What 2025 Taught Us About AI, Composable Tech, and Marketing Control

Joined by colleagues James Gent and Neil Hughes, the conversation draws on real client work, delivery experience, and the patterns emerging across enterprise teams as they plan for 2026.

Rather than chasing the latest trend, the episode focuses on a few realities that kept coming up throughout the year:

• The growing pressure on marketing teams to prove value, not just capability

• Why simplifying and consolidating MarTech stacks has become unavoidable

• How AI is adding value today and where expectations still outpace reality

• Why decisioning is becoming as important as any new technology

The discussion also explores how AI is shifting the problem space. Moving away from the idea of perfect datasets or “golden records”, teams are now grappling with how to provide the right data, in the right context, at the right time, specifically to enable agentic orchestration layers to reach what they need.

This episode is an honest look at the trade-offs, foundations, and ways of working that teams are navigating as they move into 2026.

If you’re thinking about how to build momentum without adding unnecessary complexity, it offers a grounded perspective from the front line.

Listen now on your preferred platform:

Apple Podcasts
Spotify
YouTube

Intro – 0:21: Welcome to The CX Equation, a podcast by Tap CXM. We share actionable insights and real world case studies to equip you with the tools you need to drive loyalty, engagement, and sustainable growth. We’re your hosts, Chantelle Casey and Mark Clydesdale.

Mark – 0:35: Welcome to The CX Equation. Today, we’re joined by James Gent and Neil Hughes, both of whom work with us at Tap CXM and bring a wealth of experience working with leading brands to design and deliver marketing technology solutions that drive measurable value. Today, we’re going to get their insights on 2025 and look ahead at what trends, technologies, and lessons are likely to shape this year ahead. So, Neil, James, hello, and welcome.

Chantelle – 1:03: Just to kick us off then, let’s take a bit of a rewind back having a look over the past year. What are the things that you’ve noticed most in the way that marketing teams worked, which have changed over the last year?

Neil – 1:14: For me, personally, I think there’s been a real focus on value. So a bit more grown up approach to proving ROI for spend on technology and on resources. That’s definitely a pattern I’ve spotted when talking with clients. They’re looking to either consolidate their technology spend where it grows a little bit organically, either through their own M&A purchases or just directly from kit that’s been bought by different departments, different teams, where there may be some kind of overlap. So I think this year for me has been one of checks and balances on how companies spend, and I think that chimes a little bit with the consumer market at the moment, personally.

Chantelle – 2:13: I think that comes in with people reviewing operational processes as well and seeing where they can save money in those areas. With new assessing of their MarTech, that comes with assessing operational stuff as well. So I think we’ve definitely seen a trend in that this year.

James – 2:27: Definitely. I think we’ve been working on quite a few projects that are consolidation of tech. People want to simplify what they’re using, rationalise what they’re using as well. I think it’s trying to get, certainly larger businesses, people doing things in a way that’s common, standardised. I’m sure we’ll get on to it in a second, but if you’re looking to layer in things like AI technology then having a centralised platform, data in one place, all of those good things are needed to be able to leverage the latest technology that’s being released. I think a lot of businesses are still trying to work out what’s the best way for them to set up themselves from a technology perspective and people and process around that as well.

Chantelle – 3:15: I think in re-evaluating or reassessing their MarTech stack comes with cutting out a lot of the noise. In a lot of these legacy platforms our clients work with, there’s a lot of unnecessary redundant processes that aren’t actually giving them much of a return.

Mark – 3:33. I’d say the last two or three years now, we’ve been seeing big international global businesses certainly centralising technology more, trying to have a standardised set of licenses and technologies that they know that each of their regions or brands can use that are a leading bit of enterprise technology that provides the capability that they need, trying to move to common data structures. And it’s all in the drive for better knowledge sharing, better cost efficiency, better kind of quality control. I think the bit that’s interesting to then get the balance is if as a global team, global function, global business, you’re trying to centralise the data and the technology, how much freedom is still allowing the regions to work within that? So because people in your marketing teams in your regions will know the customers and the trends in that region better than any centralised team. So I think it’s important that even if we’re trying to save cost and make it easier to use technology, we’ve still got to then empower teams using it in the local places to actually deliver the best customer experience.

James – 4:47: Yeah. And I think a lot of businesses we’ve been working with are going through reorganisations, and people are having to do a lot more with a lot less as well. So being able to rationalise their processes and really look inwards and see how they can better use what they’ve got to get at least the same output or more. I think that’s where it’s gonna be interesting to see how people can maximise that tech they’ve now invested in and really generate that ROI that has been  promised so far. Because I think a lot of the time, business cases are put in place, but to actually get to that golden nugget of understanding what the business value is is the hardest bit. And next year is probably gonna be a year of reckoning to see ‘are they gonna get that value that has been promised this year through the investment that’s been made throughout 2025’?

Neil – 5:45: That could be more interesting when people are looking at best of breed and composable stack as well. I’ve seen people that have bought certain large clients that have bought all of the kits and the tools and plugged them all together, but now they’re having to make some decisions on, right, well, which ones are these are core or the spine of our stack and which ones may we have to actually lose to just save a bit of license costs across the board. And therefore then looking at some of these tools to pick up the slack or some of the teams to then pick up the slack. If a kind of function isn’t there before, then maybe it goes into a design department or something like that.

Mark – 6:23: For the non technologists in the room, what do we mean by a composable stack?

Neil – 6:29: I think it’s a bit of a false premise in some regards. So over the years from a sort of sales point of view, some of the technology vendors have come together under kind of banners like the MACH Alliance, and where they aretechnology that should be simple and easy to plug into each other, doesn’t tie you into one vendor and doesn’t tie you into a lot of features and functionality that you don’t necessarily need. So the idea being that you buy the best in breed for a particular purpose and plug it into more of that. And then if you do need a mobile measurement platform, then you buy the best one there, plug it into your stack. That’s great. If you don’t need it, you don’t buy it and you don’t pay for it. But the point that they don’t necessarily like to make a big deal of is that most of the hubs have come from this kind of approach anyway. They just ended up buying the best of breed tool or a similar type of tool and integrating it into their stack. Unfortunately, that integration is not always brilliant and isn’t always the best tool. So, really, if you’re an enterprise business, you’ve got to weigh up that kind of trade off, basically. But it can introduce more vendors and more admin overhead in that regard.

Chantelle – 7:49: I guess with that is the trade off that the complexity has shifted to internal systems or the integration layer between all of those systems and, obviously, added complexity of handling multiple platforms and the governance around that, that’s a trade off there.

Neil – 8:05: Yeah. There’s definitely a feeling that you just use a key and plug it in, and off you go. That’s every sales guy’s pitch, but we know from the hard yards that it’s not always as easy as that.

Mark – 8:28: So we don’t wanna get into specific vendors, but when you say the hubs, we’re talking about the big Adobes and Salesforces of this world. For me, a few years ago, every client I worked with would be on Adobe or Salesforce, really. Certainly any large company. Whereas it does feel like there is more competition now.

Chantelle – 8:57: I’ve seen more of a trend in use case driven sort of tech investment or implementation rather than just going with the Big Bang Theory. People are wanting to prove the value of a platform before jumping the gun and moving over and I think that’s easier to do with these sort of more specific platforms rather than the big players.

James – 9:52: I think you’re seeing other platforms grow alongside Adobe and Salesforce, so the likes of Braze are almost a billion, and they’re recording revenue now 25% year on year growth. I think the interesting part is, previously, people were doing a lot more in the tools, like with Adobe Campaign or Salesforce Marketing Cloud. There’s been a lot more investment earlier on in the data chain, let’s call it, so people are spending more on Snowflake, Databricks and then do more of the data work upfront, which means that you don’t need such a heavyweight marketing automation tool later on. And you’ve got more of a flat customer profile that’s bringing in events being attached to that customer profile, which is a different way. More people were working in the past with tools like Adobe Campaign and Salesforce Marketing Cloud. So we’ve seen that evolution of the stack. Certainly, there’s been the growth of those types of tools. Adobe have Adobe Journey Optimizer, which is a bit more similar. I think it’s the marketing automation tool capability. I think what’s gonna be interesting there is journey tools, marketing automation tools, essentially, their orchestration engine. There’s gonna be more overlap with this central agent orchestration layer that’s gonna live in the stack now going forward. So I think the architecture is evolving right now to have the agent capability built in, and it’s not yet clear how much of that is gonna replace what’s being done in the marketing automation tools versus what will be done within that new agent orchestration layer.

Chantelle – 11:19: That kind of goes hand in hand with the trend we’ve seen with more clients and companies wanting their marketing teams to have more autonomy, and the vendors have gone in line with that as well. So a lot more marketing friendly platforms where they don’t need to have the level of technical knowledge that previously they would have had to have on the likes of Adobe and Salesforce. So I think we’ve seen that coming through in the products as well, and also in the requirements from our clients when they’re looking at new software.

Neil – 11:43: Clients are expected to be technical in certain other ways. Like, back in the day, I don’t know many kind of customer marketing API was or an SDK or these kinds of things, but some of these newer Challenger vendors have brought some of this terminology in, and the lay marketer knows that a little bit more now, less about normalising your database structures and your tables within the actual product. And I think James hit on a really good point about the fact that Adobe and Salesforce have been through this maturity curve, and they just reached a point where they plateaued slightly in a sequel based, all-data-in-the-platform approach to delivery. And both of those were the best of breed at doing that. There was just a bit of an architectural shift, which was more sort of microservices, web, smaller packets of information and data being passed around as events. So what they’ve done within the onboarding journeys, they’ve said to clients “if you want to take advantage of the capability and you want it to be super easy for marketers, you reduce down your data model. You make it very simple with a customer profile and with events, and, therefore, it is simpler, and you can make the UI then cleaner, easier”, and that, for a marketer, is great. Where it becomes problematic is once the vendor goes off the pitch after onboarding and you’ve got a data engineering function that’s moved elsewhere and is focusing on other things, that the next bit of data or the thing that you forgot to ask for in onboarding, you want to get into the platform. And that’s where you can reach a bit of a plateau as with some of these modern tools. But you’re right, the kind of AI element to it is, like, where is that gonna live? Who’s going to own that within the business? And what does your infrastructure need to look like in order to make that happen? It’s gonna be a key thing for 2026.

Chantelle – 13:44: AI is the big elephant in the room at the moment, isn’t it? So just on that, is it all hype, or is there actual practicality? Are we seeing actual practicality to agentic AI and gen AI or whatever is in the MarTech landscape? What are your thoughts on that?

James – 13:57: Yeah. I think it’s interesting a few things that come out over the last month or so in our industry. So we’ve got the MarTech report that’s produced by Scott Brinker. They were talking about AgentForce, I think, 18,500 deals closed in 2025 for people buying AgentForce. And their annual recurring revenue has now gone to $500 million just for AgentForce, that other revenue on top of that, which is AgentForce capabilities on top of other Salesforce stack. We can’t really quite see any Adobe numbers yet. They’ve got this new metric in there, which is AI influence, which is showing that they’re building in AI to the products everywhere, and they’re showing that that’s now attached to I think it’s about a third of their revenue, but they’re looking to try and get that to 100% of their revenue as everything is attached to AI. So I think what that shows is people are buying the tech. They’re starting to use it, but I’m not convinced yet we’re at the point that we’ve been able to demonstrate the value that’s been promised from the capability. I think it’s also interesting reading things like Reddits, subreddits,forums to show what people on the ground who are using the technology actually are thinking about. And I think with AgentForce and other capabilities, everyone was saying, what’s being demoed is just the very top of the iceberg. It looks very great and very shiny, but, actually, what’s it underneath that, if that’s in a real world production organisation, it’s not actually realistic to be able to produce those use cases yet. So there’s a lot of infrastructure and underlying work that needs to be done to be able to get to that value. And I think that’s borne out by, say, Salesforce buying Informatica during the year as well, where they’re heavily investing in data infrastructure to make sure that people have got that capability to be able to get the data in the right place in a clean way. I see this is a foundational year where people are putting in place the capability to try and get the value. I’m hoping in 2026 that we’ll start to actually see some of that value being borne out through some more real life case studies because I think there’s a lack of real life case studies at the moment. A lot of it is sales demos still built on synthetic data and not real data at the moment.

Neil – 16:05: Definitely. I’ve heard instances of 2026 where large enterprise clients want to test having a CRM agent team versus their own in-house team. It’s slightly scary that we’re gonna be competing directly against the machines in some sort of squid game style, but it is where the market feels like it’s going a little bit. But that, to James’s point, can lead to a bit of a trough of disillusionment if it doesn’t quite meet the needs or it’s too difficult to get off the ground. So, yeah, there’s been a lot of proof of concept, a lot of showing how this could influence, but most of it is from what I’ve seen is going into efficiency gains in production. And like you said earlier, being able to do more with less, that seems to be the place where people are focusing most and getting the most success at the moment. But I think a lot of that is opening up the infrastructure so that the agents can operate in an autonomous manner. So some of the seeding ground to some Challenger vendors in certain areas, that could be because Salesforce might have their eye more on the endgame of AI agents itself as they see it. And maybe they don’t mind that some of their share kind of eaten away in that area.

James – 17:25: I think it’s definitely the internal content production or internal production of marketing assets and being where the gain has been so that’s shown so far. I don’t know if you guys saw there was the AI Turing test that Lloyds Bank and Ogilvy did recently where it pitted against three different teams together. So there was human only, AI only, and then human and AI. And then they were all given a brief to then go away and respond to the campaign and then produce the assets and produce the campaign. And it showed that, clearly, the humans plus AI produced the better output across a few different categories. I think one was strategy, creativity, and forget the other ones, but the AI only underperformed against the human only from the overall output. It produced way more content, but it just didn’t perform to the same level for the end output, essentially, just producing a lot of slop, which we’re seeing at the moment. And you have to have the human plus AI part of it to be able to get the value, which I think is great to see that. So I think over the last year or so, there’s been a lot of noise about cutting out teams or resources to replicate it with AI. But I think we all know now at this point, we’ve all played with it long enough to know that there’s just really not a bad point where we can do that.

Neil – 18:52: The concern there is some CEOs have maybe jumped the gun in shedding staff or they haven’t quite got the learning or the models internally fully trained from the experts before they’re saying, maybe we don’t need you anymore. So I think there might be a little bit of a rollback on that in some places.

Mark – 19:12: Well, I did see a cartoon on Instagram that was, who are we? CEOs. What do we want? AI. What do we want it to do? We don’t know. When do we want it? Now. I do think what frustrates me is AI is this it’s a huge area. There’s an obfuscation being taken advantage of by companies. So back to new tech vendors, everything now is powered by AI, and everything says powered by AI. But what does it mean? And unless you really understand what you’re getting, it’s almost just being sucked into this this new thing. We have clients ask us, what’s an AI strategy? What should our AI strategy be? How should we use AI? And it’s kind of a meaningless question. Using AI isn’t an outcome. I would like to see the trend move back towards the use cases and the things that we’re trying to do. So as you’ve just said, James, the idea of an agent being able to effectively, what we’re saying is automate personalised testing faster. Right? I can’t get humans to come up with all of these different test ideas and put them into practice and measure the result in a really kinda quick real time way, figure out the best one, and then kind of run with it. Agents can very much help us, but it’s not like ‘great, I’m using agents’; it’s ‘great, I’m doing more automated personalisation testing’. I worked with a telco last year that was using the power of large language models to enter their brand guidelines and their legal and compliance guidelines, and then that to read the marketing that marketers were producing and checking that the content met the brand guidelines and met the legal guidelines. And the AI could come back and say, I’m not clear whether you’ve applied the promotional discount in the price that you’ve given me. Please can you be clearer about that? That’s a process automation benefit, right, where I can now very quickly just make sure that there’s a quality assurance to all of my marketing. So I would like to see us moving more towards the what am I trying to do? What am I trying to do more quickly, more powerfully? Because it’s not the case of all of a sudden AI is just gonna start writing all of the content that we send straight to a customer without human oversight, as your example’s just been, James. It’s gonna come out with slop, and I absolutely, at this point in time, wouldn’t trust it to write something to a customer that I don’t then see beforehand. But then we’ve had machine learning for a long time. Machine learning is technically part of AI. You will have technology vendors come along now, and there’s a tiny bit of the platform that does a bit of machine learning, and it’s AI powered so that they can jump on the hype bandwagon. And I think we’ve got to be much clearer about what are those use cases that we’re actually doing and why they add value to the business rather than it just be about using AI.

Chantelle – 22:08: And, actually, that makes it so much easier to prove the value of it as well rather than just having that open ended question.

Neil – 22:14: For me, there needs to always well, not always, but mostly for a successful kind of application of AI, there needs to be a problem that you’re trying to solve that you know reasonably well. Some of the best ones I’ve seen with clients that I’ve worked with, like a train company that has cows on the line that can cause problems and can have impacts on their commercial business, on passenger experience and things like that. Stock levels of food and waste as we moving towards carbon neutral society. The idea that you can use AI to solve those kinds of problems is actually helpful, but just having an open ended goal of applying AI is not actually particularly useful. And to your point about LLMs, I did an artificial intelligence elective at university in 1997. This is not new stuff. It’s got a real wave of sales and marketing behind it, and I’ve seen a lot of vendors just rebadging something that they already had, which was some kind of model scoring mechanisms as to an AI suite, and it simply isn’t really.

James – 23:23: Yeah. So healthy so a healthy dose of cynicism, I think, is helpful. But it’s also I think we’re in a place where it’s amazing how quickly you get used to some of this technology as well and been part of quite a few demos this year where people are demoing AI capability. And already, you can see people like, ‘Okay. That looks great.’ But when you think about what we had a few years ago, this is amazing. Thinking about how we might be able to use prompts to generate a whole workflow, and you can do that via MCP server and not have to actually go into a tool, for example. And and it can try and, upfront, understand what’s gonna be the best way to A/B test before you even send something out based on the data it holds and all this stuff, I still find it’s pretty impressive. But it’s amazing how quickly you get used to it. I think in our industry, we kind of do lean towards being quite cynical about this stuff because we’ve we’re used to the sales messages, and we have to develop it. But I think there’s still a degree of magic to it as well. I think it’s gonna be amazing to build more of this stuff next year and see how we can actually build out the clients and get some value from it because I think it’s an amazing opportunity for us to make some massive changes, particularly on the distinction between how you’re using it internally versus how you’re using it for the end customer or how the end customer’s using it. I think how the end customer uses it, obviously, carries greater risk, and there’s brand reputation and trust issues there that you need to make sure you’re not gonna erode any brand trust you built over the years by suddenly allowing some agent capability on the customer that doesn’t do what it’s supposed to do. But, internally, I think there’s a massive opportunity to just improve the workflow and the way people are working with the technology we’ve been working with for a number of years, and I think we’re at that point now where you can literally come up with an idea and produce then the assets and produce the workflow and test things in a matter of hours, whereas before, it used to take weeks and weeks. So I think we’re already at that point where you can produce things at a lightning speed, but it’s having that quality control and making sure you’re not producing crap at the end of it that is actually gonna be the key thing here. And that will be how to make sure there’s that layer of sense check and making sure that what’s being produced actually is gonna deliver the results you want.

Neil – 25:50: And guardrails are important, aren’t they?

Mark – 25:52: I think there’s definitely depending what industry you work in, the use cases for AI change as well. And, James, I like what you’ve just said about the way that the end customer’s using it. So let’s pick on travel and hospitality for a moment. There was a report earlier this year that now it was a relatively small number. It was about 4% of consumers are now asking ChatGPT to plan their holiday for them. But then if you look at the demographic breakdown in younger audiences, it might be more like 10% who are asking ChatGPT to plan their holiday for them. So then as a travel and hospitality company, I guess there’s a few questions there. How can I how can I be part of that? How can I pick up on what consumers are asking for and understand how I can then better serve their needs because I know what they want? How can I get them to come to me to ask me, my company, to plan their holiday for them, which might be through some form of large language model or some integration with ChatGPT and myself or another potential model? But there’s a whole area of …we used to go and speak to people as travel agents, and they would help us plan our holidays. Right? Young people don’t wanna do that. They also don’t want to do the ends and ends of Google searches for it. So how do I provide the user a tool that I, as the travel and hospitality company, trust from a brand reputation point of view that can help them plan their holiday and just overall enhance that experience. So depending what industry you’re in as well, there are lots of different benefits. I think we need to be getting to that specificity as to how is this gonna help my business, and how is it gonna help my consumer.

Neil – 27:31: It’s a whole new area of marketing that has sprung up, hasn’t it? This GEO side of things with generative content. You’re going to have to really define a content approach that’s specific for that engine of optimisation, if you like. It’s not 100% different from what already exists: it’s about providing good quality content, being a trusted source of information and somewhere that people would naturally go to investigate – and then making that content available in the right kind of formats and structures where it can be easily consumed. But you’re going to have less control over that at the start. And there’s going to be a real gold rush of people working out how to do that to the best ability to please the agents and doing that legwork on behalf of the consumer. It’s gonna be an interesting journey, but I think you can probably pick up a lot of good habits from how you’ve been managing your content before, but you may get punished a little bit on old and invalid content or poor quality content being available, which you don’t then necessarily have control over that the agent is going and picking up and utilising. And then that will be frustrating for an end user if it’s not good quality for them.

Mark – 28:54: Sticking to the AI point, and then, again, even maybe with travel and hospitality as an example, one of the most challenging parts of any of this is getting the data in a good enough quality that any testing I’m trying to do, any learnings that I’m trying to teach something, that the data is available and the data is of good quality. I think Adobe released a report maybe last year or early this year that said travel and hospitality companies tend to be broken down into their products or their departments, and the data is disparate depending on what sort of holiday you’ve booked in the past. And because I can’t unify that customer profile, my ability to then use AI as a concierge or something like that is limited. And we’ve always known bad data in, bad result out. Right? So data will continue to be a sticking point. How do you think we’ll see companies trying to react to that and further enhancing what they have in terms of a clean, compact, integrated set of data on customers?

James – 29:57: There’s all these new roles or terms that are coming up. Like, before, we had a prompt engineer, and there’s a go to market engineer. I think the latest one is now a context engineer. But I think the key thing is if you’ve got an agent or you’ve got an orchestration layer that’s got several agents, it’s not just having the data, but it’s also having the right data in the right context. And I think we’re gonna see more of that capability where in the past, it might have been everyone’s trying to get a single customer view or that golden record, and a lot of of companies still haven’t been able to get to that Nirvana because they’ve got so many different silos of data all over the place. And it might just need a different solution to actually bring the right context to that LLM or whatever tool or agent is doing the reasoning based on that context of the data. Then there might just need to be a centralised, let’s call it, context engineering team, but some team or some people who are responsible for making sure that the right data, the right context, and the right knowledge base is available. So I don’t know if you guys have if you built out your own AI agents or built out anything, most of all the tools you can use can build in workflows, and you can do different things. But it all comes back to what is the knowledge base it’s got access to, what is the context, and what’s the quality of that context. And so that’s the key thing in being able to make sure you can bring in that right context at the right time to then make sure the output is relevant and consistent. That’s gonna be the bit to crack that people are trying to do. So I think it is a data problem, but it’s also then a context problem and making sure that it’s got the right context based on whatever data is available to that.

Neil – 31:37: Chatbots used to just rely on your FAQs data repository, basically, and try and essentially automate that. It’s like that on steroids, where it’s really the next level, but having perfect data in every context is an aspiration that everyone has, but very few people are gonna get anywhere near it. So it’s how can you get as close to that as possible with the resources that you have. And I think the world is constantly changing, growing, new things happening. You’re not ever going to be able to get to that perfect dataset. So it’s what’s gonna deliver that 80-20 of what you need for right now and making sure that you have a team that’s supporting the rest of the functions of the business to be able to get a hold of the 20% when they need it.

James – 32:29: And to your point, earlier around some of this stuff’s been available for years anyway, I think what we’re seeing is just generally now that people are moving or looking to move away from a traditional ‘if this, then that’ type logic, and it’s moving from now more of a reasoning logic. We’re seeing that people are now looking more and more at ‘Okay, what is the decisioning, and what should be doing the decisioning? Where should that live, and what technology should we have?’ So I see that as definitely a trend at the moment. It’s something that I think will continue as people that are trying to work out where that should sit. Should we have more than one decisioning engine? Should it sit in different places for different reasons? Should it live centrally, as close as possible to your core data source, or should it live closer to the end where you’re delivering something? Let’s say if you’re producing an email and then you’re having positioning at the point of sending the email rather than at the point of where the data’s sitting centrally, or do you need both? So I think that’s definitely a topic that is more prevalent at the moment.

Neil – 33:29: Well, that’s definitely something that’s been around for a while, but people (Mark, you’ll know this as well as anyone) people of an enterprise scale really struggle to make that happen, and it’s been more on a grand platform transformation scale. You’re changing your team’s ways of working, ways of thinking, and ways about embedding those decisioning approaches into your business. I feel like now what’s happening is the decisioning is becoming more readily available. And actually some vendors are looking at how they can make that smaller and more manageable, easy to digest for a business and provide value quickly on a smaller outlay, and prove the outcome. And actually, if they do that, that gives them the right to then attack more use cases and gradually layer on further functionality. But that’s just more a change in approach than in a hugely different concept.

Mark – 35:11: I’m really glad to hear you mentioned ways of working, Neil, because it is such a big change for people. If I’m a marketer and I’ve been used to the idea of campaign management, right, and I’m thinking about ‘what are the things I wanna say and who do I want to say them to’. That is an entirely different way of working to thinking ‘how do I build up all of these experiences and try to build a system’. And by system, I mean, more than technology. I mean, the technology and the ways of working and the process around all of this that can say, for that customer given that context and those attributes and this recent event, this is the right thing for them right now. And it’s a much less predictable world. You don’t know how many volumes you’re going to send out at what time, and it’s harder to say that I’m gonna get as exactly this much revenue uplift on that product in this time frame. But you will do better as a business overall, and it’s a big, big mindset shift. And so I think a key part to embedding AI and agents and automated testing and all of this into businesses is making sure that the people on the ground have got that way of thinking and that way of working in the business to actually allow it to happen.

Chantelle – 36:22: Yeah. I think just back on the decisioning point, one of the biggest challenges we’re seeing at the moment is that it’s not the actual models and how things how the content is chosen. It’s the gap between the decisioning layer and the activation layer and actually how are they talking to each other and then the content management on top of that. I spoke to a client last week who were like, ‘we’ve been doing decisioning for ten years. Why is it such a buzzword at the moment?’ And I think it is because we see vendors that are trying to do a lot of it in the same within their platform now. So they’re talking about it a lot more, whereas, like you guys have said previously, decisioning was thought of quite separately, I think, outside of the marketing automation platform.

Neil – 37:04: I’m with you there, Chantelle, and I think it’s quite amusing in some parts, because when you say it’s part of their platform, some of the platforms I can see a menu option, but actually it’s fully off-site, ‘send us the data. We have a consultancy team who will build out the models and train it and work with you to make sure that it delivers that result that we’ve promised you will’, but it is not an integrated part of the stack, and it’s not plug and play. It’s something that they’ve purchased and are pushing, and it may work for clients, and that’s great, but it makes them a little bit more like some of the traditional marketing hubs and buying companies or buying features in to eventually integrate into their stack. So it’s just about being careful that you understand exactly how the solution works and whether it is all that different to another option.

Mark – 38:08: So, fast forward a year, we’re summarising 2026. Are we gonna be having the same conversation, or what are we gonna be talking about as everything that’s just happened in that last year?

James – 38:21: I think technology is gonna become sentient and we all have to worry.

Mark – 38:28: This podcast will be hosted by AI.

James – 38:30: Yeah. Exactly. And CDPs won’t want to ingest messy data, we’ll have to hire a CDP therapist or something because they’re they’re now sentient . The rate of change is just incredible, isn’t it? So I think it’s gonna be something that we can’t even comprehend right now in a year’s time just because of the rate of change. But I think this time next year,  I think we’ll have a lot more case studies to talk about, which is in the new architecture where we will have companies that are benefiting from this concierge type service.

Chantelle – 39:03: What about away from AI? Do you think there’s anything that’s gonna be a shift in anything over the next year in any other areas?

Neil – 39:09: I think a continued trend of honesty about what companies can achieve with the resources and the budgets that they have, continuing along that track, I think there will be maybe a simplified org structure in some companies if they can manage it. I think slimming down and making sure that if we are going to use AI, that it’s being used in a way that helps the business streamline rather than causes complexity.

James – 39:45: We’ll reach minority report type levels of predictive analytics that companies will start to know what customers are gonna do before they even know about it themselves, which might lead to some interesting use cases.

Mark – 39:59: I have an alternative take. I have a non AI related one. I think, sadly, we are in a slightly more polarised world nowadays. I don’t think this is true of everyone, but I certainly think in the younger generations, companies matching their personal values is becoming more important to them, and I don’t see that going away. Now I do think we’re at a particular point in history as well that we’re not very good at disagreeing with each other, and we’re a bit tribal about our thoughts. So I think companies will have to be clear what they stand for sometimes, and I do think companies will risk losing customers if they don’t reflect the right sorts of values. I hope that doesn’t turn into no one really saying anything because of fear as to how one person might react. But I do think being clear about what your values are as a company and younger consumers, in particular, choosing who they shop with, who they buy from, whose services they take up based on their personal values and their alignment with that will continue to become bigger. And I’ll be interested to see whether marketing teams how much of a role they feel they need to take in talking to customers about what they stand for, rather than just what they’re selling and the price it is.

Neil – 41:23: Yeah. I can see that. And I can see the artisanal approach. People have kind of worked out how to bake bread to the nth degree and make it sort of last for weeks on end and do it at a high volume, low cost. But you have seen over the years, the sourdough brigade and people willing to pay more for a different kind of service. And so maybe you start to see some of these, like, human curated approaches to things like, when you’re talking about travel earlier, Mark, that, there’s gonna be a split into the people that go down the influencer route versus the people that go through the concierge kind of approach. So maybe there’s something for us humans there.

Mark – 42:12: And do we still care about sustainability? I mean, I guess I wasn’t going to bring it back to AI, but AI is using so much resource and power and just feel that it certainly felt like a couple of years ago, any big enterprise company was investing in departments around sustainability and how they were being more sustainable. Has political climate made us stop caring about sustainability or companies maybe rowing back on some of those sustainable things at the moment?

James – 42:40: I think, actually, your point earlier, on being polarised. It’s how much of it is being politicised as well? Like, how much do people believe? And I think that’s part of the problem is people are now so distrusting of information and what’s true and what’s not. It kind of feels like in the last couple of years, it’s taken less importance than it did a few years back, and maybe that’s because people are not as trusting of the information and doubting it, which is not ideal if it is true.

Mark – 43:08: Authenticity does feel like it’s becoming a keyword. Right? When a brand’s selling me something, how true are they being? And I do think we need to remember to be authentic. You can’t just stick a strap line on an advert and assume that that message is gonna get across. You gotta kinda follow through on some of this stuff now, I think, because otherwise, you will get you will get found out.

Chantelle – 43:29: And I think people are interrogating brands and their content more for authenticity now they know a lot of it is being generated by AI. People will spot it more easily, the more they see AI content, etcetera.

Mark – 43:42: Cool. I think we’re we’ve been talking away and we’re almost out of time. So should we ask the guys our usual closing questions?

Chantelle – 43:50: Yeah. So we finish on the same question every episode, so we’ll still put it to the both of you. Who has been the greatest professional influence on your career and why? And what do you think you learned from them?

Neil – 44:01: Probably one of my old bosses, Alex Naisby, used to always tell me to question everything, “why?”. And curiosity is a real skill, in this industry, generally in life, but I’ve always taken that with me wherever I’ve worked, that ‘why are we doing this?’ Not because I don’t want to do what’s being asked, but often you get into group-think or you continue doing the same sort of pattern that people are used to doing. So just taking a step back and clarifying the why for me has always been something that I’ve looked to do. And I think that’s genuinely something that’s been a big influence on how I approach projects and things like that.

Chantelle – 44:48: I can vouch for that as well. I work very closely with Neil on a lot of things, and he loves the question, loves the why.

James – 44:55: There’s a few people, but I guess, people that come to mind, again, the first boss I had when I first started working in the city, and he was very disciplined, old school approach, where it was: every meeting you go to, you have to have meeting notes afterwards fully typed up, and if you don’t, then, you’re reprimanded. You had to be suited and booted. And if you weren’t, you’d get kicked out of the office and couldn’t have anyone being late for any reason. There’s no excuse for being late whatsoever. It was like the height of rudeness to be late for anything. I think just having those principles instilled early early on in my career for making sure that you just get the basics right, – which is turn up on time, make sure you produce good quality documentation that other people can follow, make sure that the way you’re presenting things is presented in a good way as well. It’s just getting the little details right will lead to better work. So I think that springs to mind. And then although I work with them day to day and give them a lot of stick, Andrew and Thomas, who started Tap as well, I’ve worked with them now for eleven years, and they’re both very different people. And I learned lots of things from them over the years as well. Andrew’s got a very sharp focus on getting things done and trying to focus on what is the next thing you can do,what is the thing that’s gonna add value, your next action. I’m naturally more of a procrastinator and will think around things. So I think taking that learning just to focus on what’s important next and get that thing done has been very beneficial. And then Thomas, on the other side,it’s amazing what stuff he knows because he’s spent so long going down different rabbit holes on different information. But the way he’s able to get information very quickly from different places and then formulate some intelligent response off the back of that has been a key skill to be able to pick up from him as well, which is certainly helpful in our industry. We have to be slightly thinking ahead and making sure that you’ve got the information available to be able to add value to your client interactions as well. So I think those three people are probably people that spring to mind.

Neil – 46:59: Definitely, I’ve worked with Sam Taverner for a long time as well, and I give him a lot of stick, but he’s always very good at being a sounding board and getting you to raise up your thinking to more of a client level. Sometimes we can be guilty of getting so into the weeds that we forget about some of the story and how we engage as humans. And so Sam’s very good at holding that mirror up and saying, ‘what are we trying to get across for people to understand’, and ‘what’s going to matter to them?’ So, yeah, that’s it. He’s definitely someone I’ve learned from over the years.

Chantelle – 47:40: Lovely stuff. Well, it’s great to have you guys on. Thanks for having a little chit chat.

Neil – 47:44: Yeah. Thank you for having us. Thanks so much.

Mark – 47:46: And thank you everyone for listening to The CX Equation.

Chantelle – 47:53: That was a lovely chat with our colleagues. Colleagues sounds so funny! With our friends, there we go.

Mark – 48:00: Yeah. It was different, wasn’t it? Having a chat to two people that we work with most days, I felt like I knew what some of their answers were going to be beforehand whereas when we’ve interviewed some of our other guests, we haven’t known exactly what they were going to say. But it’s always nice to chat, and I enjoy working with James and Neil because they are both incredibly knowledgeable in their areas. It’s always good to get their take on things.

Chantelle – 48:25: Exactly. I felt like we were just having a chat about the office. But, yeah, it was good to have a little wrap up of the year from them. Obviously, there was a lot of talk about AI and where we are with things at the moment and not necessarily being able to prove the value yet, but people have started to experiment and lay the foundations for next year.

Mark – 48:42: Yeah. It’s difficult to avoid conversation about AI in anything at the moment, isn’t it? And I can’t see that going away in 2026 at all. So as, in particular, Neil was saying, the focus on ensuring that we’re actually generating some benefit from it and seeing metric improvement, I really hope that when we are looking back on 2026, it is, as he said, it’s more about having proven use cases, and ‘are we getting to see the benefit of implementing AI and having a clear reason for why we’re implementing AI’.

Chantelle – 49:15: Not just implementing for the sake of implementing. Indeed. The other focus we had was, the idea of composable architectures, of a composable martech stack. Obviously, previously, we’ve seen a lot of people have the one stop shop with the likes of Adobe and Salesforce and other legacy players.

Mark – 49:35: And you’re more of a technologist than me. Like, part of me wonders whether, we’re just going into the next phase of architecture. Right? And in three or four more years, it’ll it’ll be something else. You see these trends come and go, right, headless, composable, hub. Yeah. So, I do think getting your data right as a business is a fundamental thing. Right? And if/when the platforms allowed you to get away with not having great data, I understand why that is a benefit, because it meant that you could practically do something right now. But I think we’re now seeing that that is catching up with companies, and so more focus on getting a good data structure that platforms can sit on is is worthwhile.

Chantelle – 50:13: And then that’ll give more autonomy to marketing teams, or I guess they’re becoming more marketing technologists now where they’re doing more of the doing that the technical teams used to, and they don’t have as as much of a dependency on them. So operationally, it works in their favour as well.

Mark – 50:28: I hope you’ve enjoyed listening to this month’s episode of The CX Equation, and we you all have a great 2026.

Outro – 50:38: The CX equation is brought to you by Tap CXM. To find out more about what we do and how we can help you, visit tapcxm.com. And then make sure to search for The CX Equation in Apple Podcasts, Spotify, or wherever you usually find your podcast. Make sure to click subscribe so you don’t miss any future episodes. On behalf of the team here at TapCXM, thank you for listening.