions and unused benefits

Involve: Revenue / Reservations / Finance

For a recent reservation package, how were included benefits tracked and any guest changes or unused benefits handled? Explain what your team means by breakage, if you use that term.
Who adjusts the reservation, entitlements, guest charges or revenue allocation where relevant? Which systems receive the change, which steps are manual, and who checks the outcome?

Related context: OPS-19 and REV-10. Neither explicitly assesses package reconciliation.

Restaurant reservations and changing capacity

Involve: Food and Beverage / Restaurant reservations

How are restaurant reservations booked and adjusted against availability? When staffing or capacity changes, who updates availability, where does that change appear, and which steps are manual?

Related context: OPS-26. Forecasting covers and staffing is distinct from updating bookable availability.

Writing job descriptions

Involve: HR / Hiring manager

Walk through the last job description: who drafted, reviewed and approved it, what tools were used, and which steps were automated or manual? Who checked that it reflected the actual role?

Related context: OPS-29. Recruitment before an offer is accepted is outside the onboarding question.

Reviewing applications

Involve: HR / Hiring manager

Without sharing applicant records, explain how resumes are reviewed: who sets the criteria, what tools assist screening, which steps are automated or manual, and who reviews recommendations and makes the decision?

Related context: OPS-29. This is a separate recruitment workflow, not an additional onboarding scoring requirement.

Step 04

Score against the evidence

Every question is scored from 0 to 10 against the same five levels, so the number reflects what is actually happening rather than anyone's impression of it.

Level	What it means in practice
Not in place	It does not exist, and it has not been considered.
Early or ad hoc	One person does it informally. Nothing is written down.
Developing	Some structure exists, but it is not consistent, connected or measured.
Established	Structured, consistent, and delivering results. Room to optimise.
Leading	Best in class: measured, largely automated, improving continuously.

Those roll up into an overall position — Initial, Emerging, Developing, Established or Leading — plus a breakdown by area. The breakdown is the part worth reading. A single headline figure is useful for getting attention in a leadership meeting; the detail underneath tells you what to do on Monday.

Question bands: Not in place (0–2), Early or ad hoc (3–4), Developing (5–6), Established (7–8), Leading (9–10). Use the lower band unless stronger evidence supports the higher one.

STR · Strategy and leadership: direction, vision, priorities and leadership ownership of change.

Not in place · 0–2: No stated direction. Not discussed at leadership level, or dismissed.
Early or ad hoc · 3–4: Interest exists but informal: conversations, no written intent, no owner.
Developing · 5–6: Direction articulated verbally and broadly shared, but not written down, not linked to KPIs, no review rhythm.
Established · 7–8: Written strategy or clear stated priorities, linked to business objectives, with a named owner and periodic review.
Leading · 9–10: Strategy written, KPI linked, reviewed at least quarterly, visibly championed by leadership and understood at department level.

GOV · Governance and decision rights: policies, approval routes, compliance, risk and clarity over who decides.

Not in place · 0–2: No policy, no defined route. Decisions ad hoc or blocked. Compliance reliant on memory.
Early or ad hoc · 3–4: Informal understanding of who decides. Policies exist somewhere but unread or outdated.
Developing · 5–6: Documented policies and routes exist but with gaps or inconsistent application.
Established · 7–8: Clear documented decision rights, current policies including data and tool use, followed in practice, periodically reviewed.
Leading · 9–10: Governance enables rather than blocks: fast defined routes, living policies, evidence of testing, staff confident in what is permitted.

SYS · Systems and data: the tools in place, how integrated they are, and whether data flows without re-keying.

Not in place · 0–2: Paper, memory or personal spreadsheets. No system of record.
Early or ad hoc · 3–4: Standalone systems in place but isolated. Heavy manual re-entry between them.
Developing · 5–6: Core systems established with some integration. Meaningful manual bridging remains in known places.
Established · 7–8: Integrated stack with a recognised single source of truth. Re-entry is the exception and is on a fix list.
Leading · 9–10: API first, integrated, data clean enough to feed analytics and automation. New tools connect rather than bolt on.

PRO · Process maturity: how consistently a way of working is defined, followed and improved.

Not in place · 0–2: No defined process. Outcomes depend entirely on who is on shift.
Early or ad hoc · 3–4: A way of working exists but lives in one or two heads. Undocumented, inconsistent.
Developing · 5–6: Documented but followed unevenly, or documented long ago and drifted.
Established · 7–8: Documented, trained, followed consistently, with basic measures of whether it works.
Leading · 9–10: Consistently followed, measured, automated where sensible, and actively improved on a cycle.

PPL · People and capability: skills, confidence, time and role clarity among the people who use and supervise the work.

Not in place · 0–2: No capability in place. Nobody owns it and nobody has been trained.
Early or ad hoc · 3–4: One or two capable individuals. Knowledge is personal, not organisational.
Developing · 5–6: Some training delivered but uneven adoption. Capability not linked to role or workflow.
Established · 7–8: Role based capability with named ownership, practised in the workflow and supported when it goes wrong.
Leading · 9–10: Capability is designed into roles, refreshed on a cycle, and the organisation can absorb change without depending on individuals.

MEA · Measurement and insight: whether performance is tracked, understood and used to drive decisions.

Not in place · 0–2: Not measured. Decisions on instinct or anecdote.
Early or ad hoc · 3–4: Some numbers exist but are rarely reviewed or acted upon.
Developing · 5–6: Regular reporting reviewed by managers. Insight describes what happened rather than driving action.
Established · 7–8: KPIs defined, visible to the right people, reviewed on a rhythm, with documented actions arising.
Leading · 9–10: Near real time visibility, forward looking, and demonstrable decisions changed by the data.

Overall positions are Initial (0–20%), Emerging (21–40%), Developing (41–60%), Established (61–80%) and Leading (81–100%). The percentage is points earned divided by ten times the number of scored questions. Questions without sufficient evidence are excluded from that denominator and reported as gaps, never as zero.

Keep Revenue and Operations separate within the operational report. Measurement is a cross-cutting confidence indicator, not a seventh dimension. The REV-03 question intentionally uses the measurement rubric while reporting to D1.

Calculate scores in the workbook

A person chooses each score using the rubric. In Question set, enter the supporting evidence and score, then select the Evidence state. Open Results to see the percentage and coverage. The formulas check entries and calculate accepted scores; they do not judge the free-text answer.

A supported zero counts as a scored answer. Clear the score for insufficient-detail, not-applicable and not-covered answers.
Untouched rows remain visible as not-covered. Resolve flagged input issues before using a result.
Results show points earned, eligible points and coverage. No scored answers produces no percentage; fewer than half scored means evidence is too thin for a reliable maturity result.
Operations and Revenue results stay separate. The existing scope profiles are also shown separately; Adoption checks do not enter any score.

You can also enter human-reviewed AI scores to cross-check the arithmetic. Keep the scoring method identified and review differences before accepting a report.

Generate a full report

Before using a prompt or the linked GPT: follow the recording and data-handling checklist ↑. Upload only a safely prepared copy to an approved service and account. If sharing is inappropriate, use the manual workbook route instead.

Upload the transcript and only the workbook for the half you are scoring to your chosen language model, then paste Prompt 1. Keep names and sensitive guest information out of what you share. Review the cited evidence and disagreements before accepting a score. An AI-scored run may differ from a self-scored run; always label the method.

The report has two parts: a short management summary explaining the priority and first move, followed by the evidence and handoff details for the team responsible for investigating it.

Prompt 1

Analyze full report and transcript

This operational version uses the shared evidence and scoring rules, with instructions tailored to this assessment.

You are scoring one half of an AI readiness assessment for a hospitality business. The two required attachments are the session transcript and the matching assessment workbook. Use only that workbook for the instrument and only the transcript for evidence.

Do not add, reword or infer questions. Do not assume what a business like this probably does. A detail elsewhere in the transcript may support a question, but every conclusion must trace to words the participants actually used.

Step 1 — Identify the assessment. State the assessment type, workbook filename and version, transcript name and session date if available, and label the method AI-assisted. Confirm that only one assessment workbook is attached. If the workbook and transcript do not match the same assessment, stop and explain the mismatch.

Step 2 — Read the workbook before the transcript. Note every question ID, Asked of role, scoring key or rubric, reporting field and scoring rule. Work through the workbook in question order even when the conversation happened in a different order.

Step 3 — Assign exactly one state to every question:

- scored — the evidence is specific enough to place against a rubric band
- insufficient-detail — the topic was discussed but cannot be placed defensibly
- not-applicable — the business genuinely has no such function or capability
- not-covered — the session did not reach it

Evidence is sufficient when it establishes the present state against the applicable rubric. Useful evidence can include a named system, page or channel; a frequency or quantity; a described sequence; a responsible role; a dated check; or a concrete instance of success or failure. One authoritative fact may be enough for a genuinely binary question. Several vague statements are still insufficient. Never stretch thin evidence into a score.

Step 4 — Score only questions marked scored. Apply every workbook rule: score the weakest evidenced element; treat planned work as the current state; treat adjectives as no evidence; and use the lower band unless stronger evidence supports the higher one. Validate that every score is between 0 and 10 and that its band agrees with the workbook.

Step 5 — Record control. For every question, use the evidence to record Property, Corporate, Outsourced, Mixed or Not established. Do not assume that the property controls a capability merely because a property participant discussed it.

Step 6 — Calculate transparently. Show the scored numerator, denominator and formula. Percentage = points earned divided by ten times the number of scored questions. Exclude insufficient-detail, not-applicable and not-covered questions from the score denominator, but keep each visible in coverage. Never convert an unknown into zero.

Step 7 — Produce an auditable full report in two clearly labeled parts:

Part 1 — Management summary
1. Assessment identity and evidence sufficiency warning, if any.
2. Coverage: counts and percentages for all four states.
3. Results required by the assessment-specific instructions below.
4. The weakest supported area, why it is the priority, and one immediate first move. Keep this to a diagnostic next action, not a sequenced implementation plan. Explain it in plain language for an owner or GM.

Part 2 — Evidence and team handoff
5. A complete question table in workbook order with every ID, state, score, band, control, evidence excerpt or reference, and the Asked of role for unresolved questions.
6. The control split, including Not established, and the calculation details supporting the summary.
7. For the priority finding, identify the relevant question IDs, workflow or page/system, what was observed and when, the responsible role and control route, and what evidence that team needs to check next. Mark missing details as not established. Do not invent technical fixes or implementation instructions.
8. Quality checks: list score/evidence conflicts, unresolved disagreements and any calculation assumptions. Keep this detail available without repeating the full table in the management summary.

If fewer than half the questions are scoreable, state at the top that the evidence is too thin for a reliable maturity result. Still provide the coverage and complete question table, but do not present a confident maturity label.

Constraints — Never blend this assessment with the other half of the framework. Recommend no vendor, product or consultancy. Do not design workflows, estimate benefits, calculate ROI or sequence a roadmap.

Operational assessment instructions:

- Confirm that the workbook contains the 44 property-level OPS questions and 10 revenue REV questions.
- The Adoption checks worksheet is an unscored planning aid. Exclude it from question counts, coverage, scores and maturity labels.
- Optional facilitator follow-ups clarify evidence for the original question and rubric; they are not additional scoring requirements. The Additional workflows worksheet and observations outside an original question's scope are unscored. Keep them separate from the scored question table, with no invented IDs, states, scores or effect on coverage, denominators or diagnostic priorities. Manual work and deliberate verification controls are not automatically weaknesses.
- Keep the property-level Operations result and the Revenue result separate. Do not calculate or present one combined operational headline score.
- Within the property-level result, report D1 through D6 using the workbook's Reports to field. Where a question reports to more than one dimension, state the allocation rather than silently double-counting it.
- Report MEA alongside the six dimensions as measurement confidence, never as a seventh dimension. If MEA is weak while another result is strong, say that the business cannot yet verify that claimed strength reliably.
- Use operational evidence: named workflows, systems, handoffs, frequencies, owners, controls and observed examples. A bare yes or no is not enough for a spectrum question.
- Report property-controlled work separately from corporate, outsourced and mixed responsibilities.

Or use this tool

Hospitality AI Readiness Assessor

Use the GPT with your completed workbook or session transcript to turn the evidence into a readable readiness summary and next-step outline. Review the evidence before sharing or acting on the result.

Use this tool ↗

Complete every assigned section before reading the report. Keep unassigned and skipped questions visible in coverage, and start a new workbook copy when the underlying evidence changes.

Step 05

Read the result and choose a first move

You will get	You will not get
Where you stand by department and by dimension, with revenue and operations reported separately.	A designed workflow for any gap the assessment finds.
A named priority area — the one to start with, and why that one.	A value or savings estimate for any use case.
A first move, expressed as something to do rather than a number to admire.	A sequenced multi-year plan with dependencies.
How much of your weakest area you actually control, versus what sits with corporate or an outsourced provider.	A recommendation to buy any particular product or work with any particular firm.

That boundary is deliberate. A diagnostic can tell you where you stand and what to do first. Designing the work, costing it and sequencing it depends on your property, your people and your commercial position — and an assessment that claimed to do all of that would be sending you down a plan that was never built for your operation.

This is a self-assessment, which means nobody independent is in the room checking the answers. Assessments completed this way are systematically generous — not through dishonesty, but because people describe the version of the operation they intend rather than the one that runs on a wet Tuesday in February when two people have called in sick.

We would rather say that plainly than imply that a self-assessment and an assessed result are the same thing. The assessment is worth exactly as much as the candour you put into it. An inflated score costs you nothing and tells you nothing.

It is worth saying to the people you invite, too. They may assume a low answer reflects badly on them. It does not — it reflects the state of a system nobody has been asked to fix yet, which is the entire point of measuring it.

When the result names your weakest area, there are three legitimate routes and the Alliance ranks none of them: work the breakdown internally as a checklist, engage a vendor, or engage a consultant. This assessment names no company and sells nothing.

The complete worked band anchors are not yet available. Treat the result as an introductory self-assessment, not a validated certification.

Continue with the other half when the relevant team is available: Digital readiness →.

Step 06

Generate a 12-month roadmap

Use the separate advisory prompt with the scored assessment. It sequences work by dependencies, not simply by the lowest score, and does not make the roadmap an AIHA assessment result.

Check whether people can adopt the proposed changes

For each initiative selected for your roadmap, review these five checks with the people who will own and use it. If it has not yet been piloted, record what you need to test rather than assuming people will adopt it.

Leadership alignment. Do managers agree on what the change should improve, and who owns adoption and results?
Workflow adoption. What must staff do differently, and what evidence shows they can sustain it under real service pressure?
Human handoff. When should a person take over, who receives the handoff, and can staff intervene or pause the workflow?
Service consistency. How will you check that service remains consistent across shifts and departments, including when something goes wrong?
Guest impact and feedback. What guest and staff feedback will you review, who reviews it, and how will it lead to changes?

Use the Adoption checks tab in the operational workbook to record evidence, gaps or follow-up, an owner and a review date. Duplicate the tab for each selected initiative.

These are unscored planning checks. They do not change your assessment score, question count or maturity label.

The roadmap keeps the same two-part format: a brief management summary of the proposed sequence, followed by a supporting table with evidence, responsibilities, dependencies and completion checks for the same items.

Apply the same data-handling checklist ↑ to the scored report and any supporting evidence before sharing them with an AI service.

Prompt 2

Generate a 12-month roadmap

This operational version keeps the shared planning safeguards while using dependencies and ownership appropriate to this assessment.

You are helping a hospitality business turn a completed AI readiness assessment into a twelve-month advisory roadmap. Use the scored assessment and its question-level evidence. For an operational roadmap, also use a completed Adoption checks worksheet if supplied, solely as unscored planning evidence. Every proposed item must trace to specific question IDs, scores and evidence.

First validate the input. State the assessment type, version, coverage and scoring method. If fewer than half the questions were scored, or the assessment does not contain question-level evidence, do not produce a confident four-quarter roadmap. Produce an evidence-gathering plan instead and explain what must be established before planning.

Unknowns are not zeroes. Questions marked not-covered or insufficient-detail are matters to investigate, not established deficiencies. Separate property-controlled tasks from corporate or outsourced conversations, escalations and contract points.

Prioritize by dependency, evidence strength, control, operational or guest risk, and practical ability to act — not by score alone. Do not assume that the lowest number is automatically the first move.

Output in two clearly labeled parts:

Part 1 — Management summary
1. Where the business stands: three or four sentences using the assessment's actual bands and coverage.
2. The four quarters: three to five items maximum per quarter, described briefly in plain language. Use stable item labels so the reader can find the supporting detail below.
3. The dependency spine: the two or three items that other work depends on, with the evidence-based reason for the order, and what is deliberately not in the plan and why.

Part 2 — Evidence and team handoff
4. A supporting table for those same items, not an additional task list. For each item include its label, quarter, action, supporting question IDs and scores, evidence, accountable role, control route, prerequisite, evidence of completion and review checkpoint.
5. What to find out before the next review. Identify missing facts rather than inventing system details or technical fixes.
6. Assumptions and confidence: identify every inference and any weak evidence that could change the sequence.

Name no vendor, product, platform or consultancy. Do not invent costs, savings, revenue impact, ROI or payback. Do not promise an outcome by a date. Keep the diagnostic result distinct from the advisory plan.

Close with: This roadmap is advice built on the AI Hospitality Alliance readiness assessment. It is not part of the Alliance framework, which stops at the diagnosis.

Operational roadmap instructions:

- Keep the property-level Operations and Revenue findings visibly separate, even when one dependency affects both.
- Treat MEA as evidence confidence. If measurement is weak, put the ability to establish and review evidence ahead of claims that another capability has improved.
- Sequence operational work using the actual dependencies found in the assessment: decision rights, process ownership, data, systems, people, governance and measurement. Do not impose a generic technology-first sequence.
- Name the department or corporate counterpart responsible for every item.
- If Additional workflows observations are supplied, label them as unscored context to validate. They may inform a supported item's implementation questions, but cannot create a scored finding, change the diagnostic priority or justify an initiative without scored question-level evidence.
- For each selected initiative, include an unscored adoption check covering leadership alignment, workflow adoption, human handoff, service consistency, and guest impact and feedback. Record supplied evidence, gaps or follow-up, owner and review date. If the initiative has not been piloted, identify what must be tested; do not claim adoption has already been demonstrated.
- Use the Adoption checks worksheet if supplied. Where evidence, an owner or a date is missing, mark it not established or to be agreed. These planning checks do not change any assessment score, question count or maturity label.

Contributors

People whose substantive inputs shaped this guide.

Anisha Yadav ↗

First Ten Advisory · Founder

Revenue readiness additions, business profiles, Revenue and Operations separation, and alignment of the revenue questions with the shared scoring tags.

Catherine Donaldson ↗

Canary Technologies · Head of Marketing and Brand

Nontechnical operator audience and practical usability feedback in the workgroup discussions.

Donna Rougeau ↗

Protected by ALFIE · Founder and Chief Architect

Digital readiness questions and methodology, domain-control distinctions and evidence requirements for the teams responsible for a website.

Fran Kiradjian ↗

Boutique & Luxury Lodging Association · Founder and CEO

Independent and boutique-owner usability and stakeholder perspective, including an offer to bring owner feedback into a future pilot.

Francesca Wood ↗

FAI Consultancy · Founder, AI for Hospitality

Operational questions, the FAI AI Readiness Index scoring method and archetype rubrics, the facilitated, evidence-led assessment approach, and review that clarified manual calculation and safer handling of assessment evidence.

Jamie McBride ↗

Grevon · Marketing Director

Feedback on clear, actionable outputs, useful examples and the first practical move after an assessment.

Reg Lenney ↗

Executive Performance & Stability Advisor, Rijooviness

The five-part human-adoption framework, adapted into unscored operational roadmap checks for leadership alignment, workflow adoption, human handoff, service consistency, and guest impact and feedback.

Rick Posey ↗

Cendyn · VP Product

Discussion input on prerequisites, sequencing and the technical and regulatory context for readiness.

Safa Rahal ↗

Accor · Cluster Director of Marketing and Communications

Feedback on organizational differences and distinguishing capability, adoption and outcome evidence.

Todd Herion ↗

AeroPath Consulting LLC · Founder

Workgroup facilitation and consolidation of the two-workbook package, website build specification and published assessment content.

Reviewers

Steven Marais ↗

Noble House Hotels & Resorts · Vice President, Room Operations

Reviewed the guide and contributed practical operational follow-ups on guest data, reviews, room updates, upselling, governance and system reliability, plus unscored discussion prompts on package benefits, restaurant availability and recruitment.