All set? All right, we're going to get started. Thank you all for joining us. I'm Maggie Nolan. I'm the IT Services Analyst here at William Blair, who covers Endava. I am required to inform you that for a complete list of research disclosures or potential conflicts of interest, please visit our website at williamblair.com. Endava is a next-gen technology services company. We're excited to have with us here today Mark Thurston. He's the CFO of Endava. He's going to be giving a presentation. We'll do a little Q&A, and then immediately after the presentation, we will host a breakout session with additional Q&A that you all can participate in. That's going to be in room Burnham B. With that, I will turn it over to Mark. Thank you for joining.
Thank you very much, Maggie. Thank you, everybody. Go through the disclaimer. About Endava. As Maggie said in her introduction, we're characterized as a next-gen tech business. We used to have the tag name as digital, basically. We work basically at the forefront of technologies historically, helping our clients adopt them and basically had a focus on speeding them up in terms of their go-to-market or building productivity. Quite different to some of the more traditional players who would engage in managed services and support legacy IT estates. Very much change was part of our agenda, and we've been certainly impacted by the changes in the technology space over the last two years with the advent of AI. In terms of the background to us, and I'll focus on the bottom rectangle in terms of the timeline. We're a founder-led business.
We were founded by our CEO, John Cotterell, in the U.K. You can see we've had a fair amount of M&A over that sort of 25 years, which has been bolt-on. When I say bolt-on, it was basically to accelerate our strategy around diversification, because certainly in the earlier years, we had a preponderance of focus on financial services and particularly sort of payments. It was to build out the end industry verticals in which we operated in, build out the geographic footprint in terms of where we delivered from, and also the geography of the markets into which we delivered to. In terms of our expertise, which is sort of top left, as you can see, we are heavily an engineering-based organization.
We tend to operate at the moment in a distributed agile methodology, but we have a strong engineering heritage, which plays to our strengths in the marketplace in terms of building product and delivering change for clients. Underneath the capabilities, we've started to build out accelerators. Accelerators is something that has always been part of what Endava does, and one way to characterize it as being tools. For instance, Chronos, the first tool that is listed there, is something that we used to deploy across a client's estate to identify weaknesses or hotspots, et cetera, before we actually started work. The other elements of it, like Ray, Maps, and more importantly, sort of Compass and Morpheus, which are more agentic and more part of the Dava.Flow methodology, which I'll touch on later.
Accelerators is starting to move into a place of being, let's call it minimum viable software being built. It's not a product per se. It is a modular approach to how we will deliver solutions, and that is increasingly going to be part of our AI narrative as we go forward, which as I said, we'll touch on later. In the middle, in terms of the industry verticals we cover, which is important to us, we think it's important to have domain knowledge of the industries in which we operate. Our biggest segment still is payments at 23%, but it used to be a lot larger, and payments has been facing a headwind certainly for the last two or three years. Part of our M&A history has been to give us positions of strength in healthcare and mobility, for instance.
Another is what we use as our incubator for future industry sort of verticals. We've diversified. If you looked at this split, say four, five years ago, it would've been very heavily financial services, very heavily sort of payments. One of the important things also in that box is the tenure of the client relationships. We tend to have very sticky relationships with clients once we are engaged with them. We tend to be in the product development change area, which is discretionary in nature, has been part of the headwinds the company's been facing over the last two or three years. Once we are with a client, those relationships endure because we are seen as a trusted partner. You can see there's a sort of stat there, about 90% of our revenue is coming from the same cohort of clients in any one year.
We still have quite a lot of go get in terms of pipeline to get, but we do have an underlying sort of base of revenue on which we rely on. As part of our heritage, we have built out a global footprint. We're around 11,000 people. We're distributed globally, we'll touch on them later in the presentation, about 32 countries. Also importantly, we're developing relationships with some of the hyperscalers, in particularly sort of OpenAI, who are one of their founding sort of partners in this AI native delivery world that we're in. We're starting to lean more on partners as part of the AI narrative going forward, more of that later. Why is Endava attractive potentially to you guys to invest in? I think we have a potential large tailwind coming through AI adoption more at the enterprise level.
From what we have been seeing and others, since ChatGPT was launched onto the consciousness, is that it's moved very quickly, the technology, but the adoption is not high. There has been a lot of experimentation, a lot of proof of concepts, but it hasn't got into scale production systems, which is where we would play. I think there is a big potential spur to the market for those companies that are AI native, which is the second point. We think we are very well-placed for this. We are the perfect size. We're not a huge organization, 11,000 people, but we have fully embraced becoming AI native. All of our engineers are going through a process of training on Dava.Flow, which is our scaled delivery methodology in the AI world.
That Dava.Flow delivery methodology, which I'll also come onto later, I think is a key differentiator when we compare ourselves with the peers that would like to be successful in this market. I think the middle bottom point at the bottom of the grid there around industry alignment, we think is very important. We've always gone to market around the end industries. I think it's important because you have to understand the solutions that you're building in a particular industry segment. You have to have a viewpoint on how technology is going to impact them, and it makes the solutions that you develop all the more compelling. We go-to-market through our industry verticals, which is also, we think, a differentiator with the marketplace.
As I already said, we do have deep relationships with clients who trust us as a tried and tested producer of solutions for them and help them navigate this disruptive phase that we're in, which is the AI-native transition. Again, I think part of our pedigree is we have been around 25 years. As I said, we are AI-native. There has been a shift since the early 2000s when John founded the business. Certainly, the digital transformation shift that started to occur in the 2010s, as we're saying here on this timeline, is where Endava positioned itself to be a cutting engineering technology-led business that could deliver change through our methodology, which was largely distributed agile. We internally called it teams. We have experienced these shifts in technology and how they can accelerate a client's business model before.
I think the place we're in today, AI native, it is moving a lot more quickly and much more impactful. We think we are very well-placed to help advise our clients navigate that shift. We have the right mindset and culture to deliver through the shift that is occurring now and will do for the next few years ahead of us, certainly. What we are seeing at the moment in terms of this AI shift is that whilst, starting on the left-hand side, AI adoption is reasonably high, it is very thin. Penetration is very uneven, so it can be deeper in some places, parts of the economies and certainly industries than it is in others. I think what we're seeing is not quite the trough of despair, which some commentators call out.
The actual delivery of the return on investment on the investment so far, which has largely been through in-house teams in enterprises, is they've come up with very interesting proof of concepts and shown what the capability of AI is. The real return on these investments, which is enterprise-led, has yet to be demonstrated. We sense frustration, certainly at the C-suite level in terms of the return on investment on AI, which plays to the skills that we have at Endava, where we have a track record of rolling out cutting-edge technology at scale across organizations. One of the things that is also holding back that deployment is the governance around it. What are the guardrails around the legislation? How do you prevent hallucination, which is all part of the scaling, the repeatability of the services that you're delivering.
Our sense in the market, there's a slight tipping point at the moment. There's been a lot of experimentation with AI. A lot of budget has shifted from what would traditionally be externally with service providers such as ourselves, has been experimentation internally. I think largely it hasn't delivered the benefits that clients have been expecting, certainly at the C-suite level, which creates an opportunity for AI-enabled service providers such as Endava. We, as I said earlier, are developing our own methodology for delivering AI. We don't see it as a platform. We see it, again, as a methodology where you're deploying tools to deliver AI, which plays to being agnostic about the tools that you use, because clients will have a view on the various technology in terms of the solutions they want to use.
Our key point is that you still require humans in the loop to provide that sort of governance. You have a model, in terms of Dava.Flow, where you are using agents, but it is key that you have humans in the loop to manage that process. It requires a big change, so we are investing heavily not only in the tooling in terms of the software that we use, but also in retraining the personnel, and then shifting the sort of business model to do that, as well as putting the governance and the rails around that adoption. The other thing that we are also doing internally in terms of this shift that AI is putting on our business model is traditionally, Endava had not relied on partnerships basically to deliver revenue for us.
We didn't have a stronger relationship with a Google or an AWS that some providers have because we produced excellent solutions and didn't see the need for it. We see that they are driving the agenda in terms of the AI world, that they are going to be an increasingly important part of driving revenues for Endava. We will also have a focus on developing those relationships with those hyperscalers, and in particular, we have a good, strong relationship with OpenAI and Google. Dava.Flow is our AI-native delivery methodology. It isn't a platform. It isn't something that clients plug into. It isn't something that clients use for capacity. It's a tool-agnostic methodology with humans in the loop. Basically, it's a circle. It's a flow how it works.
We initially start with a process which is called Signal, which basically, if you go back to the old time and material world, we would call that phase exploration or discovery. It would be a bunch of Endavans sitting down with a client, looking at architecture maps, looking at any information around the client to come up with a discovery piece of work, which could be something like 8-12 weeks. Initially, it was to discover a way forward to address a particular solution a client was working for. Signal can actually start as part of the sales process with us. We can turn up to meetings, and we have done, where we've absorbed information from the client, and we can actually have an idea as part of the pitch about what the route forward is for that client.
The next stage, Explore, is where you're starting to take those Signals and start to construct agent-ready backlog. You are starting to think about what it is you're going to build and what that solution will look like. There's a triage process from Signal. I'll make this up just for illustrative purposes, 20 potential options to address the solution the client is looking for. You can triage it down to 10, let's call it proof of concepts, let's call it minimum viable products very quickly that can be put into production by agents in that Explore phase. The client is then working with us to spec it further and get more certainty before it actually goes into the Govern phase. Govern phase in the old time and material world would have been production.
We would have had, say, 20 people going through the phase Signal, Explore, and then moving into production, which could be scaling up teams to some of our larger clients where we would have 300 or 400 people working on it over a period of time. Govern for us is a different word. It is production. It's basically where you have agents and humans working alongside each other, and Govern is about making sure it stays within the guardrails that we have established, whether it's regulatory, security-wise. It is a very fast process. That would be where code is produced.
We think that that process, once you're into Govern, the speed at which work is delivered is multiple times quicker than we would currently get under a time and material world where even using some of the AI tooling that is out there where you get 20%-30% improvement in productivity. This is a level of magnitude a lot higher. You move round into Evolve. The product or system is in production, and Dava.Flow is monitoring how that product or system is performing. We began then producing imports into the Signal process to then start to iterate it. Dava.Flow is very different to time and material, which is all people.
The use of agents working alongside the human in the loop, as we call it, to basically produce the guardrails and sense check what is produced there, and it delivers work very quickly. That speeding up and the sort of certainty of the eventual outcome you're going to get because you're going through that Signal, Explore process very quickly means that you would price this work or the commercials around it would be very different to Time and Materials, which is you pay for as long as it takes because you don't quite know where you're going. If you've got a lot more certainty about where you're going, you can price it as fixed price or outcome based. Certainly in some of the more recent contracts that we've put through.
We mentioned our earnings call two weeks ago, Tyl with NatWest, a big U.K. bank, where that construction is there's a fixed price element to the services, but there's gain share basically on cost efficiency and revenue share, because they are getting the product very quickly, but there is also the benefits that will accrue to both the client and ourselves. Because we've got to that solution with a lot more certainty, we're able to price it that way. We wouldn't price Dava.Flow work on time and material, because it would really sort of create massive headwinds in terms of the sort of revenue. It's a big shift that we're going through at the moment, and certainly in terms of Endava. Now as I said, we increasingly will rely more on partners as part of our go-to-market.
I mean, the typical one is the second tier down, where it is the likes of the hyperscaler, the Microsofts, the OpenAIs. We will use them for introductions, but also use their technology and the solutions. It is a little bit wider than that. As you can see, going to our payments and our financial services pedigree, the second from bottom, where we will partner with specific sort of product companies and evolve their product set, but also enable them in terms of their go-to-market. We are plugging into a wider sort of ecosystem to drive revenues as you'd expect us to do. The next few slides are just a quick take through our sort of financials and where our capability is. We have moved, as you can see from the left-hand side, from largely a U.K.-centric business.
We're still about 1/3 of our revenues there. North America is increasingly a larger proportion of our revenues at 38%. Rest of world is Asia Pacific largely, and Middle East. Middle East was part of the reason we reduced our guidance at last quarter as we are doing work, particularly in the sort of payments and financial space there. Where we deliver from, which is the locations on the right-hand side. As you can see, we are global. We're in 32 countries. Again, we deliver predominantly from what is Central Europe, EU. Those places like Romania, North Macedonia, et cetera. That is the heritage that Endava sort of delivers from. We do have delivery capability in APAC, EMEA. APAC is largely Vietnam and Malaysia. We do have in there as well India. Central Europe, non-European and non-EU is places like Serbia and Moldova.
Latin America is obviously places like Colombia and Argentina. We do have a reasonably high level of onshore in North America, and that's particularly sort of linked with our healthcare business, and we also do in Asia Pacific. Sorry, in Western Europe. You can see it's very much a nearshore model where we have people working from same time proximity into the geographies where the end revenues are delivered. I can keep going, or we can take a quick-
Can we do the quick version of the quarter to level set, and then we can do some Q&A?
Yeah. Our last quarter, which you can see here on the revenues, we were at GBP 178.5. We were sort of down on what we were expecting. Again, given the sort of transition we're going through, a lot of our pipeline is a lot more lumpy than it was. Part of this transition that we are pushing for in an AI-native delivery is that we want to go for larger multi-year engagements. The deal lifecycle is a lot longer, a lot more elongated due to the complexity of the arrangements than we have expected and anticipated. The revenue pipeline conversion is proving more problematic for Endava at the moment, and certainly some of the macro elements that we're experiencing, particularly sort of the Middle East at the moment, is also impacting pipeline conversion.
In terms of if you look at our profitability and focus on the adjusted PBT, so we adjust out in there share-based payments and amortization of intangibles and realized effects. Our profitability has been particularly impacted by the pivot that we're going through at the moment. Very low adjusted PBT margin in Q1, which is 1.8%. We were anticipating we'd be around the sort of 5%. Part of that impact that we've been feeling this year, we did headline it to the market, is that the AI pivot that we're going through requires an awful lot of investment. There is retraining of personnel. There is a lot of investment going into Dava.Flow itself, which is mainly around sort of the software spend that we have at present. The Q3 profitability was low because of the revenue miss.
If you lose revenue and you can't tweak the cost base quickly enough, then that falls straight through to the sort of bottom line. Profitability is low, but it's part of this transition that we're going through from a, let's call it, a Digital transformation T&M business to an AI-native delivery business, which in the quarter was about 15% of revenue. It had grown very quickly from 5% on a 12-month sort of basis. It's triple in sort of size, but it's coming from a low base. The profitability on the AI-native business is significantly higher than the traditional T&M business. You've got two models or business models at play where you've got, let's call it a Digital transformation as a simplistic sort of take on it.
You've got a T&M business with margins, gross margins at around the sort of high 20%, which is slow growth under pressure. You've got a fast-growing AI -native business, which is growing more quickly from a small base, which is a lot more profitable. In terms of, you can see the sort of pressure in terms of our clients' relationships. We are reducing the number of clients that we engage with, so 584. This is part of trimming out our smaller client tail. We're focusing more on larger clients, multi-year engagements. That has been the focus of our sort of sales process. As a result, you can also see our sort of client concentration also picking up at 40% in the quarter just gone. There's quite a lot of stability in that top 10 clients.
You can see a little bit of pressure when you look at the average spend, certainly on the sort of top 10. There are sort of puts and takes in there. We are heavily dependent on our top 10 in our payments clients, Mastercard and Worldpay, which is now Global Payments. They are at inflection points with their spend profile, so a little bit of a sort of pressure there. You see the remaining spend from the clients outside the top 10. You've got some stability there. There's been a lot of comments about pricing pressure in the industry, given the sort of time and material nature of what we do. Is AI creating sort of downward pressure on rates? We've seen rates be broadly stable, which you can sort of see there.
I think with us, it's the volume of work coming through, which again, I think is caused by this, let's call it a trough of disappointment, in terms of corporates have not been spending with IT services providers such as ourselves, as they have been directing work inwards internally. The C-suite is not seeing the benefits of that. I think the work is going to come through to IT services as we go through that point where we're going into production scalable sort of systems, and that plays very well to our Dava.Flow narrative. Again, I think this is just replicating what we had on the first sort of slide. It's our geography mix and our industry sort of mix. I think that's about it really.
Yeah. We have time for a couple of questions, and then we invite you all to join us in Burnham B directly after we wrap here. You're talking about this pivot, and there's a desire for these AI deployments, but at the same time, a great concern for what the ROI is going to be with your clients. Can you help us distinguish what is different about where you do move things into production at scale versus pilots that stall with your client base?
I think there's been a lot of experimentation. I think that a lot of the proof of concept stuff clients have been doing themselves, or they've been helping us to Explore where they go. I think the rate of change in terms of the products, let's call it, for want of a better word, certainly coming from the likes of OpenAI, has changed. A client wants to move on something, the technology changes again. There's a little bit of a sort of hold back. I think we're getting into this sort of stability view at the moment. I think the change is always going to be quite rapid. It's a bit like you better get used to it. You need to sort of move. I think the investment that's gone in over the sort of two-year period, there was nothing of any scale.
There hasn't been a killer application that people have been sort of waiting for, I think is the wrong word for it anyway. People have been waiting for the adoption of AI. That hasn't happened. I think that frustration has been building up, certainly some of the CEO conversations that we've been having. We now need to sort of move. I think the conversation has become more strategic. We need to move because it is going to impact, and they will look for partners such as Endava to do that.
One quick one before we wrap. You mentioned the 15% of revenue, great growth rate there, obviously profitable for you all, kind of trending in the right direction. What are the circumstances that make clients more willing to shift towards an outcome-based or fixed price model, which appears to be more advantageous to you versus the traditional Time and Materials?
I think there's been a change. If you go back 20 years ago on that sort of that graph, or most IT services used to be fixed price, et cetera. There used to be a lot of time spent in specifying what you were going to get, and it would cost X, a lot of change control management around it. Doing the spec used to take forever, and there used to be big disappointment at the end because you'd either get something you didn't expect you were going to get, or it would cost a lot more. I think the agile journey was that process doesn't work, so there was a lot of iteration. You didn't quite know what you want.
You would move around, and then you would alight on a direction of travel, which was open-ended in some respects because the client would pay for that discovery before actually starting on it. It wasn't completely out of control, which is what I was sort of maybe implying there. I think with the start of Dava.Flow, where you're using agents to get that agent-ready backlog, and you've explored a lot of prompting, there's a lot more certainty around it. You pretty much know what you're going to get, and it's going to get there very sort of quickly. That sort of plays more to the traditional way that IT services are bought in the first place, which is around it's fixed price. We'll work out exactly what we want, and then we will sort of build it.
I think it's removed a lot more uncertainty, and certainly in terms of the speed at which it can be delivered. I think the models, the commercial models are definitely sort of changing. If I think about the heritage that Endava had, which was around sort of payments gateways, et cetera, and building product for clients. It would take, let's say, 18 months, cost GBP 30 million, and it would be completely bespoke. Now people want it quicker, and they want it at a less price. Now for the economics to work for us, it's okay, well, you will get it quicker. We've got a lot more certainty, but we want some of the upside you're going to get from that.
Sure.
Cost savings, et cetera. The construct is something that works for both sort of sides. I think it's the nature of the way that actually IT services will be delivered anyway.
Very good. Well, Mark, thank you for sharing with us, and hopefully we'll see you all in Burnham B.
Cheers.