Where Should AI Stop in the Hospitality Guest Journey?

TRANSCRIPT: Where Should AI Stop in the Hospitality Guest Journey?

The Michelle Pascoe Hospitality Podcast

---

Michelle Pascoe: James McIntyre is an Australian-based customer experience, loyalty and growth strategist. He specialises in customer journey mapping, lifecycle marketing, retention, loyalty strategy, CX design and AI-powered personalisation across retail, e-commerce, financial services and membership businesses. For many of my listeners in the hospitality industry, loyalty and membership are key areas, because they allow us to target and connect with our customers. James has held senior consulting and leadership roles across some of Australia's largest loyalty and customer brands, including Wesfarmers, OnePass, Flybuys, Medibank and Cash Rewards. He is the creator of the ACTIVATE Framework, which we're going to hear about later, an eight-step methodology for AI-enabled CX transformation. Welcome, James. Thank you so much for joining me today.

James McIntyre: Thank you, Michelle, that was a wonderful introduction, so I'm very grateful for that, and also to join you on your podcast. It was a great chat offline where we shared our mutual passion for customer experience, so I'm really looking forward to the chat today. We're going to talk about journey mapping, we're going to talk about tech, and of course we're going to talk about AI. I'm really excited to get into it, and thanks again.

Michelle Pascoe: Perfect, excellent. Well, let's start with the first question. You've worked across, as you said, OnePass, Flybuys, Medibank and Cash Rewards. Where does AI genuinely earn its place in the guest journey, because that's what's really important, isn't it? And where does it need to stay out of the way? How do hospitality leaders draw that line between wait times, upsells and routine requests, versus service recovery and special occasions?

James McIntyre: It's a great question, Michelle, and I think it's something a lot of people are grappling with at the moment. If you think about things broadly, a great customer experience is a great customer experience, and it comes down to three things. Is it personalised, is it frictionless, and is it rewarding?

What's probably changed is that customers have new expectations around customer experience. Looking through the Genesys 2026 State of CX report, they talked to about 10,000 people. It's just come out and it's online if you want to have a look. They found that 92% of consumers expect every organisation to match the best experience they've ever had. So regardless of what industry you're in, the expectation is high. And 76% of consumers expect AI to improve service quality. People know AI is here, people are using it, and they have that expectation that things will be better.

So the opportunity for CX leaders is to use AI to close that gap, not to widen it. It is a tool. There are a number of tools within our toolkit, and it's one that will actually help you deliver the experience customers expect.

But where do you draw the line? There is a line, so where do you draw it? AI definitely owns that functional, transactional, low-impact layer: wait times, routine requests, standard upsells, anything predictable with a predictable answer. Humans continue to own three things. Service recovery, because in my opinion you don't want an agent doing that. Moments that matter, like birthdays, anniversaries, special things and what guests have told you about. And then anything ambiguous, where you want a human in the loop, and actually a human leading it.

So how do you draw that line? You need to plot that customer experience. You can't decide what's going to be automated and what's not until you understand what the journey is and what the moments are. Most people haven't done that map, and that's the first step I normally work through with clients. That's why I developed the ACTIVATE Framework.

Michelle Pascoe: Yes, and look, I love those three points you made. First, the service recovery. I think we've all been there, and I'm speaking on behalf of all my listeners here. We've all been online when a problem has come up, and we go to support or to contact, and you're not talking to a person. I love how they haven't answered your question and then they say, "Please rate us." How do you think I'm going to rate you? You haven't even helped me. And then there's never any phone number. There are just no people to contact.

What goes from a customer who's disappointed that the product or the service delivery wasn't what it was meant to be, a bit of an annoyance, maybe a seven out of ten, becomes something bigger. As that gap broadens, because that chatbot or whatever it might be isn't a person dealing with them, it just goes down to the bottom, where you get one out of ten or even minus out of ten. So that's definitely where you've got to have that human contact.

When it comes to birthdays and those sorts of things, sure, AI can get that online because we ask people those things. But I think it's the nuances. A server is walking by, James, and he hears that it's a wedding anniversary, or it's Bob's birthday and Bob is a little bit shy about it, or Jane's just got her driver's licence, and they haven't put that on their booking, or they haven't made a booking at all. Those are the nuances where all of a sudden a waitperson can make a decision that makes the difference. 

James McIntyre: That's the moment of magic. That's what you remember, and you remember it for years. And guess what? You tell others. You don't tell others via a survey, you tell others in real life.

Michelle Pascoe: That's right, yeah. I think that's what makes the difference. I know so many people think AI is going to take over. Well, it's not. I actually think we're even past that. People in business realise AI is there to help, and it's a fantastic resource, but we will always need that human interaction, particularly when it comes to customer experience.

So when we look at an AI chatbot, for instance, or a booking agent, it promises something. A guest expects a room upgrade, a dietary accommodation, a fast turnaround perhaps, and the team can't deliver it even though it's been promised. Who actually owns that broken promise? Is it a tech problem, a training problem or a leadership problem? What are your thoughts?

James McIntyre: I think we're going to see more and more of that, and increasingly the role of a customer leader is to help manage what I call promise governance. If you think about a human in your team and what promises they're actually making to customers, then think about what an agent is making. What promises are they making, and why? Having those guardrails is really important.

So it's a leadership and a strategic question first. Where are the guardrails? What promises are going to be made about your brand and your service? Then it's training. Think about training for agents as well as people, and ongoing training. For an agentic agent it's not a one-and-done. You need to train it and make sure there's no drift on an ongoing basis. And then tech is last. It really starts from that strategic and leadership perspective.

Michelle Pascoe: It's interesting when you talk about keeping the AI up to date, because it's very true. As humans, we keep up to date with what's happening because we're living that experience. But I think sometimes people just set and forget. It's a bit like when we all first got websites. It was, "Just set and forget," a bit like the on-hold music promoting raffles that are six months old. Let's just put it up there and forget about it.

But you're right. We have to keep updating the AI with what's happened, because things move so rapidly that it may not pick up on them, or you can't expect it to. You've got to keep feeding it that real-time data and information, don't you? Even if your guest demographic starts to change, or your offering has changed, you really do have to feed it that information, otherwise it's going to be out of date.

James McIntyre: Well, that's right. Ultimately it pulls on data, and it pulls on data that you have provided, and depending on the instruction, it pulls on wider data as well. So getting your data right is really important, particularly if, as you were saying, you've got duplicate pages, instructions, manuals and so on. Like a human being, it can get confused between version 10 and version 21. So there's a big job to be done in having the right data for it to seek knowledge from.

Michelle Pascoe: So it's definitely not just set and forget. Look, there's a real risk right now, isn't there, of AI tools over-informing or over-promising just because they can. Where's the line between AI giving guests the right information and AI giving them too much, perhaps offers, options or assurances the business never intended to make?

James McIntyre: Absolutely, and there are a couple of really notable and recent examples. There's a real risk without clear guardrails, a clear strategy and all the correct data we were just talking about.

A very unfortunate one: Air Canada was recently sued after an agentic agent provided incorrect information on their bereavement fare policy to someone. That was a real moment that mattered to that person, and the incorrect information was surfaced when it never should have been provided via an agent in that instance.

Another example is the Chipotle restaurant chain. They had a chatbot called Pepper, and it was meant to be for the menu, ordering and customer service. Some developers, and I don't know how they figured this out, started asking it questions about complicated Python coding. Obviously that type of thing chews through a lot of tokens, and it's expensive. They figured out they could ask it all these questions, then told people within their community, and it went viral. So having the right guardrails is just so important, so you can avoid things like that.

Michelle Pascoe: So poor Pepper went really off cue. It didn't know the 56 items on the menu, but it knew everything else.

James McIntyre: Absolutely, exactly. It knew all sorts of things.

Michelle Pascoe: You really wonder sometimes. I've even done this with my daughter. When we first got AI, I put all my brand into it, so it knew who I am and my company. Then I said to my daughter, who owns a psychology practice, "Oh, you can share it, just put in what you want to know." And all of a sudden, the next time I went on it, I'm now a psychologist of 32, and I'm thinking, no, this is not right. Obviously somebody had gone in and it had just mixed it up. It has to be treated as that one purpose for your business when it comes down to it, doesn't it? Do you think?

James McIntyre: Yeah, absolutely. And I think it's really important that your website and your content are now suitable for AI agents to crawl, and that you're instructing it in broader terms about who you are, what you do and the problems you solve. It's less about optimising pages for particular keywords and more about a wider phrase as well.

You can test it yourself and see what comes up when you search within an agent. Increasingly, that's the first point of call people are going to go to for information. So if you're running a hospitality venue, you need to make sure you're showing up in the correct way, and that your website is set up for agents to crawl your content.

Michelle Pascoe: Yes. It sounds a bit creepy, but that's what they need to do. So it's just making sure of that, and if you don't know about it, get somebody who does to help you.

James McIntyre: That's right.

Michelle Pascoe: So for venues running loyalty or membership programs, which many of my clients do, it's more than the card where we all started off many years ago, have your seventh coffee and you get it free, that sort of thing. It creates loyalty not only from the customer to the business, but from the business to the customer. So is a guest experience with an AI-driven touchpoint, like a chatbot booking or an automated review reply, fundamentally different from a human one? Does it build trust the same way, or does it erode it? Because that's key, isn't it?

James McIntyre: That's a great question. If we look at a really human-centric answer, trust is built up with your brand over a number of interactions, over a period of time, and it could be over a long time. I don't know that people always know whether they're dealing with a human or with AI or an agent, and in some cases perhaps they don't mind.

But trust can be broken when any of those touchpoints fail, and that can also be when they're not remembered. Particularly if you're on a loyalty program and you're providing information, content, data and transactions, you really expect a brand to know you. It's this idea of carrying context through the interactions. A great example is if you say you're celebrating a special occasion at any touchpoint, then when you arrive in person you would expect people to know about it. If that doesn't happen, whether it's AI or a person, that's when trust is broken. So really, it's the sum of all parts, in my opinion.

Michelle Pascoe: Yes, and that loyalty is so key. As I said, we've always expected loyalty from our customers, but it's the reverse of that too. We want our customers to return to us, and we ask them all these questions, so we really should be checking that information before the person comes through the door.

And it doesn't just have to be the VIPs, your top customers who spend the most money with you or stay most often. To me, everybody's important on that loyalty rung. We have a little bit of a joke about the spoon stealers on the lowest rung of the loyalty program, tongue in cheek, but they are just as important.

They may have loyalty with the program, where AI can interact and tell them what's happening, what's coming up and how many points they've got to use. But it still comes down to when they walk in that door. As you rightly said, James, it's that the person remembers who they are, uses their name and reminds them to use those loyalty points. I think that's the biggest thing. We expect people to just remember these loyalty programs, but we all know our wallets and purses are full of cards and our phones are full of apps for different loyalty programs. You think, what have I boosted or not boosted this week, and what am I up to? To me, that's the key point where the human comes in and says, "Oh, did you know you've got this amount of points?" What are your thoughts about that?

James McIntyre: That's a great point. I think the average Australian is a member of between eight and eleven loyalty programs, which is a lot. As human beings, we've got more things we need to remember in our lives, and we're getting more messages than ever. So how are you going to stand out?

In terms of loyalty, where it really comes to life is when the loyalty benefits are redeemed. That's the massive moment that matters. That's the aha moment when members see the benefit of being a member. So getting people to that point and reminding them, as you say, is super important. The likes of Coles and Woolies do that really well at the point of sale, when they say, "Hey James, you've got a ten-dollar reward to use. Would you like to use it now?" And you think, fantastic, I'll do that. Other programs need to think about how they're going to do that, because once someone redeems, they feel great, the process starts again, and they tell others.

Michelle Pascoe: Yes, and I do like that about Flybuys with Coles. I've been a member of that since day dot. I've actually still got it in my maiden name, that's how old my card is, and I haven't bothered changing it. I've been married a long, long time. But it's more the fact of, "Oh, I forgot about that ten dollars." Or with Woolworths, "Oh, you've got another thousand points," which I've transferred over to Qantas. It's those little reminders. Not expecting the customer to remember, but that little prompt: "Oh yes, that's why I've shopped here. That's why I've waved my card there." We need those little reminders for people to remember why they are a member, rather than the next time they get the SMS or the email, they just unsubscribe. They've got to know what that benefit is off the cuff.

What are your thoughts about being loyal to your long-term customers? This has been a real bugbear of mine for many, many years, and it constantly comes up. I've been a user of the same telco in Australia for the 33 years I've had my business, and I get nil loyalty whatsoever, not a brass razoo. But hey, James, if you go from your telco to my telco, you'll get a special deal. Where do you think that loyalty lies, or doesn't it?

James McIntyre: It's a real friction point, and it can come down to silos in an organisation. If you're not thinking holistically about the customer end to end, you might have a department that's focused on acquisition at any cost, and a growth and retention team doing something different. So some of the work I do is to lift up the thinking around CX and look at members end to end, making sure you're talking to them on an ongoing basis.

It can be a real friction point to see a great offer in market that's only for new customers. I think we all know it's more expensive to acquire a new member than it is to retain one. We often forget about that, and with short-termism we look at just putting people in at the top of the funnel. Sometimes we forget there are a lot of existing members who have perhaps become disengaged, but we still know things about them. We still have their contact details. At one point they liked us enough to transact with us and join a loyalty program.

So some of the work I do is making that visible to brands and coming up with strategies for those people who have become disengaged, and how you actually put them back into the funnel, as opposed to spending money on acquiring new customers. Obviously you need to do both, but it's always a point of friction.

Michelle Pascoe: It is, but it is so interesting. When you go into shops, I like this one: "Oh, we can't print the receipt, but if you give us your email." And you think, here we go, now I'm on the trail of getting 1001 marketing emails. But when Covid hit, my gosh, I think people must have sat in their home offices and delved through their last 40 years of any customer who had given them an email or a phone number. I don't know about you, James, but I got hit constantly with, "Welcome, you're a customer of so-and-so." I'm thinking, I can't even remember the product I bought from you, let alone when. You've not contacted me for years, and all of a sudden you're showing me this loyalty? You're not showing me any loyalty, you haven't contacted me. So how do you keep these loyalty programs fresh?

James McIntyre: Well, I did think of something in regards to the problem you just mentioned, which is actually a really good personal use case for AI. If you're using Claude Cowork, where you're basically instructing Claude to do things on your behalf, you can get Claude to go through your Gmail or your email, look at all the emails you haven't opened and unsubscribe from them, which I did recently. So that's a great use case.

But how do you reactivate customers? It's not easy, particularly if they haven't interacted with your brand for a long time. People often feel they don't want to remove someone from their email database, because it makes the database smaller overall, and who knows, maybe they'll buy one day. But there's a certain amount of cleansing you can do through your CRM system to see who is actually interested.

Then it's looking at different strategies to try and reactivate them based on what you know about them. Did they come in at a particular time of year? Do they come in because it's the footy finals? How are you going to be relevant? Do you know what they ordered last time? Then think, okay, let's look at six different experiments we're going to run with different cohorts. That's probably the way to approach it, and it's different for every brand. Come up with a couple of different hypotheses around what would help reactivate Michelle, and what we're going to do for James, then test those with small cohorts and see what works.

I often find it's not what you think, so be prepared to change your mind. Again, it can be a good use case for AI, to come up with those concepts, give you more data to run those experiments, and analyse them as well. So test and learn, I would say.

Michelle Pascoe: I really like that point. As you said, find out a little bit about that customer, so when you do reach out you can remind them, "Oh, when you bought that, it went with widget X or Z." Give me something, not just "buy, buy, buy from me now" when I haven't heard from you. Why should I give my loyalty to you when you haven't shown me any loyalty whatsoever, apart from now you've got a new sale? So it's interesting, as you said, to look a little bit further. You can do that with AI, whereas in the past it was just all too much, because you'd have to have Flossie in the back room going through thousands and thousands of names. With AI you can do it really quickly.

James McIntyre: And one thing you can test is new channels as well. There might be someone who's not opening your emails because they don't like email. For younger cohorts, email is less prevalent, and research I've done recently shows, and I didn't expect this, that some cohorts like to see information and offers from brands through their social media. I wouldn't have expected that.

Michelle Pascoe: Right, yeah.

James McIntyre: That's their discovery platform, so perhaps that's the way to communicate with them. But again, it's that experimentation and knowing your customers.

Michelle Pascoe: Yes, and you've got to be able to do that, but AI certainly makes it a lot easier. Let's talk about ACTIVATE, the framework you've created, James, for AI-enabled CX transformation. What's the one human skill in hospitality that the framework assumes will always stay human, and why does it become more valuable, not less, as AI takes over that transactional layer?

James McIntyre: What I love about hospitality is that we talk a lot about marketing and customer experience, and a lot about personalisation and how we can be more personalised and one-to-one. When you think about a great experience overall, and the experience you get in hospitality, it's really the ultimate in one-to-one interaction, where a person is interacting with another person. That doesn't go away.

Thinking about the ACTIVATE Framework, I developed it to think about the customer journey and to understand where the touchpoints are, where it's relevant to integrate AI and where it's not. It's different for every customer and every category, but it's an eight-step methodology for AI-enabled CX transformation, and it's one I use with brands to figure out where AI comes to life.

Something like 92% of AI pilots fail. What I'm seeing across organisations is a lot of pilots happening across lots of different teams and departments, popping up left, right and centre. So what's a way we can think strategically and coordinate where AI is going to come to life for a customer?

There are eight steps, and we'll go through them really quickly. It's on my website as well, which I'll share with you, Michelle, and you can share with your listeners if they're interested.

The first is to audit the real journey, so that's the A. Map what customers actually experience, including AI-influenced discovery and automated loops, the things you know about and the things you don't. You might not know about some of the ways your customers are using AI to interact with your brand. That's the first thing you need to do.

Then you want to clarify the moments of disproportionate importance. What are the touchpoints that influence trust, churn and retention decisions? We talked about some of them, like special moments.

Next, how are you going to triage those AI opportunities? Avoid the pilot trap of doing 101 different pilots, and look at some commercial criteria.

The fourth step is to integrate your data, CX and marketing, so you have a data layer that's integrated. Validate that through commercial metrics. Assign accountability, which is really important: who's going to have accountability across which parts of the journey, and who's going to have AI governance and oversight. Then test in priority journeys, so how are you going to set up some small tests and test and learn.

And then, of course, embed the human overlay. Where are humans coming into this? Automation removes friction, but it shouldn't remove empathy. Technology should handle the efficiency, and humans handle the complexity.

If your audience is interested in the framework, there's also a free diagnostic, which takes about three minutes, where you can go through and audit your business against it.

Michelle Pascoe: Perfect. So what's the website address for that, please, James?

James McIntyre: It's jm-cx.co/activate-framework, from memory, but I'll send it through to you, Michelle, so you can share it.

Michelle Pascoe: We'll put the link in our show notes so everybody can get to it. Thank you very much for that, because I think it's important. When you've got a methodology like that, people can follow it and understand all those very specific components, James, and they can work through it with you, which is excellent. And to have that diagnostic as well. So yes, listeners, please go on to James's website and do that. As he said, it only takes three minutes, and it will give you some real insights about the CX journey and that connection with AI.

Before we wrap up, James, if you were advising a club CEO or a hotel GM starting from scratch today, what's the first guest-facing moment you'd hand to AI, and the first one you'd protect fiercely for a human? And what does that decision actually say about the business's values?

James McIntyre: What I would say is that it's so important to start with the customer. That's your first step. Understand the customer and the customer journey. How are they interacting with your brand? Talk to your customers. Understand how they're using technology and where they're using AI.

I think the first step would be: how are you showing up when people use an LLM to look at your product and service? What sort of questions are they asking, what are they getting back, and is it correct? That's probably the first thing I would look at.

The second step is that I'd probably do an audit of the technology and tools you're using at the moment. We've talked a little about CRM and sending out communications and emails. There are some great enhancements. Pretty much all tools have put on an AI layer that enables you to be more efficient and more personalised in what you're doing, whether that's a subject line or the best time of day to send a communication. So if I was a CEO or GM, I'd be saying to the team, "Okay, you've got AI now in tools we subscribe to and pay for. How are you going to leverage that?"

But definitely start with the customer. I'd start with the customer journey and then go back to the tech from there.

Michelle Pascoe: Perfect, thank you so much, James. Your conversation with me today has certainly enlightened me and got me thinking about a few areas, even in my own business, where I can utilise AI more, but certainly without reducing the amount of face-to-face time and connection we have with our wonderful clients. Is there anything I haven't asked you about this topic that you would like to share?

James McIntyre: It was a great conversation, Michelle. Thank you so much for having me on your podcast. This is an evolving space. If we were to talk in three or six months' time, which I hope we have the opportunity to do, it could be a very different conversation. But what won't be different is the emphasis we both have on the customer and the customer experience. The tech, though, is certainly changing very quickly.

Michelle Pascoe: It certainly is, it's very rapid. Thank you so much for joining me, James, I really appreciate your time. And listeners, please don't forget to have a look at James's website and his diagnostic, as well as catch James on LinkedIn, because, audience, that is my favourite platform to connect businesses with businesses. Take care, James.

James McIntyre: Thanks, Michelle.

---

END TRANSCRIPT