Hotel AI Success Starts with Data Quality and Clear Goals

This series looks at how AI is showing up across hotel operations, especially through the growing number of tech suppliers in the space. Think of it as a practical snapshot based on what we have explored so far — not a comprehensive review of every player.

1. Introduction

A Moving Target

It is important to say upfront that this landscape is evolving quickly. New features, integrations, and even entire products are changing all the time, which means some observations or criticisms in this series may already be outdated by the time you read them. That is simply the nature of the space right now. Feedback from suppliers is very welcome — whether to clarify, challenge, or add nuance — and posts will be updated or corrected where needed.

Why Data Quality Matters

From a technical standpoint, AI performance is fundamentally limited by data quality and data architecture. While many use cases promise strong results, the real constraint is often not the model itself but the underlying data layer: how data is captured, standardised, and governed across systems. In practice, fragmented sources, inconsistent schemas, missing fields, and weak integration pipelines introduce significant noise that reduces value. Without robust processes for data ingestion, normalisation, deduplication, and reconciliation, even advanced AI models will produce unreliable or low-value outputs.

The Integration and Connectivity Challenge

What really stands out is not just the number of solutions available, but how hard it is to make them work together. Hotels rely on a wide mix of systems: from Property Management Systems (PMS) and Channel Managers to Revenue Management Systems (RMS), Point of Sale (POS), Customer Relationship Management (CRM), and Guest Feedback tools. Add marketing platforms, housekeeping and maintenance systems, plus accounting and financial tools, and the supplier list quickly becomes extensive. The real challenge is not selecting each tool in isolation, but integrating them into a cohesive ecosystem that genuinely simplifies operations while keeping costs under control.

What Should Artificial Intelligence Achieve for Hotels?

The core goal of hotel AI is to create a seamless, personalised guest experience while improving operational efficiency and revenue performance. AI should automate repetitive tasks and support better decision-making, freeing staff to focus on high-value human interaction, stronger service, and more sustainable operations. If implemented correctly, AI should enhance hospitality rather than replace it.

2. PMS and CRM at the Core

In my view, the Property Management System (PMS) and Customer Relationship Management (CRM) platform are the two most important elements in the hotel tech stack when it comes to reaching your goals. They hold much of the most useful information about guests and hotel activity, and ideally, all other systems should connect to one of them. PMS and CRM should have profound two-way connectivity.

The PMS Challenge

Building a reliable guest profile. We have recently been working with PMS such as Mews, Cloudbeds, and Opera, although my experience is strongest with Mews. One of the things I appreciate the most with Mews is its API documentation, which makes it easier for new solutions to connect. For repeat guests — who may book through OTAs, direct channels, GDS, or corporate accounts — hotels need both automated and manual processes to merge guest information properly.

The limits of current matching. Mews is strong in this area because it allows two email addresses under one guest profile, which is already an improvement. In many cases, that is still not enough. OTAs often provide single-use email addresses, leaving hotels with only one real contact point. In an ideal world, hotels would be able to store three or four email addresses and multiple phone numbers under one guest profile to facilitate profile merging. Since automated merging is usually based on name, email, or phone number, the ability to store more identifiers makes guest matching far more reliable and the service experience more memorable.

The reporting challenge. The same issue appears when running production reports by company, agency, consortium, distribution channel, market segment, or rate code. Mews offers a range of segmentation fields such as Company, Travel Agency, Origin, and Segments, however pulling a clean total for OTA production is still not always straightforward. In the reservation source, for example, you may get default labels such as "A-Expedia" or "A-Expedia Affiliate Network" at the same level as "Telephone" or "Booking Engine" — making precise reporting harder than it should be.

Why granularity matters. A guest who works for a company like Barclays might book the same hotel by phone, through the hotel website, via an agency using consortium rates, or through the GDS. Capturing that level of granularity and being able to query it accurately is still difficult in many hotel systems. For a medium-sized hotel chain, this often means exporting PMS data into a data warehouse just to produce reliable reports with the right filters and level of detail.

The CRM Challenge

Beyond in-house guests. A PMS is excellent for collecting information about guests who have already stayed at the hotel, but it does not capture the full picture. It often misses prospects, bar and restaurant customers, and potential leads who have interacted with the business without booking a room. That is where the CRM becomes essential.

B2C and B2B data under one platform. For advanced users, a CRM solution such as Salesforce offers flexibility and personalisation options that dedicated hotel CRMs often lack. Many hotels use a distinct CRM for B2C, Sales, and management of Groups and Events rather than forcing everything into one system. A single CRM can provide better visibility across the business, helping identify leads and measure the true value of agencies or corporate accounts more accurately.

Turning signals into opportunities. For example, if a guest books with a corporate email address, that can be detected through the domain name and automatically routed to the sales team for follow-up. That kind of logic turns simple data into commercial opportunity — and it is exactly the type of process AI can help improve when the underlying data structure is strong enough.

3. Why This Matters

PMS and CRM are used here as examples of how complicated it is to unify hotel information and use AI effectively — but there is much more to say about the wider hotel tech stack. Every department has its own needs, and every system brings its own integration and data quality challenges. The point is not that one platform must do everything, but that the data between systems must remain usable, connected, and trustworthy.

4. Looking Ahead

In the next posts in this AI for Hotel series, I will explore the specific ways AI is already implemented within your current solutions and can support different hotel departments and business goals. Topics will include Guest Experience, Customer Service, Personalisation, Revenue Management, Business Intelligence, Marketing, CRM, Reputation Management, Reservations, Sales, Operations, Front Office, Housekeeping, Maintenance, HR, and Staffing.