AI Won't Fix Your Customer
Data Problem.
It Will Expose It.

Everyone is talking about what AI can do for customer experience. Faster service. Better recommendations. Smarter sales conversations. Personalisation at scale. And increasingly, AI agents capable of not simply answering questions, but making decisions and taking action.

But there is a less exciting question businesses need to answer first: What customer truth will that AI be acting on?

If your CRM says one thing, your booking system another, your marketing platform something else and your service team holds information none of those systems can see, adding AI does not automatically create a connected customer experience because it cannot compensate for a fragmented understanding of the customer. It can simply make disconnected decisions faster.

And that is why customer data strategy is becoming more important in the age of AI, not less.

Why does customer data matter for AI?

AI systems need more than access to data. They need reliable customer identity, current context, business rules, consent and permissions to determine what action is appropriate.

If that foundation is fragmented, outdated or poorly governed, AI can automate irrelevant or contradictory customer experiences at greater speed and scale.

Better AI does not automatically create better customer experiences. Better AI working with reliable customer context has a much better chance of doing so. That distinction matters.

What does fragmented customer data look like in practice?

Consider what this looks like in hospitality. Imagine a guest planning a long weekend at a boutique resort. She discovers the property through Instagram, visits the hotel website several times and eventually books a premium room through an online travel agency. Before arriving, she emails the hotel about airport transfers and mentions that she is travelling to celebrate her anniversary. During the stay, she books a spa treatment. She also reports an issue with the air-conditioning on her first evening. The hotel resolves the problem.

The following morning, however, its marketing automation system sends her a generic promotional email offering a discount on the room category she is already staying in. After checkout, another automated message promotes the hotel's spa, as though she had never used it.

Every individual system may technically be working. The social campaign generated discovery. The booking system recorded a reservation. The CRM captured contact information. The spa system processed an appointment. The service team resolved the complaint. The email platform successfully triggered a campaign.

And yet, from the guest's perspective, the experience is disconnected. Now add AI. An AI concierge has access to one set of information. A marketing agent works from another. A revenue-management system sees booking behaviour but not service interactions. A loyalty platform holds another view. The business has increased the intelligence of its individual systems without necessarily increasing their shared understanding of the guest.

That is the problem. The question isn't simply whether the hotel has AI. It is whether its systems understand enough of the guest's current context to make an appropriate decision.

How is the technology market responding?

The major CRM and customer-data platforms are already moving in this direction. Salesforce's AIforce, announced in September 2026, is designed to make the data, workflows, business logic, permissions and governance already held inside Salesforce available to AI interfaces and agents.^[1]

HubSpot has introduced a similar idea through Growth Context, combining customer information with knowledge about the company and its teams so its AI tools operate with broader organisational context.^[2]

Klaviyo has taken another route through what it describes as headless CRM, opening hundreds of tools and APIs so external AI systems can interact with customer data and take actions through Klaviyo without necessarily requiring a marketer to work inside the platform itself.^[3]

Different vendors. Different terminology. But the direction is similar: customer data is evolving from something businesses analyse into context that machines increasingly use to make decisions. That raises the stakes.

Which industries are already experimenting with AI in the customer journey?

Travel and hospitality are already experimenting with this shift. Hilton introduced the Hilton AI Planner in March 2026, allowing travellers to describe the kind of stay they want using natural language and receive destination and property recommendations.^[4] Marriott followed with Ask Bonvoy, a natural-language travel search experience designed to help customers explore destinations and properties conversationally rather than relying solely on conventional search filters.^[5]

Both examples sit largely at the inspiration and planning stage. But the implications become more significant when AI moves deeper into the customer relationship. Air India announced in September 2026 that it was expanding Salesforce Agentforce across customer service after initially using it for refund processing. According to the companies, the system is intended to interpret customer intent, validate requests, connect information across enterprise systems and act within service workflows.^[6]

That represents a meaningful shift. An AI system answering a question is one level of maturity. Interpreting the customer's situation and taking an operational action is another. The reliability of the underlying customer and operational context becomes much more important.

Is personalisation only as good as the context behind it?

Turtle Bay Resort in Hawaii provides a useful hospitality example. According to a Salesforce customer case study, the resort connected guest information within its CRM and used it to support personalised website experiences and AI-assisted concierge responses. The concierge system draws on the resort's knowledge base to help answer questions about dining, amenities, activities and transportation, while staff retain the ability to review and edit responses.^[7]

The more useful lesson is architectural. AI is not operating independently of the customer-data environment. Its usefulness depends on the information, knowledge and operational context available to it.

Recent academic research from the Indian hospitality sector points to the same broader issue. A 2026 study found that perceived value and service expectations played important roles in customer satisfaction, and highlighted customer-data security and trust as important considerations in AI adoption.^[8] Guests do not evaluate AI because it is technically sophisticated. They evaluate the experience the technology creates.

What happens when AI has incomplete customer context?

Return to our hotel guest. Suppose the hotel's AI marketing system sees: premium room booked, spa page visited, high-value customer. Its recommendation may seem perfectly logical: promote a premium spa package. But perhaps the guest has already purchased that treatment. Perhaps she complained about it that morning. Or another system knows she has asked not to receive promotional communications during her stay.

The AI has not necessarily made a poor decision because it lacks intelligence. It has made a poor decision because it lacks context. This becomes increasingly important as businesses move from AI that recommends actions to AI that executes them.

What does an AI-ready customer data strategy require?

This does not mean every organisation needs a massive transformation programme or another expensive technology platform. But it does mean businesses should examine six fundamental areas before scaling customer-facing AI:

• Identity: Can we reliably recognise the customer across the systems and interactions that matter?
• Context: Do our systems understand enough about what has recently happened to make the next interaction appropriate?
• Consent: Do we know which information may be used, for what purpose and through which channels?
• Governance: What customer information and business rules can an AI system access, and which actions is it allowed to take?
• Orchestration: What happens when marketing, sales, service and operations all want to take different actions with the same customer?
• Measurement: Can we determine whether the automated decision actually improved the customer experience or commercial outcome?

Only some of these are technology questions. The others concern ownership, processes, operating models and decision-making. That is why AI transformation cannot sit solely within IT.

Is this just an argument for buying a CDP?

It would be easy to interpret all of this as an argument for buying a Customer Data Platform. It isn't. CDPs can play an important role. BBVA and Vodafone, for example, have described using customer-data platforms to connect real-time customer signals with analytics, personalisation and AI-driven use cases.^[9][10]

But the strategic lesson is not: buy a CDP. It is: work out what customer context your business needs to make better decisions, then determine the people, processes, data and technology required to make that context available reliably. The architecture should follow the customer and business problem, not the other way around.

What should businesses fix before implementing customer-facing AI?

Instead of beginning with "Where can we put AI?", leadership teams may get more value from asking: "Where would better decisions materially improve customer experience or revenue — and what information would an AI system need to make those decisions well?"

A hotel may discover that its immediate opportunity is not an AI concierge. It may be recognising returning guests consistently across direct bookings and OTAs. A wellness resort may not need a recommendation engine first. It may need to connect booking history, spa usage and guest preferences. A travel operator may discover that the biggest customer-experience problem is not answering enquiries faster. It is customers having to repeat information when they move between digital support and a human agent.

Those are experience problems first. Technology comes later. Before automating the journey, understand it.

AI will undoubtedly change how businesses acquire, serve and retain customers. But adding intelligence to disconnected systems does not automatically create a connected customer experience. Before asking what an AI agent could automate, organisations need to understand the journey it is entering. Because AI isn't making customer-data strategy less important. It is making weak customer context much harder to hide. And as more customer decisions become automated, the businesses creating better experiences may not simply be those with the most powerful AI. They may be the ones giving it the clearest understanding of the customer in the first place.