管理層發言
Good day, and welcome to the Endava Third Quarter Fiscal Year 2026 Conference Call. Please note, today's event is being recorded. I would now like to turn the conference over to Laurence Madsen, Investor Relations Manager. Please go ahead.
Thank you. Good afternoon, everyone, and welcome to Endava's Third Quarter of our fiscal year 2026 conference call. As a reminder, this conference call is being recorded. Joining me today are John Cotterell, Endava's Chief Executive Officer; and Mark Thurston, Endava's Chief Financial Officer. Before we begin, a quick reminder to our listeners. Our presentation and our prepared remarks today include forward-looking statements including, but not limited to statements regarding our guidance for Q4 fiscal year 2026 and for the full fiscal year 2026, the impacts of headwinds facing our industry and business, trends in our industry, including with respect to developments with AI, enhancements to our technology and offerings, the benefits of our partnerships, demand from clients for our technology services, our ability to create long-term value for our clients, people, our shareholders, our long-term strategic positioning and our business strategies, plans, operations and growth opportunities.
These statements are subject to risks and uncertainties that could cause actual results to differ materially from those contained in the forward-looking statements. Actual results and the timing of certain events may differ materially from the results or timing predicted or implied by such forward-looking statements, and reported results should not be considered as an indication of future performance. Please note that these forward-looking statements made during this conference call speak only as of today's date, and we undertake no obligation to update them to reflect subsequent events or circumstances other than to the extent required by law. For more information, please refer to the Risk Factors section of our annual report filed with the Securities and Exchange Commission on September 4, 2025, and in other filings that Endava makes from time to time with the SEC. Also during the call, we'll present both IFRS and non-IFRS financial measures.
While we believe the non-IFRS financial measures provide useful information for investors, the presentation of this information is not intended to be considered in isolation or as a substitute for the financial information presented in accordance with IFRS. Reconciliations of such non-IFRS measures to the most directly comparable IFRS measures are included in today's earnings press release as well as the investor presentation, both of which you can find on our Investor Relations website or on the SEC website. A link to the replay of this call will also be available on our website. With that, I'll turn the call over to John.
Thank you, Laurence, and welcome, everyone. We appreciate you joining us for our third quarter fiscal year 2026 earnings call. I'll address first the issues that are top of mind for the investment community today. This has been one of the more challenging periods Endava has faced in recent years. Demand conditions remain uneven across several sectors, deal cycles continue to be extended and clients are scrutinizing technology spending more carefully than at any point since the macro slowdown began. Against this backdrop, the primary driver of the quarter's miss and the lowered Q4 outlook was a slower-than-expected pipeline conversion. Factors impacting the revenue miss in the quarter and the lower revenue guide include clients located in the Middle East delaying work due to the ongoing conflict, a slowdown in overall client demand due to the macroeconomic environment arising from the conflict, and finally, large complex outcome-based contracts taking longer to execute than planned.
During the quarter, we took a goodwill impairment of GBP 364.6 million, which is a noncash accounting adjustment, and does not impact our liquidity, delivery capability, client commitments or ability to invest in the business. Mark will provide additional details on these items shortly. Although we are disappointed by these outcomes, we believe it's important to distinguish clearly between near-term execution challenges and long-term strategic positioning. Over the past several quarters, we have accelerated our transition towards AI-native delivery, expanded relationships with leading hyperscalers, deepened our presence in payments transformation and increased engagement with senior client decision-makers, pursuing enterprise-scale AI initiatives. What we are seeing now is a market moving beyond experimentation. Clients are increasingly looking for partners who can help them operationalize AI securely, integrate it into complex enterprise environments and connect investment directly to measurable business outcomes.
Each new wave of technology change has triggered the same entrepreneurial reflex inside Endava: move first, learn fast and scale what delivers impact. The rapid emergence of artificial intelligence is simply the latest inflection point. And in recent quarters, we have concentrated talent, investment and partner collaboration on embedding AI across our delivery model to ensure Endava is ready to lead clients through what comes next. Robust enterprise-grade IT services are essential for enabling AI leaders to scale their products safely and quickly. And thanks to our deep AI-native delivery framework and expanding partnerships with both OpenAI and Google, we believe Endava is well positioned to provide the secure integration, cloud orchestration and compliance layers that make that growth possible. This quarter, we made strides in our go-to-market approach and in the evolution of our business model.
We are transforming our go-to-market approach by engaging directly with key decision-makers to show how AI can accelerate their transformation priorities while deepening partnerships. We're moving to outcome-based contracts. For example, PGX and a modular accelerator core for next-generation payment platforms delivered through Dava.Flow ties our success to measurable improvements in clients' payment operations. We're continuing to progress selected client engagements as part of our AI-native shift with Dava.Flow. We now have 12 clients where Dava.Flow is deployed as compared to 3 last quarter. We're seeing progress in our shift from a traditional digital transformation business towards an AI-driven business. These initiatives and others like them have moved our AI-driven business up from 5% of total revenue a year ago in Q3 FY '25 to 15% of total revenue in Q3 FY '26, or GBP 27 million. This shows the scale of the pivot Endava has undertaken during the past 12 months and now gives us an AI-driven base that we believe will continue to expand.
Margins on our AI-driven business are higher than our traditional digital transformation business. Let me share a few headlines for progress on this shift in the quarter. As part of our go-to-market pivot, we expanded our strategic partner network, enlarging our market reach and solution set. I want to highlight our recently announced collaboration with Mastercard which combines Endava's AI-native engineering and industry expertise with Mastercard's global reach and data-driven products and services. Together, we believe we have a powerful engine to accelerate the adoption and scale of next-generation payments and immersive experiences for Endava's clients worldwide. We aim to unlock value at pace, bringing solutions to market faster for Endava's clients with initial focus on high-growth sectors such as insurance and health care, with additional attention on telco, mobility and travel. On AI adoption, clients are moving beyond isolated productivity pilots.
They now aim to create AI-native initiatives inside their existing organizations and to launch entirely new businesses that embed AI in both build-out and day-to-day operations. Although these engagements sit at different ages of maturity, we're seeing increasing numbers implemented into production. Adoption is becoming more operational, more open and more tightly linked to measurable results. Over the coming quarters, we will focus on expanding our delivery portfolio with the goal of turning this interest into larger outcome-based programs. As part of our go-to-market strategy, we are investing strongly in partnerships, particularly those with the hyperscalers. By combining our depth of industry expertise with the scale of AWS, Google Cloud and Microsoft, we are producing accelerators and marketplace solutions that tackle our clients' most complex challenges. We expect to launch more than 15 marketplace offerings this year, of which 10 are already live.
And we are aligning Dava.Flow with each hyperscaler's platform. Together, these initiatives are expected to cut time to value and help clients realize measurable returns on their technology investments. Today, I want to share some of the momentum we're achieving with Google. Through our collaboration with Google, we have added new clients this year, particularly in financial services, retail and gaming. Enterprises are partnering with us to accelerate their cloud transformation and harness Google's AI capabilities. For Longbrook Insurance, we migrated and set up the AI security guardrails on an AI-driven underwriting platform for warranty and indemnity insurance, making a transformative step forward in digital underwriting built on Google Cloud. The solution uses advanced AI to automate key stages of the underwriting cycle from submission triage and risk assessment to pricing and due diligence while keeping underwriters firmly in control through a human-in-the-loop dashboard.
Their innovation in AI-powered insurance and InsurTech is designed to support considerable productivity gains, cost reduction and speed to revenue for Longbrook Insurance. Building on our enterprise AI progress, Google has invited Endava to participate in the Google AI Agent partner program, a program traditionally limited to their largest global system integrators. The initiative, now open to a small cohort of AI-disruptive partners recognized for expertise at Gemini Enterprise and Vertex AI, is already generating new strategic engagements in North America, APAC and Europe. For example, we recently finalized an agreement to implement Gemini Enterprise at a leading U.K. high street bank. The project is expected to deliver an enterprise-grade agent gallery that lets the bank's developer community register, govern and discover custom-built agents, fully integrated with Google data platforms such as BigQuery and Cloud Storage.
The solution is expected to provide timely, actionable data that improves efficiency and supports revenue growth at scale. A year ago, we began applying our AI-enabled engineering expertise to long-standing client needs in payments, a domain where we have more than 20 years' experience modernizing gateway and merchant services estates. We believe the sector now faces three concurrent requirements: one, lowering the marginal cost of scaling; two, tightening operational efficiency and control; and three, keeping room to innovate around embedded commerce, omnichannel acceptance, platform consolidation and marketplace models. Our answer is PGX delivered through Dava.Flow. PGX provides a reusable core, spanning digital acceptance, orchestration and routing, merchant portals, onboarding, settlement, fraud management, developer tooling, partner/ISV enablement and back-office services so clients can modernize selectively and still differentiate at the product and experience level.
Shared configurable components cut scaling costs. Standardized orchestration and back-office services boost efficiency and leave headroom to innovate. Crucially, PGX supplies the data and workflow layer needed to introduce agentic AI-enabled operations and agentic commerce across the front office, onboarding, servicing and back office. Built with agentic AI and strict human-in-the-loop governance, PGX demonstrates the accelerated AI-assisted engineering can meet the quality and compliance demands of complex regulated payments environments. Early market reception is encouraging, with new signings in the last three months. Interest is already expanding beyond financial services into other sectors where modern payment capability is becoming central to customer engagement, efficiency and growth. First, we have been selected as a strategic partner with Tyl by NatWest, NatWest Group's merchant payments arm, to modernize and expand its payments acceptance platform.
Under the partnership, Endava will deploy Dava.Flow together with components of PGX to speed the rollout of new fully integrated products and services while improving flexibility, scalability and end-to-end performance across the payments life cycle. Working jointly, the two companies have mapped a business and technology roadmap that links specific feature deliveries to defined market opportunities and associated revenue targets. For NatWest, the partnership represents a material investment in strengthening its merchant payments offering. For Endava, this partnership adds an additional and significant large-scale complex engagement with a leading U.K. financial institution, reinforcing our credentials in payments transformation. Second, PGX continues to gain momentum with two additional wins, one with a global payments provider and another with a pan-European energy retailer. Both clients chose the accelerator to cut operating costs, simplify estates and accelerate time to market.
Dava.Flow supplies the delivery engine that converts these modernization programs into measurable commercial value and seamless integration to ecosystem partners such as payment schemes, acquirers, POS hardware and compliance providers. Some other client wins: we have also recently renewed our long-standing partnership with Slovenia's Ministry of Finance and Financial Administration through to 2028, bringing the relationship to more than two decades. Under the new agreement, we will continue to operate and enhance eDavki, the national tax portal that serves hundreds of thousands of taxpayers, integrates over 200 tax-related services and processes more than 12 million electronic documents each year. eDavki delivers a secure integrated experience that has eliminated postage costs, accelerated processing times and given the authority near real-time visibility across its core revenue systems, demonstrating Endava's ability to modernize mission-critical high-volume platforms at a national scale.
The insurance company North Standard now regards Endava as a trusted extension of its technology organization, combining strong cultural alignment with deep technical expertise to deliver consistently high-quality outcomes. The success of the partnership and the value delivered by our team gave our client the confidence to extend the engagement for a further two years and expand into additional roles and delivery teams. Our collaboration with a global brand and vehicle manufacturer has progressed from stand-alone engineering projects to an embedded partnership that has designed, built and operated cloud-native data platforms for connected vehicle services, real-time performance monitoring, provided around-the-clock support services and delivered logistics systems covering more than 80 facilities in nearly 30 countries. We are currently using AI-enabled delivery frameworks to create modern production operations systems designed to improve life cycle management, enhance data visibility and raise efficiency in production-critical environments, all underpinned by our disciplined approach to high-performance, scalable and compliant architecture.
Let me turn to Dava.Flow and AI projects. Over the quarter, Dava.Flow has shifted from exploratory use to enterprise adoption. We enhanced the framework through a combination of partnerships and by applying it in two large-scale live engagements. Firstly, a large-scale implementation engagement in a regulated high-assurance environment. And secondly, the TechNexus technical operator program, a previously announced engagement in the payments field. We have also continued to advance an AI-enabled human movement analysis platform for a leading high-performance sports organization with the quarter focused on validation, robustness and operational readiness. Working closely with domain experts, we refined evaluation logic to improve alignment between system outputs and expert expectations, strengthened the core analytics pipeline, and expanded synthetic data sets to improve performance across real-world scenarios.
We also introduced more structured measurements through accuracy dashboards, regular comparisons to previous versions and standardized evaluation against labeled data. Although still early, the increasing level of stakeholder validation underlines that the program is moving steadily towards a production-ready solution. We applied the same agent-centric principles to a very different challenge, streamlining engineering knowledge for a European-based media and entertainment group. The client struggled with fragmented engineering knowledge locked in Jira, Confluence, GitHub and Microsoft 365, which lengthened incident resolution, delayed sprint planning and hampered onboarding. Endava delivered a Google Cloud agent-based pilot that unifies these sources behind a secure role-based natural language interface, automatically retrieving the most relevant tickets, code and documentation in one view.
Since go-live, engineers report a roughly 60% reduction in time spent locating material and cut onboarding time by 30%, translating into faster coordination and measurable gains in overall delivery productivity whilst also validating our agentic approach and further strengthening our partnership with Google Cloud. To conclude, I want to thank our employees across Endava. Our teams continue adapting quickly to technological change while supporting clients through increasingly complex transformation programs. We remain focused on disciplined execution, operational accountability, clients' delivery quality and positioning Endava for long-term relevance in the next generation of enterprise technology services. With that, I'll hand the call over to Mark, who will walk through this quarter's financial performance and our guidance for the rest of the fiscal year.
Thanks, John. Revenue for the quarter ended March 31, 2026, was GBP 178.5 million. The revenue miss in the quarter was due to opportunities moving beyond March. As John mentioned, there were several factors that impacted revenue this quarter, along with the revised Q4 outlook: mainly clients located in the Middle East delaying work due to the ongoing conflict, a slowdown in overall client demand due to the macroeconomic environment arising from the conflict, and finally, large complex outcome-based contracts taking longer to execute than planned. This compares to GBP 194.8 million in the same period in the prior year, representing an 8.4% decrease. In constant currency, our revenue decreased 6.4% from the same period in the prior year. Loss before tax for the three months ended March 31, 2026, was GBP 372 million, which includes a noncash goodwill impairment of GBP 364.6 million compared to a profit of GBP 13.6 million in the same period in the prior year.
Our adjusted PBT for the three months ended March 31, 2026, was GBP 3.2 million compared to GBP 24.6 million for the same period in the prior year. Our adjusted PBT margin was 1.8% for the three months ended March 31, 2026, compared to 12.6% for the same period in the prior year. Our costs increased in the quarter due to higher go-to-market investments and an increase in the bench as we are training staff in AI and Dava.Flow skills. This is a key investment in skills for our new AI-driven business. The market capitalization of the company and the reduced outlook has required us to assess the carrying value of goodwill and the deferred tax asset for U.K. tax losses in the U.K. As a consequence, we have taken an exceptional charge of GBP 364.6 million in relation to the impairment of goodwill and GBP 23.2 million regarding the derecognition of the deferred tax asset. Those charges are noncash and one-off in nature.
The deferred tax asset derecognition, because it has occurred partly through the financial year, impacts our adjusted tax rate, which for Q3 is 17% and is expected to rise to 37% in Q4, leading the estimated full year adjusted tax rate to around 25%. These adjusted tax rates do not change the amount of cash tax we are paying. Our adjusted diluted earnings per share was 5p for the three months ended March 31, 2026, calculated on 52.2 million diluted shares as compared to 34p for the same period in the prior year calculated on 59.4 million diluted shares. Revenue from our 10 largest clients accounted for 40% of revenue for the three months ended March 31, 2026, compared to 39% in the same period last fiscal year. The average spend per client from our 10 largest clients decreased from GBP 7.5 million to GBP 7.1 million for the three months ended March 31, 2026, as compared to the three months ended March 31, 2025, representing a 5.6% year-over-year decrease.
Foreign exchange contributed to a 3.7% year-over-year decrease due to U.S. dollar weakness in the quarter. In the three months ended March 31, 2026, North America accounted for 38% of revenue, Europe 23%, the U.K. 33%, while the rest of the world accounted for 6%. Revenue from North America decreased by 5.5% for the three months ended March 31, 2026, over the same period last fiscal year. The decrease was driven by an FX headwind of 6.1%. Comparing the same periods, revenue for Europe declined 3.6% due mainly to weakness in payments and TMT, and the U.K. decreased 15.4% due mainly to the reclassification of the large payments client from the U.K. to North America as the relationship with the client is now based there, which was mentioned last quarter. The rest of the world decreased 1.8%, driven mainly by the payments and other verticals. Our adjusted free cash flow was negative GBP 3.1 million for the three months ended March 31, 2026, from a positive GBP 17.5 million during the same period last fiscal year.
Free cash flow was negative in the quarter, mainly due to an increase in receivables as a large proportion of the billing for the quarter was issued in March. We anticipate collecting the majority of this by the end of June. Our cash and cash equivalents at the end of the period totaled GBP 48.4 million at March 31, 2026, compared to GBP 59.3 million at June 30, 2025, and GBP 68.3 million at March 31, 2025. Our borrowings increased to GBP 195.8 million at March 31, 2026, from GBP 180.9 million at June 30, 2025 and GBP 136.5 million at March 31, 2025, primarily to support the funding requirements of our share repurchase program. Capital expenditure for the three months ended March 31, 2026, as a percentage of revenue was 1.6% compared to 0.6% in the same period last fiscal year. Turning to the guide for the remainder of the fiscal year, as John mentioned earlier, we have lowered the Q4 guide due to slower-than-expected pipeline conversion, which is most marked in banking and capital markets across all of our regions.
Now moving to our outlook. Our guidance for Q4 fiscal year 2026 is as follows: we expect revenue to be in the range of GBP 181 million to GBP 185 million representing constant currency revenue decrease of between 3.5% and 1.0% on a year-over-year basis. We expect adjusted diluted EPS to be in the range of 9p to 13p per share. Our guidance for full fiscal year 2026 is as follows: we expect revenue to be in the range of GBP 721.8 million to GBP 725.8 million, representing constant currency revenue decrease of between 6.0% and 5.0% on a year-over-year basis. We expect adjusted diluted EPS to be in the range of 45p to 49p per share. The above guidance for Q4 fiscal year 2026 and the full fiscal year 2026 assumes exchange rates on April 30, 2026 when the exchange rate was GBP 1 to USD 1.35 and EUR 1.16. This concludes our prepared comments. Operator, we are now ready to open the line for Q&A.
分析師問答
Our first question today comes from James Faucette at Morgan Stanley.
Thank you very much. I wanted to dig in quickly into two topics — a little bit unrelated, or I mean, always related, but separate. First, in terms of customer decision-making: obviously, there's a lot of AI evaluation going on. You talked about projects moving to production, but we're still seeing pressure on the rest of the budget and spend. You talked about extending decision cycles. Can you just help us bridge those, and when or under what conditions you would expect to see that movement to AI production start to benefit you so we can start to see real movement on the bookings side? And then on capital allocation, can you just talk about how you're thinking about what you should be doing around your debt and borrowings, especially — obviously you've tried to take advantage of where the stock is with buybacks, but just wondering if delevering is an increasing priority, et cetera.
Thanks, James. So let me pick up the customer decision-making question that you had. We are seeing much more substantive AI-driven deals coming through. We announced NatWest and the collaboration we have with Mastercard in the opening remarks. And it's visibly growing as a proportion of our business compared to what it was a year ago, taking it to GBP 27 million in the quarter or 15% of the total business. So it's now becoming a substantive element of the business that we expect to grow from. We're seeing that in respective deals that are coming through. They are more complex in nature, outcome-based, looking at serious transformation across the customers' business, and they've taken longer to close and get started. We do use AI very much as part of that sales process, so actually establishing what is going to be done is a very AI-driven process. You are correct that there is pressure on discretionary spend.
I think in Endava, we are more exposed to that than many of our peers; a lot of our business is more in the discretionary camp. And we continue to see downward pressure on that. You can see that in the underlying shift of our business from our traditional digital transformation business towards the AI-driven business that I highlighted in the opening remarks. So obviously, the digital transformation business has been declining whilst we've been seeing the AI side ramp up. However, that is the shift. That is the pivot that we are deliberately making as a company. And we're very comfortable seeing the AI-driven arena growing. Mark, do you want to pick up on the capital allocation?
Yes. So the cash generation in the quarter was disappointing. As I said, most of the billings arose in March. So the collections will take place between now and June. So we anticipate a significantly better cash flow generation in Q4. But notwithstanding that, leverage is something we want to focus on reducing. We do have a refinancing coming up during the course of FY '27. But looking at the wider funding of the business, it is something that we will consider as part of that.
Our next question today comes from Bryan Bergin at TD Cowen.
So maybe a bit of a follow-up as it relates to the unplanned pressure here. I understand the Middle East volatility causing the discretionary issues and large deal opportunities taking longer than planned. But in addition to that, is there vendor consolidation and broader shifts in client priorities playing out where the offering just isn't as robust yet as competitors? I'm really trying to understand how much is maybe a transitory timing dynamic versus a function of clients setting programs and shifting those priorities elsewhere or consolidation share loss or even other factors like over-competitive pricing in the market? If you could just comment on that.
So we're not seeing a huge vendor consolidation headwind. The pressure seems to be coming from as we deliver more productively. We've been talking about our shift to AI-native where more than 75% of our staff are now using AI in their daily work, and that is driving higher productivity. Clients are harvesting a little bit more of that benefit than we would prefer as we are not reinvesting it. But I think that is also part of the shift as they're looking to a much more substantive AI-driven transformation, and that's very much part of the pivot that we are focusing on. Those projects are taking longer to come through. They continue to take longer. The thing that I'm highlighting is that they are coming through, and we are getting them signed there.
Okay. And then my follow-up as it relates to some of the actions by the foundational model providers. With news flow around their deployment companies and some of the joint ventures, they are looking to set up consulting and engineering talent. What's your perspective there? Considering your base of engineering talent, it seems like an attractive potential opportunity for them to lean into partners like you more. But as you think about kind of competition versus cooperation, where do you land? What are your thoughts on them?
Yes. So number one, I think the foundational model companies are actually showing that they need services partners for implementation in the real world, and they're looking for how to accelerate that and push it along, and that's the reason for some of these deploy co models that they're coming up with. We are in conversations with them very, very much. The expectation is that that will become a new channel to market for us as they utilize our skills and capabilities in order to drive the commitments that they will be making to clients. So we see it much more as being a collaboration opportunity, a go-to-market opportunity than a competitive activity. A lot of what they're focusing on is complementary to what we do in terms of the heavy-lift engineering capabilities that we've built in the AI space, and they recognize that.
And our next question comes from Puneet Jain with JPMorgan.
I want to follow up on Bryan's question on revenue weakness. I want to focus on both related to your estimates as well as your peers. I understand that the Middle East surprised you in this quarter, but revenue has come in below your expectations many quarters in the last three years. Do you think you need to change anything in the planning process to get a better handle of quarterly revenue or the quarter as well as full-year guidance?
Yes. I mean the Middle East arena was not something we saw coming. We'd actually invested pretty heavily over the past 12 months and had deals ready to sign, literally about to kick off when the conflict kicked off. And it literally stopped all activity across our client base. And that had a noticeable impact on Q3 and a bigger impact on Q4. Your question about the timing of how quickly we get these deals through the pipeline and into revenue is one that we're paying a lot of attention to. We're very sensitive to it. If you look at our Q4 guide, we've put a lot of work into the client conversations and the project plan, if you like, of getting these deals signed and the revenue ramping. So it is something that continues to need a lot of attention, and we definitely got caught by that in Q3.
Okay. Got it. It's been about a couple of years since you acquired Galaxy, and now you are also pushing ahead with this AI-first model, AI-first delivery. Talk to us about change management within Endava: how are you motivating employees to embrace AI to increase this new way of delivery while also the stock obviously has been down so much, and some of those employees might be worried about their jobs given the news flow around AI. So talk to us about the change management within Endava, how are you managing all those things?
Our approach to change here has been a pioneer-and-rollout model. So in each area, as we're driving change, we get a smaller group of people who pioneer what good looks like and then roll it out across the organization. So the first of those that we talked about around 18 months ago was the shift to AI-native. That was done by getting small teams across each part of the business to engage with AI at that stage. It was the generative AI that was in play and how to create GPTs and how to drive usage across each part of the business. We then moved into a rollout phase where adoption was pushed right across the business with everyone having access to GPT Enterprise as our standard across the business. Over the following three to four months, we saw usage across our staff base move above 75% of people using it every day in their job, which was our objective to have that AI-native shift. The big shift that we're pushing at the moment is Dava.Flow.
We've been developing Dava.Flow over the last 18 months or so. You'd be aware that we came out with our own agentic solution ahead of some of the large vendors coming up with agentic models. We were using that to initially define how Dava.Flow would work. Dava.Flow is our method around how you develop business solutions and ultimately software and agentic solutions in an agentic world where most of the work is done by agents rather than by people. That shift from traditional agile has been pioneered by small groups that defined Dava.Flow, created all the prompts that go into it, created the context warehousing and all of the pieces that go to make Dava.Flow work. We pushed that into our payments gateway so that we had an internal project where we could really drive not only the payments capability but also the development of Dava.Flow. Over the last six months, we started to spread that step-by-step across the organization.
I highlighted we've now got 12 clients using Dava.Flow, up from 3 last quarter. So that's the rollout speed. Within the organization, we've got over 1,000 engineers who are actually using or training on Dava.Flow now — over 10% of our direct staff — and that's in anticipation of the greater use of Dava.Flow that we anticipate coming through both Q4 and as we move into Q1. All of that is managed within a change framework we call our Keystone management program that is driving that change.
And our next question today comes from Nate Svensson with Deutsche Bank.
I wanted to ask about another one of the factors you called out as driving the miss and the guide down specifically — the outcome-based contracts taking longer to execute. Hoping you can give some detail around what exactly is taking longer to execute here. Broadly, you've clearly talked a lot about the shift to outcome-based contracts over the last few quarters. So I'm wondering if these problems or the things that are taking longer to execute are actually fixable or transitory? Or is there any sort of dynamic where clients just don't want to shift to outcome-based models to try and realize benefits on pricing or efficiency versus your traditional time-and-materials contracts?
We're not seeing the latter problem. These are, by nature, very large transformative engagements in the tens-of-millions category and nailing down scope takes time. We're using AI to help give clarity on what it is that we're going to be delivering and getting that much earlier in the cycle and expecting that we would see a three- or four-month sales cycle from having shaped what it is that we're going to be delivering and how AI is going to be making an impact. But we're seeing that turn into five or six months to get the deals closed. There is an element of clients being on a learning curve — their legal departments worrying about issues, worrying about regulation, worrying about how to contract these deals — and that is becoming visible and it's taking longer. We expect that to ameliorate as people become more familiar with the issues and can get these things through faster on their side. We're not seeing it because they don't want to engage on outcome-based deals. These things are progressing. We announced some in the opening remarks, and there are others under the covers that are a little smaller. We are seeing them progressing. They're just taking longer than expected.
Got it. And then for a follow-up, I wanted to ask specifically on two of the verticals. First, what's happening in banking and capital markets? That vertical had been growing nicely for you and growth fell pretty dramatically in Q3. I think you mentioned worse pipeline conversion, but more color would be helpful there. And then in health care specifically, I think on the call last quarter you talked about a large health care client slowing down spend in Q3 but expected them to return to spend in Q4. So it looks like that played out in Q3, but is that specific client still expected to return to spend here in the fourth quarter?
On the health care side, yes, we expected one of our larger clients to slow, which they have continued to do. So they came in as expected. And we expect that to continue slowing into Q4. There is some offset to a certain extent as we go into Q4 because another client is actually growing quite quickly. The trouble is the decline of the larger client has happened more quickly than anticipated with the ramp-up of the newer client. We do have a ramp down from an existing client from Q2 through Q3 and Q4. So you've seen a step down because of those dynamics, but it stabilizes as we go into Q4 as anticipated in the guide. In banking and capital markets, we were pretty stable through Q1 and Q2. We did see a step down as we went into Q3 — I think a couple of million, roughly GBP 1.5 million or so — partly due to one client coming off project work that we've been doing for them and also some lumpiness in the delivery profile for another client. But we do expect recovery in banking and capital markets into Q4. The point is it's not as strong as we were anticipating in the guide we set in February. That slowdown in banking and capital markets is most pronounced in the U.S. and the U.K., although we are feeling it to a smaller extent in other geographies.
And our next question today comes from Jonathan Lee at Guggenheim Partners.
Given what we saw in the quarter versus the mid-February commentary around 95% contract and committed visibility, what are you seeing quarter-to-date in April and May on both demand and slip contracts? What's the coverage on the Q4 range today? What gives you confidence in that sequential improvement into Q4 that's implied in the outlook?
If you go back to Q3, we had a range of GBP 185 million to GBP 182 million. We were saying the contractual coverage at the high end was about 90% and it rose to about 92% for the low end of the guide. So the pipeline to convert in both high and low was about GBP 19 million and GBP 16 million, and we converted about GBP 13 million. So you had a conversion which was below what we anticipated — at the low end it's about 80%. Now for the Q4 guide, for the high end we have contractual and committed revenues of 95%; for the low end of GBP 181 million, we have 97%. That leaves about GBP 9 million to convert at the high end and GBP 5 million at the low end. We have three or four opportunities that are sizable. But we have taken a view that some of those are not going to convert as part of the high guide, and in a severe downside one converts when we get to the low end of the guide. So we have been conservative.
I know we missed in the quarter with that outlook. In terms of the step-up, it's something like at the top end of the guide about I think 3.5%. We do have some movement in terms of working days between the quarters that actually helps us somewhat, so the step-up is not as strong as it may appear when you look at it on an absolute basis. But there is always pipeline in our figures; it's the nature of the business model. The issue, going back to John's initial comments, has been the predictability of when opportunities convert.
Thanks for that color, Mark. Just as a follow-up, can you help us think through some of the earlier comments around AI productivity harvesting? As clients become more aware of the efficiency gains that AI is enabling, how do you think about the structural durability of pricing and contract profitability over the longer term, particularly as clients may look to extract more of those gains at the table? What's the offset mechanism there?
I think the offset mechanism is the change in the business model John was outlining in terms of AI-driven models, which are basically outcome-based. Yes, there's been pressure in the traditional time-and-materials space. We are being more productive; it erodes revenues. But we're moving more to outcome-based, longer-term duration partnership arrangements with large clients, where we have stronger visibility of revenue and we can capture more of that benefit from the rollout of Dava.Flow. Therefore, we can protect margins. The key thing is building the new AI-driven business more quickly to offset the decline that we're seeing in the traditional digital transformation business, which is largely T&M. The new model revenue margins are significantly higher than our existing T&M margin figures, which are under pressure. It's a question of can we accelerate the new, more profitable work to offset that decline.
I think the other thing to add is it's not specifically around productivity and pricing. Our utilization availability is running lower as we're going through this pivot; we're investing in retraining and skilling. We need to do that to prepare our workforce for the pipeline coming through. It's not optional, and that is a big part of the margin compression you're seeing rather than a pure pricing issue. Pricing has been pretty stable when you look at it on an average work-day basis. The usual metrics of billability and average rate per work day are shifting as we go through this change.
And our next question today comes from Matt Desert at William Blair.
This is Matt on for Maggie Nolan. To follow up on that last point on AI, how are you defining your AI revenue? What growth trajectory are you underwriting there? When do you expect that could become a majority of the business mix?
We've pulled this out as what we're calling AI-driven business where AI is at the core of the business transformation proposition, often outcome-based and typically sold at the top of the C-suite. Examples like NatWest and the collaboration with Mastercard and a number of the Google Cloud deals fall into that category, and we're very focused on developing this type of pipeline. It needs forward-leaning organizations who are up for this acceleration, and that is a subset of the market. It's not everyone who's up for that right now. But where we find those customers, we're getting really good traction around the AI-driven change. I would highlight it's different to the AI-native measure that we've previously published, which has stabilized in the 75% to 80% mark and measures how people are using AI in the organization daily. That enables strong productivity, but it's not the same as the AI-driven business transformation that we are classifying here.
Got it. And then on the margins, what specific levers do you have to protect or expand margins given the revenue pressures you're seeing as you pivot the business? How do you think about that going into next year?
Margins: we're managing two dynamics — the traditional T&M business and the new AI-driven business. For the traditional business, the lever is cost management and getting better visibility so that changes are not disruptive. The other offset is building the new AI-driven work more quickly where you have longer-term visibility year to year and more control over how you deliver that work because it's not T&M. Deploying Dava.Flow helps capture that benefit. Those are the two levers: manage costs and accelerate the more profitable AI-driven work. We expect over coming years the split between fixed-price outcome work and T&M to shift, with T&M proportion of revenues starting to come down as the new work grows.
And our next question today comes from an analyst with HSBC.
I just want to ask on the update regarding your go-to-market with OpenAI. You have a partnership. Do you have any update on how it's going?
Yes, we continue to have a really strong relationship with OpenAI. It's global in nature, driven out of the U.S. The conversations we had around the new deploy code are a part of that relationship, and through that we're putting together thoughts and plans on how we'll work together with the new deploy code. We continue to get early sight of some of the models they are putting out so that we can prepare go-to-market capabilities alongside them, and we continue to bid together on opportunities, some of which are in the large complex space.
That concludes our question-and-answer session. I'd like to turn the conference back over to John Cotterell for any closing remarks.
Yes. So thank you all for joining us today, and I look forward to speaking to you in September.
Thank you, sir. That concludes today's conference call. We thank you all for attending today's presentation. You may now disconnect your lines, and have a wonderful day.