Ocado Group plc

Good morning, everyone. Very good to be with you here today to talk about FY '25.

And FY '25 was a year of real tangible growth for Ocado, but one that also saw the business mature in a number of important ways. And while we've seen robust growth in the business and good progress across most of our global operations, we also worked to help some partners address a number of key challenges in their early network decisions.

This included constructive engagement with our partners in North America, as they made decisions to close sites in areas where demand has not evolved as initially expected. And in fact, last year, we reflected that a number of our partners were looking at a small number of sites, which required a different strategic approach.

And while the decisions made in North America to close were difficult, it does reflect a mature approach with those partnerships and putting them on a stronger foundation for long-term and sustainable growth. With exclusivity now having ended in North America, we've begun the journey to reengage in many of the commercial opportunities available in that very large -- world's largest grocery market.

And Tim will reflect on this and Ocado's approach to reentering the wider global opportunity as Ocado moves into this next phase of commercial growth. So I look forward to hearing more about that soon.

I've now been Chair for just over a year. And during that period, I've spent a lot of time engaging closely with a range of stakeholders.

And reflecting on what I've learned, I remain still very excited about the significant opportunity that remains in -- to solve a range of business issues across the omnichannel journeys of our retailers. And Ocado itself is still a business that has a huge breadth of talent, a unique and world-leading technology platform, a visionary leadership team and a scaled commercial relationship with many of the leading retail brands around the world.

I've enjoyed getting under the skin of these issues over the past year. And while recognizing that executing in a competitive and ever-changing world is challenging, I do believe Ocado is well positioned to take advantage of the significant global opportunity, both with current and future partners.

I particularly value the time spent with many of our Ocado partners, including counterparts at Coles, Ocado Retail, of course, Kroger and of course, our JV partner, M&S. This engagement gives a tremendous window into the strategic thinking of our partners and in particular, a depth of understanding on how some of the world's most successful retailers think about their own long-term success and growth.

Today, you'll hear from Stephen and Tim about progress we're making towards the key priorities that we laid out at the half year. First of all, our core priority to turn cash flow positive later this year with full year cash generation in FY '27, the measures that we're taking to drive continued growth and greater efficiency with our partners, and lastly, how we're reconfiguring key parts of the business to make sure we're well set to take advantage of the renewed and significant global opportunity.

So over to you, Stephen.

Thank you, Adam, and good morning, everybody. I hope you're all well.

Thank you for joining us at today's full year '25 results. I'm going to take you through the financials.

Next slide, please. Okay.

Here are the headlines. So good financial progress across the board really.

Revenue grew, group revenue by 12%. I'm going to take you through the logistics and the tech solutions growth shortly.

We had strong adjusted EBITDA growth of GBP 66 million to GBP 178 million. The underlying cash flow, if you exclude the letter of credit, was a GBP 230 million outflow.

But if you were to take that into account, underlying cash flow would be GBP 140 million -- GBP 130 million better, driven by that receipt from the letter of credit. I should say upfront, by the way, on the closure fees from the 4 site closures and on the letter of credit, the accounting is not straightforward.

It mostly impacts fiscal '26 and future years, but we've included in the appendix a couple of charts that show you how it all works when it comes to do your modeling. So I just wanted to make that open upfront.

The retail -- sorry, the underlying cash flow in terms of credit, I talked about that. Liquidity finished the year yet again with healthier liquidity of GBP 700 million or so of cash and the access to the revolving credit facility.

This has been bolstered further by the GBP 279 million inflow that came in post the year-end. So we're sitting on very good cash balances today.

It does mean, and I'll get into it when we look at our debt chart shortly, as we approach our debt maturities with the optionality there. Certainly, in the first instance, the GBP 350 million convertible bond that's due in January '27, we can pay out of cash, which you will see our gross debt numbers starting to come down, important factor for us, particularly when you see the trend in interest costs.

Yet again, we achieved our guidance for revenue, margin and cash flow targets and hit all of those. There's probably nothing more I'll say here.

I'll take you through the detail now of each business. Here's the statutory chart, getting you to your earnings before tax of GBP 403 million, a positive number, but benefiting, of course, from the big adjusting item of the valuation of Ocado Retail of the stake of our 50% in Ocado Retail when we deconsolidated the asset and M&S took over consolidation.

As a consequence of that, we took our value that we put in at GBP 1.5 billion or so, half of that and then you adjust for the assets that are on our balance sheet to get to that adjusting items income. A couple of other callouts.

Tech solutions revenue growth, I'll go through that. Logistics, 11%.

I think it's probably worthwhile calling out the finance cost line, a GBP 48 million increase in our interest costs. As I've mentioned earlier, there are plans to address that debt and gradually reduce that gross debt level.

I'll get to that shortly. So Ocado Group adjusting items.

Here is the key item there at the top that I talked about, Jones Food, if you recall, went into administration last year. We wrote off those assets that were consolidated on our balance sheet.

We were a consolidating company. The Kroger of lessor credit pre fiscal '25.

So whilst the cash was received in early this year -- sorry, last year, in fact, there is an accounting recognition of the revenue related to prior periods, which is an adjusting item in the prior period. You'll see at the chart at the back how it works.

It's not straightforward. The organizational restructure, that is not the cost of the restructuring that we're about to do.

We did a small amount -- a relatively small amount of restructuring principally in G&A and in technology in the first half of this year. So there's a small amount in there.

The rest is pretty straightforward. So tech solutions.

Well, you know the business model, grow the average number of live modules on that point, and we'll come to it shortly. That's probably the key number in respect of becoming a cash flow positive business full year fiscal '27, turning cash flow positive in the second half of '26.

121 modules today, driven by -- the growth there is driven by new sites going live. We've got around 6 sites going live over the next couple of years, but also drawdowns in existing sites.

Those are the 2 drivers of that growth in modules. The quicker we grow our utilization of those sites, fill their capacity, the more modules and CFCs that are ordered going forward.

That's a key metric for us. Recurring revenues make up the bulk of that revenue, that GBP 444 million, growing by 7%.

And as you'd expect, that's in line with the growth in average number of live modules, but also the fees that we get, as you'll see shortly, per module that are indexed every year to local inflation. The nonrecurring revenue, a material increase in nonmaterial revenue by GBP 41 million, but there's a lot of noise within that number.

A lot of it that's in there is around the Morrisons fee that we got when we exited their 5 modules out of Erith. It's about -- I think about GBP 17 million or so there.

And then there's about a GBP 15 million number in respect to the closure fees as well. So you'll see that detail later on in the pack in the appendix.

Other than that, contribution margin for tech solutions, an improving contribution margin of 72%. Of course, the revenue does benefit from those items that I mentioned, but we've also included a potential decommissioning provision in there in respect of those site closures.

So just to make sure that we balance it out that we haven't taken all of that benefit directly to contribution. There is some provision in there as well.

These are the expense items of the technology spend and then support costs. That's the G&A costs that exist, but it's also the partner-facing teams as well.

As you'll see shortly in the slide, it's sales team, but it's also G&A, corporate overhead type teams. Okay.

left-hand chart showing the progression of average live modules. Now clearly, as we approach '26 and '27, we are going to be hindered in '26 by, of course, the sites that have closed in -- recently that were part of that 121 number.

We make that up by the sites that go live. We also make it up by Ocado Retail drawing down on further modules as well, which they will do, self-evident given we're going to show you shortly where they are on their capacity, but as they're growing as strongly as they are another year of growth, 15% revenue growth.

And that's driven by volume as well. That's a volume-driven growth.

Better revenue per module, I talked about that, that progression, indexation playing a part there drives our recurring revenues. That's the math.

So here you go, go live of CFCs and drawdowns to drive the growth following the resets. Here on the right-hand side, we call out the CFCs we expect to go live over the next couple of years.

And again, there's some bullet points there that just reinforce my earlier comments around module drawdowns and the importance of those. We expect by fiscal '27 to be at at least 125, and we're targeting over 130.

But for the purposes of the modeling of cash flow positive, this range does the job. We expect we can go further than that, but let's get there when we do.

That's our ambition. Okay.

Direct operating costs, we expect those to benefit further going forward, but you can see the progression here, continued efficiency in the operations. We do think we can get to these being below 25% and therefore, an over 75% contribution margin.

Technology spend, you can see how it's declined in fiscal '25 to a total across CapEx and P&L costs of GBP 248 million. We will be seeing shortly in the guidance that we give for fiscal '27, again the cash flow positive.

We are looking at around GBP 100 million or so reduction in that spend over the next 2 years. We have made it very clear for quite a while that the last 5 or 6 years has been a peak investment period for us.

If you recall, we launched those reimagined innovations in January 2022, and now they're going into the market with our partners taking those, Ocado Retail in the U.K., Kroger in the U.S. and so on, ordering them and seeing the productivity gains that Ocado Retail is already benefiting from.

So it's a natural part of our evolution as we expected for this technology spend to start to wind down. SG&A costs will reduce further as we focus on a leaner operating model.

You'll see that in the mix there of SG&A costs that we've actually -- I know it's split in the first half, second half split, we'll show it a little later. We've actually put an extra GBP 6 million into partner-facing sales teams and G&A costs have come down by the same GBP 6 million.

So whilst it's flat year-on-year, the mix has changed quite significantly, and we'll continue in that direction. Again, that number, we're expecting there to be about GBP 50 million lighter in '27 versus where it is currently today in fiscal '25.

That gets you to your GBP 150 million of cost savings. Ocado Logistics, okay, 11% growth in revenue, orders, again, it's volume-driven growth, as you might expect.

So what would I call out on this slide, pretty reliable EBITDA number that we're generating. EBITDA has grown a little year-over-year.

A lot -- some of that, I should say, is down to TSAs that have rolled off that we provided to Ocado Retail that we're now charging for as those transition services agreements have concluded. So we're getting more profitability out of the business.

Eaches is growth of 8%, orders per week up 10% in line with revenue growth. UPH, again, we're going to show -- we're going to see a chart in a while that shows that UPH progression, really important part of the business model and the investment case for Ocado, which is the productivity our technology can bring to warehouses, drive down labor count and drive down labor costs where labor is becoming more expensive and more scarce, a really important part of our investment case.

And going through those metrics, there you go. You can see the progression in UPH.

Luton has at a peak of 318 UPH recently, and we averaged 289 in fiscal '29 -- fiscal '25, sorry. And then DP8, pretty steady at around the 21 number, 21.5 this year.

So just close to 22 drops per van per 8-hour shift. Ocado Retail.

Another really strong year for Ocado Retail, revenue growth at 15%, strong growth in our customers and growing our market share in the online grocery market, 15% revenue growth, gross profit up by 14%. Across these cost lines, you can see the operational leverage coming through as well, with the revenue growth of 15%, CFC costs up just 7%.

Service delivery, that's hit hard by U.K. labor inflation, but also by the national insurance changes that kicked in over the last year or so.

Utility costs flat year-on-year. Support costs, you can see growing by just 8% and marketing costs growing by just 3%.

So good operational leverage in the business model. What else?

I think I've pretty much highlighted the key things there. The underlying EBITDA margin, if you were to add back the GBP 33 million of Hatfield fees is now a 3.8% number.

We expect those Hatfield fees to reduce as Ocado Retail orders more modules. There is a credit system in place that as they place more -- order more modules, for every 3 ordered, roughly there's around a 2% reduction in the Hatfield fees number, which is a 13 module count for that business -- for that -- sorry, that warehouse.

Ocado Retail, structural and volume-led sales growth, orders per week, so it's a volume-driven growth, not a price-driven growth, 13% growth in orders, average basket value up slightly at 1.3%, and I think you'll be familiar with these trends. Okay.

Again, customer growth now comfortably over GBP 1.2 billion -- GBP 1.2 million, sorry, basket items stable and utilization. There's the chart, the important one for us.

We like that chart because it means they're going to be ordering more modules and CFC shortly. Watch this space on that one.

So our cash flow, when you put this all together, this is from an EBITDA basis down to a cash flow basis in fiscal '25, there's our EBITDA. The cash received into contract liabilities, we get -- that's the cash that actually came in.

We deduct then, of course, the amount that was recognized through the P&L account, which is part of the EBITDA number and because that's the noncash item that we take that out. The working capital movements we benefit from.

There's the interest payment line that I highlighted earlier, GBP 46 million increase in that cash outflow year-over-year. CapEx pretty much in line with last year.

CapEx is principally the CFCs, of course, the MHE, but also our technology spend. OAI CapEx, that's related to McKesson.

Lease liabilities, no movements there and nothing to write home about in the other line. And then finally, the proceeds from the letter of credit of GBP 113 million, getting you to that underlying cash flow of GBP 100 million.

Net cash inflows. When we take into account the final receipt from AutoStore, from that settlement that we concluded, agreed with them around 3 or 4 years -- 3 years ago now.

That's the final payment that's now come through. On our financing, we actually paid -- whilst we paid down less debt than the amount of debt that we've raised and benefiting with GBP 58 million on the balance sheet, which you see today.

Other items, you may have read that we've sold our stake in Paneltex, which was a very small GBP 400,000 investment, and we've got back comfortably our money back from the valuation of that sale, which is around GBP 20 million sale or so. And we own 25% of the business.

Okay. The outlook for the year ahead, for the next couple of years.

This goes back to my earlier comments. We have guided for quite a while that we'd be heading towards 20% of recurring revenues.

When you do the math, this is the GBP 250 million or so that we spent across CapEx and P&L spend in fiscal '25, declining by about GBP 100 million to GBP 150 million in fiscal '27. Tim will be talking more about that during his pack.

Other than that, I will then move on. CapEx has been weighted to the Re:Imagined fees, as you can imagine.

You can see the composition of the capital items in fiscal '25 in the pie chart at the top there. And then there's more commentary around the '26 and the '27 targets outlook.

The cost that we are taking out of the business of GBP 150 million in aggregate, that will be an exercise that is commencing now. The key events will be in March and then later in the year as well, there'll be another event.

We'll just have to see how that progresses over the course of the year. It will take place -- I should have said, it will conclude by the end of November.

There'll be progressive reductions over the course of the next 6 months because a better way of framing it. Lowering and leveraging our SG&A cost remains a key focus for us.

This GBP 50 million of costs come out of this area. You can see the trend has been a downwards trend.

You can see the shift there between the blue and the green bars around partner-facing teams and corporate functions. I think I've pretty much commented on this dynamic already.

There's the GBP 150 million, putting it all together, and just reinforcing that point. And summarizing the key building blocks to be cash flow positive for full year '27.

Live modules of GBP 125 million to GBP 130 million plus, I'll explain the drivers of those. That will generate a contribution of around GBP 400 million with a cash contribution of circa GBP 3 million per module.

Total tech spend and SG&A spend will be around GBP 250 million from around, you can see, the GBP 400 million in fiscal '25. That's the GBP 150 million saving.

And then we get a variety of other net inflows, including upfront receipts from partners. We typically get about GBP 34 million a year from -- GBP 30 million to GBP 40 million a year from those.

Logistics cash inflows, the business that generates cash for us, take off our lease costs and then whatever the movement is on working capital from year-to-year. Net interest costs are we've modeled GBP 80 million to GBP 100 million, but hopefully, there'll be opportunity there to reduce that given our strong liquidity.

And as I said, when we look at our debt stack and options to address some of those maturities, some of those -- some of that debt. We've concluded here in the final bullet, there will be sufficient cash flow to be cash flow positive and to fund circa 10% module growth.

So one could argue that if we did a 15% module growth in fiscal '28 versus '27, we wouldn't have the cash flow. We might not be cash flow positive because we're putting the CapEx in.

That's a high-quality problem, I think. We'll cross that bridge when we get there.

That would be the only reason why we wouldn't hit our target if we had some big orders for delivery in '28 and '29. But I think both Tim and I would actually welcome that.

Okay. Managing our debt maturities, and I've talked about this.

You can see here, by the way, the GBP 56 million convertible bond due December '25, we paid that off, if you recall, just after the year-end. The GBP 500 million has gone away.

This is the one we'll be targeting out of cash, the GBP 350 million. And there may be opportunities as we look through with our cash balances to look at those other 3 debt gross items that we've got as well.

So I think you can pretty comfortably expect our gross borrowings number to start to come down quite significantly, which will be good news. So summary guidance, tech solutions revenue of around GBP 500 million, and adjusted EBITDA margin of around 30%.

Logistics, more of the same, no great surprises in there. We're going to be having somewhat perversely a GBP 200 million outflow in the year that we turn cash flow positive.

A lot of that is driven around the timing of our cost reduction activities that I talked about, the sequencing over the year. You get the full year benefit in '27, but you then see a partial year benefit in fiscal '26, which explains a lot of this dynamic.

CapEx will be around GBP 250 million, and those are all the key numbers of our guidance. And that concludes it, there you go.

I'll just repeat those messages. Good management of our balance sheet.

I think we've been pretty good and proactive there on debt management, strong financial performance in '25, cost and capital discipline becoming a key theme for the organization. And again, the core priority is to turn positive during the second half, but it's backed up by a very robust plan as well.

Thank you very much. Tim, over to you.

Thank you, Stephen. Thanks, everyone, for joining us today.

All right. Let's move on.

Right. So we've had a good year in a number of ways.

So I think one of the key metrics is that first one is that's the international CFC volume growth of the 26%. So we just want to keep helping our clients to grow.

We're helping them to grow more and more, and we want to see that number continue to grow, and the compounding effect of that will lead to more and more drawdowns of modules that are available in existing CFCs as well as demand for future CFCs. Just again, to give you some idea of scale, we shipped 72 million orders across the whole of the OSP platform last year with a 98 -- more than a 98% fulfillment rate, 0.7% of average OSP waste.

And then we saw quite a lot of efficiency coming into the platform last year. We got across the platform to a 21 DP8 on a weighted average basis across all of our clients.

We saw an average of a 10% improvement in CFC productivity across our clients. And as Stephen mentioned before, in the financial year, we achieved 318 UPH in Luton and since the end of the financial year have got into the 320s.

The metrics keep getting better. And to put it into context that 318 UPH, that means that a 40-item order is fulfilled in less -- using less than 8 minutes of human endeavor compared to about 75 minutes to do the same thing in a manual operation picked in store.

So as you know, we've been busy working with our partners on partner success. That's the one area that Stephen outlined where we've been spending more money, improving our partners.

We thought we'd pick a couple of examples just to show you not just what the equipment is capable of doing, but how we help our partners and what kind of results we achieve. So this is one warehouse where we've been working closely with one of our partners.

Again, because of partner confidentiality, I won't get into exactly who they are, please don't ask me. But this is an international warehouse.

This is a combination of 2 things. This is a combination of new software and operational advice.

In this particular first example, this is DP8, so this is deliveries coming out of that warehouse. And we helped over an 8-month period to get a 34% improvement in the number of drops per shift.

To put it into context, actually, the top of that graph is 25 DP8. So actually, it's higher than the U.K.

The U.K. is not a standout performer on DP8.

We have achieved greater results in some of our international warehouses. And so this is not a question of somebody who is extremely poor becoming less extremely poor.

These are actually quite impressive numbers and are significantly ahead of our partners' expectations. If we choose a UPH example, here is a UPH example, it's a 5-month period that we went in and helped a partner inside their warehouse.

We improved their labor productivity by 1/3 in just 5 months. That is a combination of better operational processes.

So we're helping partners in their planning and in their operations as well as a rollout of some of an early rollout of some of the reimagined kit into that building. But -- so quite meaningful results in short periods of time.

We brought back in Lawrence Hene, who used to run a significant part of Ocado Retail for many years, and he's leading our partner success efforts, working alongside Nick de la Vega, who's come in to run our revenue sales and partner relationships. So significant progress in those areas.

In terms of CFCs driving growth, here are some comments, which I won't read them out. I'll let you read them yourselves from some of our newer partners, Alcampo, Auchan Polska and Coles, who all have opened warehouses recently, who are all seeing strong growth in their sites ahead of the wider online markets that they operate in, are achieving incredible NPS scores from their customers.

What we're seeing is if you take an existing geography where you have existing store-based operations and move them into a warehouse, you can see not only enormous pickups in NPS, customer satisfaction generally, but you can also see 30% to 50% growth beyond the market and beyond your baseline in a very short period of time from the better performance, better availability, better fulfillment, fresher goods that arrive from putting those volumes into the warehouses. I think, obviously, we've spoken a bit about the warehouses that have closed.

You need to put volume through warehouses. These are examples of retailers that have got some volume from store-based operations and putting that volume into warehouses makes enormous sense.

So historically, we built warehouses that were designed to largely do fulfillment from order today, deliver tomorrow. We've talked before, we talked at Re:Imagined about inventing new software that would enable these warehouses to be used for order today and deliver today.

We have been rolling out that software during the course of the year. We are still in the early stages of rolling out that software.

It is currently available for rollout with all of our partners, and we expect it to be in the vast majority of warehouses before the end of the year. It's currently deployed in 9 CFCs.

We've seen the earliest deliveries from order to delivery of 73 minutes. Now 73 minutes is not an impressive number for speed of delivery of an online grocery order.

You can do that in 10, 15, 20 minutes, but from micro sites with 1,000, 2,000, 3,000 SKUs in them, with efficiency levels that are really, really poor. This is an order processed in a large-scale Ocado CFC with extremely high productivity, as we spoke about before, with range of 200, 30,000, 40,000, 50,000 SKUs available to those customers.

And it is not costing anything in productivity to achieve that. And that order is being delivered in a scheduled 8-hour route, but we are able to get the last from the customer ordering it to delivering it to as little as 73 minutes for a full basket order.

We have seen in the first warehouse we rolled this out in days where we're achieving 40% same-day deliveries in an international CFC. We think this is a game changer, and we look forward to the rollout of this across the rest of the network and our partners continuing to work with it and increasing the amount of capacity that they have for same-day ordering, which largely addresses that large shopper universe of people who find -- want to shop online, but find it hard to plan where they'll be able to take advantage of the big ranges, the hypermarket prices and same-day delivery.

We also spoke at the half year about aggregators. So we have now integrated aggregators into our platform for the first time in the past year, enabling customers who order groceries with our partners, but from wider platforms.

So from aggregator platforms, those orders are going through and then those orders are processed either in Ocado CFCs or using Ocado in-store picking software in the client stores, making significant efficiency compared to having multiple apps and multiple pickers in those stores. And these changes really reflect the evolution of the online grocery market where in some markets, significant amounts of volume are going to aggregators who don't process orders themselves or don't have stores or warehouses themselves.

But now our platform is flexible and those orders can get pushed through it. It has enabled Morrisons to increase their aggregator coverage in the U.K.

to a further 100 geographies. It's enabled Monoprix to roll out to 22 further cities in France with one of the global aggregators that they work with.

So let's just talk a little bit more about the evolution of the platform. If we went back to 2018, there's a little graphic here that described largely the platform that we sold to our early partners.

We had a largely next-day service. Partners are expected to operate OSP web shops and take 100% of the orders across OSP, web and mobile.

And they were processing those largely in warehouses at an average of a 6-module size for home delivery in vans, either directly or via spoke sites. It was a narrow but successful approach to the market that has served us extremely well here in the U.K.

for a number of years. But the market has changed and the market continues to evolve.

And today, shoppers expect to shop online with total flexibility across different platforms, lead times and shopping missions. And retailers need and want to meet those expectations without incurring the high costs associated with the traditional fulfillment.

So where are we today? Today, our platform supports all different shopper lead times from sub 1 hour, 1 to 6 hours, remaining same-day and next-day deliveries.

We support bringing orders into our platform through managed fulfillment where the clients run their own front ends through the OSP web shop that continues to evolve and deliver a market-leading experience as well as mentioned before, through aggregator sites as well. We can process those in in-store fulfillment over 1,000 stores live through our new store-based automation that can range from 4,000 to 5,000 square feet attached to a store up to about 17,000 square foot potentially unattached to a store.

In 2 to 10 module sites, for large scheduled delivery businesses in micro fulfillment centers, as I said before, from about 4,000 to 17,000 square foot that do not need to be attached to a store if they don't want to, as well as in manual warehouses or third-party DCs, serving all the different customer missions and all of them with the best economics. And then in the last mile space, we're working today delivering customers to order -- delivering orders to customers using couriers, lockers, customers collecting it, home delivery and home delivery via spoke.

It's an incredible amount of total and evolution of the platform to total flexibility for our existing global partners and future global partners. It is the product of a very large and busy R&D period for us as a business.

But we've now deployed most of this evolved platform for our partners worldwide with strong results. And with our exclusivity rolling off in multiple markets, we're focused on bringing these benefits back to some of the world's most mature grocery markets for the first time in years.

As we move into this new commercial phase, we're also taking steps to realign our business to better serve our global customer base and focus on new prospects opportunities with the biggest value. But I wanted to start by reflecting on the scale of some of our commercial footprint today as I think it's sometimes underestimated.

Most of you are aware of our global grocery partnerships worldwide. They remain the core revenue driver for our technology solutions business and the partnerships where our technology is most fully deployed.

However, our commercial footprint does extend more broadly into the wider logistics, CPG and retail sectors, primarily driven by our growing AMR business. Today, our technology is live in 127 warehouses and more than 1,000 stores worldwide with 70 commercial clients and partners.

We've got more than 17,000 bots live on grids around the world as well as 431 on-grid picking arms, that number growing probably daily, more than 2,500 truck AMRs, and we're seeing keen interest in initial orders for our new case handling product, Porter case handling AMR. So as we move into a new commercial phase at Ocado, we're building on strong widespread relationships with many recognized and leading worldwide brands.

We're also making changes to the structure of our technology solutions segment. Stephen has already talked through our progress towards reaching our steady-state cost base that we flagged over the last few years.

We've made significant strides towards our full year '27 targets over the course of the last year, and we continue to track towards those targets as we move out of this peak development cycle and into a steady-state R&D phase. The structural changes that we're making support these goals and make sure our business is properly geared towards our priorities, namely a renewed and focused go-to-market strategy, a simpler operating model, investment concentrated where we see the clearest path to value creation.

One of the first key steps is the consolidation of our commercial divisions, meaning Ocado Solutions and Ocado Intelligent Automation are now operating as a single point of sales and account management under the new leadership of Nick de la Vega, who joined us as Chief Revenue Officer at the end of last year. This change also reflects an overall shift in our approach to new commercial opportunities with a more targeted approach to the most valuable opportunities and primarily within sectors where our expertise is most needed.

Taking the example here, which shows in the blue areas of the grocery supply chain where our technology has been traditionally deployed in the CFCs and delivery to homes. One of the key lessons we've learned from the market engagement of OIA has been that there's significant opportunity to go further up the grocery supply chain and the CPG supply chain.

We see significant opportunities to expand into those areas, both case replenishment for stores and wider distribution networks, both where AMR products like Chuck and the new Porter solution can bring significant capital-light productivity improvements. Our AMR products are already deployed in upstream CPG supply chain environments, and we see a positive opportunity to build on this business supported by a single, more simple commercial structure.

We believe that opportunities like this will bring significant added value and optionality to our core OSP business, enabling us to grow an attractive new revenue stream in our tech solutions business alongside our core automation and fulfillment assets. Our solutions are very deployable in the case pick market for store replenishment.

We highlighted this next piece in the half year, but I think it's really important that we continue to focus on it, which is about bringing the right fulfillment for the right market. To be successful today, retailers need to do careful network planning to make sure they deploy the right solutions in the right places at the right time in their development of their e-commerce journey.

But we have a full toolkit to address those different opportunities. It's a framework for future growth, and it underlines some of the decisions taken in recent months.

We can do everything from low-density solutions where you use manual pick in store with world-leading efficiency using our software. We can do manual pick in dark stores.

As we mentioned, store-based automation before, we're going to focus on that in a moment, micro fulfillment centers as well as the large automated CFCs. The critical lesson that we've learned is that you do not buy a large-scale CFC unless you have a business to put through it.

They are not a profitable asset if you don't use them. But if you do use them, they're great.

So moving on with a little bit of focus on a CFC to start with. So a CFC can range from about GBP 150 million of annualized sales to over GBP 500 million in capacity.

GBP 500 million is approximately what we refer to as a 6-module CFC. So if we take a 6-module CFC as an online case study, it can do about GBP 480 million of annualized sales in a standard sales pattern with kind of similar metrics to an Ocado Retail business.

Today, to build a 6 module CFC requires both upfront fees and retailer CapEx to put in things like fridges beyond a standard developer spec shed of about GBP 50 million of investment. The benefit of running that at GBP 480 million of sales compared to doing this in store is somewhere in the region of GBP 30 million to GBP 40 million a year, meaning it is a 1- to 2-year payoff asset.

These buildings are amazing if you can use them. And that really is the key lesson.

Some of the buildings in North America were not being used and those retailers working with us have made the decision to close them, where these buildings can be used when you have an existing business or you can rapidly grow into these buildings, they are significant improvements for the same volume going through them. As I mentioned before, they deliver a far superior customer service, driving up significant NPS, resulting in significant uplifts as well in sales.

So if you go into a building where you've got, say, recent examples, we built some 3 module sites. We've launched a 3 module site recently.

We've got another one in build at the moment. If you've got 1 or 1.5 modules of business from your store pick to go into that facility, by the time you go into it and see the uplift and you've got strong growth, you're in a very good position to drive to full capacity and see significantly quick retailer cash payback.

These have been -- these principles have been a key in the engagement with all partners at the moment and are reflected in those new CFCs that we're building. And this kind of thing is reflected as well in some of the CFCs that have opened with Coles, for example, putting significant volume into their new CFCs in Melbourne and Sydney in the year that they've opened.

And you saw the positive comments from Leah just before. If we move on now to the other end of the spectrum, which is store-based automation.

Here's a nice little visual of our store-based automation sites with the external pickup ports, the same on grid -- the same robots operating on grid, the same on-grid robotic picking. The one new piece being the external ports, a small development that we are engaged in at the moment.

Store-based automation is a phenomenal product if you have a lot of customer pickup direct from store because you can process these orders really quickly and really efficiently compared to in-store. It's a phenomenal product if you have a lot of gig-based direct-to-customer deliveries from drivers picking up 1, 2 or 3 orders.

They can also interact from those ports and those products can also be picked incredibly quickly from order to delivery. It is even more important if you're doing ultra short lead time delivery because when you pick ultra short lead time orders in a grocery store, the effective pick rate drops by about 70% because you're no longer able to batch and pick in the zones.

You have to run around and pick a smaller order across the whole geography of a store for a single customer. And the pickup to doing it in our machine will be comfortably into the double -- a number above double-digit percentages in terms of efficiency improvement.

Now if you take some markets, if we take a market like the U.S., for example, 8 years ago, the average store was doing $1 million to $2 million of sales. These machines wouldn't work sensibly to do $1 million to $2 million of sales, but markets have grown dramatically.

And so here's an example of the sites that we've been talking to retailers about in the last few weeks that we've been able to speak to retailers in the North American market. And we're seeing significant enthusiasm for this product across every conversation that we've had with retailers in that market.

Typically, sales in store could be anywhere from $5 million to $40 million online. I know the $40 million sounds like a large number.

There are people who are interested in sites of those size. That's also a particularly relevant size for the French market, which is a 90-plus percent pickup market.

But if we take a case study of a store with $12 million of sales, that's a fairly common size. That's 10% to 15% of a store that does $80 million to $120 million of store-based sales, now doing 10% to 15% online.

The full retailer upfront fees to us and CapEx would amount to about $2 million upfront, we estimate, with a greater than $1 million a year operational improvement. This ignores the improvements in NPS.

This ignores the benefits of increased capacity, which is suffering in a number of the larger stores and busier stores in these retailer networks. This is just pure labor savings, predominantly labor savings in these facilities, meaning that retailer cash paybacks of 2 years are quite possible in this space across all of the stores, across markets that are doing $8 million, $10 million, $12 million, $15 million, $20 million, $25 million, which in many markets now is the kind of average.

If we look at the U.S., for example, when we entered into our exclusivity arrangements with Kroger 8 years ago, the U.S. grocery market was $30 billion in size.

Next year, we're not ready to roll out store-based automation in mass scale at the moment. We're looking to do a couple -- a few handfuls of sites at the moment to prove all the different points around the costs and the execution.

But by 2027, when we would look to roll out in scale, the U.S. grocery market is estimated to be 8x the size it was in 2018.

This is why when people rolled out what they believed were micro sites 8 years ago, they tried to roll out one site to cover 5 to 10 supermarkets worth of volume. It created incredible complexity that doesn't exist today because today, each of those sites now needs $8 million, $10 million, $12 million e-com sales from the singular site.

But also the difference today is that we can build these things in a fraction of the space that was being used with a fraction of the labor that was being used because of advances like our incredible AI-powered pixels to action on-grid robotic pick, which today is doing more than 50% of the picks in Luton across a 45,000 SKU range. Globally, since we entered into a number of exclusivity arrangements, the global market has more than doubled.

And so we are super excited about reentering a number of markets where we're having some very interesting conversations with a large number of grocers and keeping Nick in his new role very busy. So in summary, we're reentering markets, with a tech solutions business with a simpler operating model, with a focused commercial operations and a strong R&D base.

We're seeing strong interest in a massively evolved solution set with massively more flexibility, a wider fulfillment tool set and world-leading shopper outcomes. Our partners are seeing robust underlying growth, strong year-on-year improvements in operational performance, and we have learned important lessons.

We now have stronger foundations of our key partnerships and clear pathways to deliver disciplined, sustainable growth worldwide. And on that, we'll start taking questions.

Tintin knows I can't handle as many as 2 at once because I forget what they were.

Absolutely. Tintin Stormont from Deutsche Numis.

Two questions. If we look at that graphic that you showed, the manual pick in store, manual pick in dark stores, and you look at the level of activity in terms of pipeline and trying to speak to customers, where is it sort of kind of busiest?

Where is all the activity happening? And Stephen, for you, for those types of potentially new sales, how should we think about the revenue models?

Is it more upfront fees? Or are we still thinking about the recurring fee as a percent of the capacity?

That was actually one question. The second question was just a modeling one on Hatfield fees just in terms of how do we anticipate that GBP 33 million to come down over the next couple of years to 2027.

Let me just try, and I might answer yours as well.

Go ahead I thought you might.

The first initial interest from retailers in SBA is actually around their biggest sites, their busiest sites, where a number of retailers have maxed out capacity. So the kind of first focus is, oh, wow, can you do something that gives me more capacity in those locations?

The brilliant thing about that is if you do that, it's going to simultaneously show them how much economically better they are and what better experience they deliver to shoppers. So when you look at the estate, there's an amount of the estate that is just something needed because they're maxing out capacity.

And then there's the vast majority of the estate where once you realize what these things are capable of doing, you'll realize you've got a 2-year payoff. But most retailers won't turn down a product with a 2-year payoff that also gives them increased customer shopper outcomes and increased capacity.

So there's a lot of interest. Markets are evolving and growing fast.

And the more it moves to the shorter lead times, the more attractive the product is versus the manual alternative. I've tried to explain that before with the pick speeds.

Globally, 180, 200 type pick speeds, if you're aggregating orders and segmenting pick walks and stuff like that, those drop to like 60 if you're running around frantically trying to get something ready for a courier in 5 minutes. In our store-based automation products, those will be picked over 1,000 -- a human pick endeavor will be over 1,000 UPH because the humans will be doing half over 500, okay?

So just massive increases in efficiency. In terms of the fees on those sites, largely, we don't expect to have any significant outlay if we roll out that product.

So the upfront fees should cover the majority of our investment into those sites, meaning that for a retailer, they're likely to have as a percentage of sales, a higher upfront outlay. But as it's a much more phased because you roll out store by store where you need it.

So as a function where you need it, as we showed before, it's got a 2-year payback. So attractive on both sides.

Our ongoing fees are likely to be slightly lower than our ongoing fees on the OSP product because we're not amortizing and financing as much equipment. And we'd aim to make a similar percentage of sales contribution to our R&D, SG&A profitability.

The Hatfield fees have got a split that's about -- that's just about 60-40. So 60% of those fees will amortize over time as the equivalent amount of volume is taken down in new sites and 40% of those fees will remain until the end of the Hatfield lease.

Marcus Diebel from JPMorgan. Maybe just on the rollouts and ramps for your partners.

We've seen some delays in Korea and Japan. Can you just talk a little bit more sort of like what's going on?

Obviously, you don't want to like split hairs, but obviously, we had delays with Kroger and then a different outcome than we maybe thought. So if you just can tell us a bit more what actually happens there.

Yes, that's the first question.

So look, they're slightly different scenarios. One is Japan.

Japan, we're opening 3 warehouses in a contiguous geography. So in fact, so long as there are a sufficient number of live modules, the exact timing of when the third site goes live is not hugely important because the volume is being done in site 1 and site 2.

The site 3 isn't needed from a volume basis on that date, but it's about our clients building programs and about the time lines that they give us. We can get in and build very quickly.

I think we speak about the fact that we built a warehouse from scratch and went live in 12 months this year from a greenfield site. So it's really just about when those handovers to us come.

Sometimes those programs are set up and for whatever reason it is, it can be an internal reason that our client or it can be an external reason to do with zoning or commissioning or something like that. Those projects sometimes, we can't be sure at this point exactly when they're going to go live.

In Korea, it's actually the first site is in Busan. The second site is in Seoul.

Seoul is a bigger, more developed market. So we'd like to see that site live as early as possible, but we think it's a chance it's going to roll into next year.

So we'd rather be transparent about that. We are with the client in store pick, and we are now seeing good growth on that platform and excited about the first launch later this year in Busan.

Yes. And the second question is just on sort of like we talked about it before.

And what is your sense in regards to the sort of like urgency at your client base because we live in times of the technology evolves quickly, both software and hardware. So why is it now really the time to go the next step or to wait?

I think previously, you commented on the analogy of an iPhone, at some point, you just have to buy it. But I obviously hear this a lot more at stage.

So what is sort of like the kind of like situation where clients are in?

Look, I think with most of our clients, they say they're seeing significant online growth. We saw 26% through the sites.

And capacity is an issue at places outperforming their competitors in terms of shopper experience and efficiency and cost structures. And we keep coming back to the same kind of point.

If you've got GBP 50 million of business and it's scheduled delivery and it's spread across a 3-hour catchment area, the best way to do that is in-store, right? There isn't an automated product that will help you to do that well.

If you've got GBP 150 million of business or GBP 500 million of business in a 3-hour catchment, then the rightsized warehouse is the best way to do that. And the warehouses we would build today are half the size and 75% more productive than one we would have built 3 years ago, right?

And so they're an ever-increasing attractive option. If you're doing -- if you've got a wide geography with not enough density to do that or you've got more immediacy business and pickup business, if you've got $2 million, $3 million at a single site, don't build store-based automation.

It's not going to have a return yet. If you've got $5 million to $8 million annual growing, it's time to start considering building it.

If it works as well as we expect it to work and can be built in the size and at the price we expect it to, that mean the economics work well for us and work well for -- even better for our partners. That's what we have to show in the next 12 to 18 months.

But it is a question of people aren't wanting to invest in big J curves at this point. People aren't wanting to speculatively say, I think there's going to be a giant market in this place.

I have nothing. I'm going to go and build a $0.5 billion capacity facility in this small city.

When we're talking about something like, as you mentioned before, in Japan, this is not a small city. We're building facilities in Tokyo, okay, which is something like 40 million people living in the geography of those cities.

So that's just an enormous market. But the likes of Calgary, people aren't going to speculatively build a 6-module site in the Calgary going forward unless they're already doing 3 or 4 modules worth of business in their store pick operations.

Sarah Roberts from Barclays. So just my first question, the Kroger and Sobeys sites that were closed seem to be in the underperforming category of CFCs that I think you've spoken about before.

Just wanted to understand across the wider network, not having to give any details on specific partners, but how many CFCs are still in that kind of underperforming bucket? Or have those now moved to a level where you're both happy with the utilization.

Just wanted to get a sense of any potential further downside across the existing network.

I would say it's very limited downside at this point. I don't want to say there's 0, but there's very limited downside at this point.

Okay. Helpful.

And then on the side of store and automation side, obviously, it's a little bit more of a competitive market versus potentially the full CFC model where you are the only player that can do that level of automation. So I just wanted to understand how you're seeing yourself positioned in the kind of micro fulfillment area, a bit of the warehouse automation market, why you deserve to win and how you're thinking of playing there, that would be really helpful.

Absolutely. Look, I think the key things to make these work and what hasn't been understood before and where people have failed when they've rolled some out and not had the success they wanted for clients is based on a few things.

One is just actually their understanding of handling grocery and the variation in grocery, the interaction with stores. And there, we obviously are in 1,000 stores today as well as delivering whatever, we think, we said 72 million orders last year across our platform.

So we've got enormous knowledge that a number of these players just have got next to none. When we see people bragging that in the last 8 years, they've had this much volume go through a platform.

And when we look at it, we've done that volume in the last 1.5 weeks, right? So we've got a depth of knowledge, number one.

Number two is retailers want to do this in the smallest possible sites. And cubic storage is the densest storage for the products that you need to store, number one.

And we have the solution, and we've spoken about this before, that can get the highest throughput from a square meter of grid, a square meter of processing because our bots are faster, accelerate faster, deaccelerate faster and our control algorithms allow them to work in greater density, and our single space bot patent means that nobody else can achieve the density that we can achieve. So in terms of using the least amount of space to generate the highest potential throughput, we are in the best place.

That is significantly more relevant when you're trying to carve out a corner of a busy store in a city than that is when you're trying to put something in a massive warehouse outside of city. Our on-grid robotic pick is completely unique and one of the most advanced cases of AI being used in the physical world today.

And by having at least 50%, we're at 50% already in sites in Luton, for example, by having more than 50% picked on the grid at the top of the grid, which is sharing the same air space as our moving robots, it's an immensely complex technical challenge. It removes the need to put human pick stations downstairs or to put robotic pick stations downstairs, which then take up twice as much space as the human ones did.

But by removing that space, we're able to build these incredibly dense sites. So where we have spoken to retailers who have recently built or have been exploring building alternative sites to address the challenge of capacity, we are capable of building similar capacities that they've been talking to in less than half the space and with significantly higher range capabilities.

And so we cannot see anything that has the capabilities that we have in this space.

It's James Lockyer from Peel Hunt. Two questions as well.

First one, last time you spoke about how you managed to get Detroit up capacity by 50% because of some upgrades you've been doing. I think at the time, you said you think the entire estate could benefit from 30% over time.

It would be good to understand how that's been going during the period and how you're seeing that benefit your own CapEx and OpEx efficiencies.

So James, yes, look, we are over the design capacity in Detroit. We are over the design capacity in all of the -- everywhere other than Erith, in Dordon, the 2 oldest CFCs in the U.K., all the other ones are operating above design capacity.

Continue to believe that we'll see at least the numbers that we outlined to you before. We are starting to achieve those.

They are baked in as part of the plan for U.K. expansion over the next couple of years.

And it's kind of what does it mean to a new warehouse? It's part of why a new warehouse can be only 50% of the size of what we built when we built those warehouses.

So we're not able to get the full 100% uplift that you theoretically could today if you took the building again and built into it, but we are looking close to 50% in most of those sites in terms of their incremental capacity that they're going to be able to achieve. And the enhanced productivity, which is separately beneficial in terms of reduced labor also then means that if you can pick half of it with robots and you are picking 50% more, you are actually using less pickers than your original design, which means the outside of the building in the car park still works for you.

You need to save some of those spaces for the extra drivers that you're going to have and now driving that increased volume. So there's a lot of considerations in making those changes.

The one site that's most complicated in for us in the U.K. is Luton because we've got a third-party automated freezer in it.

And that's the expensive part of building. Otherwise, it's very capital efficient for us and for Ocado Retail to achieve its next step of growth.

And going forward for clients, their CapEx in a building that's half the size is materially lower, their ongoing rent rates and services are materially lower. The availability of sites is materially easier closer to customers when you start to be able to build these things in smaller buildings.

Our space efficiency versus some others is just extraordinary. So we were talking to a retailer recently, and they said they weren't interested in big warehouses, and we said, no, of course, in their market, it wasn't relevant, store-based automation.

I said, yes, that is what we're interested in, and we're looking at some sites already. And then they gave us the size that they were looking at.

That to us was a warehouse. To them, it was store-based automation, but it was between 50,000 and 100,000 square foot.

And for the volumes that they wanted, we'd have been looking at 10,000 to 15,000 square foot. So we're space efficient.

And then the second question was on the case study you gave around, I think it was the 25 DP8 on there, which is significantly better than the U.K. you mentioned at the time.

Why do you think the reasons are specifically that, that was better? Was it density?

Was it urgency from the clients? Was it just more savvy users?

Why do you think it was better than the U.K.?

We have some U.K. sites that we operate out of that are between 25 and 30 today.

So it's like if you carve out -- that particular site has a geography that more reflects some of the sites that we've got in the U.K. as it's got a small radius around a warehouse doing a lot of volume, doing the full volume of the warehouse, you can do turnaround routes.

So one of the challenges is you fill the van up, let's say, in the U.K., you can fill a van with 22, 23, 24 orders, there's -- it's then hard to go above that. So where we do 28 or so, it's because we have vans that are not working a single 8-hour shift on one route.

So you have to do things like having 6 -- so in the U.K. in one of our sites, we do a lot of 6-hour and 10-hour routes.

So we do -- the drivers alternate 3 of 1 and 2 of the other each other week. They do 38 hours, 42 hours, 38 hours rather than 40, 40, 40, and then we can offload a whole van in 6 hours or almost 2 vans in 10 hours.

In some geographies, that's hard because the stem times are not there. And we are working on some longer-term quite complex and clever solutions to that to enable us to break through those numbers.

But basket size and proximity and density and things like that come into it.

Yes. Giles Thorne from Jefferies.

Tim, there's been a lot of public discussion about Ocado and Kroger and Sobeys. And to my reading, a lot of it has ultimately been quite gentle on Ocado and most of the blame that people are looking to apportion blame has been put at the feet of Sobeys and Kroger.

But nonetheless, it would be useful to hear your reflections with hindsight on things that you could have done differently that would have led to a more orderly outcome or a different outcome. And then the second question -- go ahead, Tim.

No, no, go ahead.

Second question is on -- I'd like to hear you talk about some of the compromises you've had to make on your innovation pipeline as a result of the cost efficiency program. Are there any bells and whistles that you wanted to build that have been put back up on the shelf and we'll have to wait for another time?

Sure. Look, on the Kroger and Sobeys one, would I like to have built the warehouses that Sobeys wants to build in a different sequence so that instead of the third warehouse being in Calgary, would -- should I try to influence management to build it somewhere with a bigger population opportunity with more density that they already had in their network or something?

Did we just accept the orders? Yes.

Is that in hindsight, a bit naive? Yes.

Could we have -- we know this because we've been working on it for the last 8 years, but could we build warehouses where a client could build a 3-module warehouse where a client could turn it on with 1 module and where we're -- economically that is a good warehouse for us even if that client never grows beyond 1 module? We're there today.

We weren't there 8 years ago. So 8 years ago, we built 4s and 6s and 8s and 10s or 7s and 10s.

And we insisted on clients opening them with 3 modules or 4 modules, which economically, we needed them to do because of the CapEx that we have put into those sites, and we even needed them to grow. But it was a big outlay.

It was a big ongoing cost for the clients if they didn't fill them. We didn't have a strong enough sense they weren't going to fill them, and they haven't filled them.

And so hence, they're a drag and a burden. Today, we've got 3 module sites going live, as I say, with 1 module down where the client is already doing 1 module before we put the spade in the ground in their store pick operations, expect to be at, let's say, 1.5 modules the day the site goes live, and we're allowing them to grow them in quarter module increments as opposed to having to grow them in 1 module increments.

So we've been aware of the challenges of our early business model, and we've been working on that for the last 8 years. We obviously still have to live with the consequences of those early sites and those early decisions.

So we need something that is economic for us and our partners at smaller size. We have it today.

We need something that our partners can start with the smallest volume and the lowest kind of fixed outlays, and then can they grow it? And we're talking to clients about doing this in a more flexible way in terms of their charging and how that works with their growth that are massively useful for them and still work for us.

So we're learning these lessons. Ideally, we might have built 2 warehouses -- might have signed 2 warehouses in 2018 and 3 warehouses in 2019.

If we could have built the warehouses that we've got in a slightly more linear way rather than kind of pushed upfront, then maybe we could have learned some of these lessons earlier and helped our clients. And maybe if some of those 7 module warehouses where they were paying us for 3 modules had only been a 3-module warehouse and they've been paying us for 1, maybe there would have been a growth path to see that filling up and those would still be open.

We're not blameless as such. But again, it's not our -- there are -- our partners are the ones that need to drive the acquisition of -- we can help them, and we are helping a number of our partners to understand how to best do online marketing, how to best trade an online business, but we need our partners having made a commitment to a site to try and work very hard to put volume into that site.

And what does Nick bring that you didn't have before?

Nick's got a huge depth of experience working with technology partners, accounts, clients, just kind of a much greater focus than we've ever had in terms of experience of doing large complex technology implementations where how to work with those clients to influence them to create the behavior that means we're both successful. I think one of the kind of combined naiveites between ourselves and our partners would have been kind of their view of we bought this amazing stuff, and if we just turn it on, we'll have a successful business.

And obviously, we've been saying for a while, that's not the case, but we still don't have that element of control where even where we realize it, some of our partners have still behaved a bit like that. And we're kind of like, he's very good at getting in there and talking through that challenge and trying to get those retailers to realize what they need to do to contribute towards making their business successful and not just a view that because we've written some clever code and we've got some [ dwizzy ] robots, that means they've got a business.

And so then on the innovation puts and takes.

Look, we have that combo at the moment of some of the biggest projects that we did rolling off, reimagined and the re-platforming, which you will have noticed as -- well, I don't know you personally, but a number of you will have noticed as U.K. customers, the kind of the move -- the migration to OSP.

That e-com migration and mobile app migration that happened in the summer at Ocado Retail was Ocado Retail moving the last part of its business on to OSP. They were the last of the 13 partners, Morrisons in the U.K.

had migrated before. So that kind of catch-up of all the tiny bits of things that drive acquisition, retention, frequency, margin, basket, et cetera, are all in OSP, and it's an onward journey from here, as well as having got live in 11 markets with payments and currencies and regulations and laws and all that kind of stuff.

Together with the rollout of Re:Imagined and efficiencies that are coming through in -- with AI in terms of not just coding, but testing and various things that mean that our overall productivity per person is significantly improving. What's not on the list is hard to say, but it's the more speculative stuff is not on the list.

The core things that we want to do to deliver store-based automation, to deliver the growth in existing facilities in the U.K., in particular, to do with supply chain, and we talked about this at Re:Imagined called Orbit, continued work on helping our clients attract and retain and make it easier for shoppers to shop, continued attention on making it easier to run the platform for our partners and the rollout of short lead time orders. All these things are still in there, right?

Everything that we really, really need is in there. We are going a little bit slower on some of the other opportunities to deploy our kit in other logistics supply chain scenarios.

We have got continued spend to enable it to move up the grocery supply chain, but we could go faster on some of those points. And we may choose to, in the future, if we start to do some significant contract wins in those areas.

So there are some things that are like show and tell, like we're doing those a little bit slower, where we -- maybe if we did them a little bit faster, we could then show something and maybe win some business. But it's a very tough process.

We're being very rigorous on it. But we expect to get a significant amount of innovation in terms of features and functionality coming, and probably '27 will be a record year for us in terms of innovation.

'26, we've got some -- I wouldn't -- I think it would be naive to say we're going to have a record year in '26 because of the disruption of actually going through the reduction in size, but I expect by '27, we'll be back at a record level of innovation. And the amount of change in the platform that we have achieved with this -- with all these amazing people in the last few years is incredible.

And I tried to outline that in that slide before, the flexibility of the platform. It's not just the flexibility, it's the intelligence of the platform and so many different things it can do today.

We have done so much innovation in that time period. It's amazing.

William Woods from Bernstein. I suppose the first question, historically, you've been quite cautious on the ROI and the ROC from smaller sites.

And I don't think that was just capacity. I think that was operational complexities around the ability to store certain levels of SKU breadth and depth, duplicative picking, decamp complexities, all that kind of stuff.

Have you worked out those operational issues? And if it does work, why haven't you built a 1 module site, for example?

And I'll come to my second question in a second.

The first one is what we can build today compared to what anybody could build 5 years ago is dramatically different, okay? The lighter weight bots, the third-generation lightweight bot can be deployed in a different way to the older heavier bots in terms of safety and crash barriers and stuff.

And then the space savings that might be 1 or 2 grid spaces in a huge site is not massively relevant when you make it a tiny site is massively relevant. The weight going through the grids as a result of the bots and that weight accelerating at the same speed and therefore, the forces that need to be offset and what you need to do in the floor space is relevant.

So if we built these 3 or 4 years ago, we need to start piling underneath supermarket floors and stuff. So there's just differences like that.

The on-grid robotic pick taking up at least 50% today and moving towards 70% or 80% of the picking, again, simplifies the whole process. The remote monitoring and operations of our grids and our on-grid robotic pick, which we centralized into facilities in Bulgaria, in Mexico and in Manila, mean that we can run multiple sites, the reliability of the robots and therefore, not needing to have like live on-site engineers at every site permanently.

Just there's a whole host of these reasons why this is viable now and we couldn't have built with our infrastructure 5 years ago. Now the people that did try and build things with different infrastructure 5 years ago just didn't have the process knowledge, didn't have the automation with the right throughputs, too much capital, too much labor, too much space.

And they didn't succeed at what people wanted. But I think it's important to understand 2 things have changed.

That's the supply side has changed, what we're able to do. But if you think back because at one point, we were going to get killed by a company -- because there's been loads of companies over the years, everybody said we're going to get killed by.

And one of them at one point was a company called TakeOff, and TakeOff was the micro sites company, and they went around to the person that we didn't sell our OSP to and sold every one of their competitors, 1, 2 or 3 of those sites and said they were going to sell them each 1,000. They did sell them 1, 2 or 3.

They never sold them the 1,000 and they eventually went bankrupt. The sites were, as I say, from a supply side, they were too big, they weren't efficient enough and the process flows just weren't good enough.

The output wasn't strong enough, et cetera. But the demand side was different, too, because the demand side at the time was to make this site viable, you need a $10 million site or $20 million site and you're doing $1 million or $2 million a store.

So they were like a mini where -- they were trying to do a mini version of our centralization where you've got -- so they kind of didn't have the benefits of our centralization, and they didn't have the benefits of being where they needed to be because they were only actually where one store was, whereas because the U.S. as a market has grown -- will have grown by the next year ninefold in that time, those $1 million to $2 million sites are now $9 million to $18 million.

And so now you can have one of these things at each location. So the offsetting disadvantage of taking a $600 million site and splitting it into 60 can be offset by the advantages of being in that local geography, leveraging existing assets of that retailer and being able to do pickup in 30 minutes or in 5 minutes in store and being able to do ultra short lead time deliveries and working with the gig economy drivers that for a whole variety of reasons are significantly cheaper in the U.S.

than a unionized labor force driving your own vehicles. Does that --

Do you not think there's an issue with kind of holding a certain number of SKUs? Could you hold the whole SKU range of a store and...

Yes. So this is your second part of question.

We probably ought to take this offline if you want to. But we have a lot of experience of moving product and understanding how this can work.

We've come up with some very good ways of doing this where the majority of the -- significant majority of the velocity will be in the automation. There will be some stuff that is not in the automation, but we will not be -- but our automation allows you to do a robotic merge.

So you're not doing one of these things where you've got some pick from here, some pick from there, some pick from there and then humans need to try and marry that up and then deliver it somehow to a customer because the whole thing will be merged by robotics in our machine. But also we do carry already multi-SKU totes in our machine, so we can carry extensive ranges.

We can replenish those ranges alongside the store replenishment for things of volume. We can replenish those ranges through batch picking in the store, but not for the specific customer, like keeping a SKU of -- a single item of each SKU in the automation, but not through an optimized pick walk on a batch basis, and we can merge in the rotisserie chicken and the sushi at the point of handing it over to the customer.

So we are working through all of the challenges that we are well aware that have been encountered in this space. Today, people want to look at merging prescriptions from other sources, merging general merchandise.

Our grid and some of our patents around our grid process, ones that we didn't license to our competitors in the cross-licensing are very important here and enable us to do things that drive, we think, unparalleled efficiency. We need to deliver it all.

There's nothing that we're trying to deliver that we see as rocket science, as in on-grid robotic pick is rocket science. Obviously, it's not Starlink or SpaceX or whatever, but it's very, very, very complex.

We've done the heavy lifting because we have that technology. There's just stuff we need to do around building up those processes, leveraging other things that we do.

Like we already do store pick, so we know about optimized store pick. We just need to bring some of those bits together in the next year, build the first few prototypes for our clients, hope and believe that they'll be successful as we want them to do and then believe that there's an opportunity for thousands and thousands of them.

Great. And then just the second one was just on that prototype.

Have you got a prototype that's working today that's delivering the economics that you put on the slide? Or is this still a little bit theoretical?

And when should we get to a point where we can see one?

It's probably -- it's on a spectrum. So you're kind of outlining 2 things, something nobody has ever built and something that's live and working in the exact size and format with all the pieces of equipment.

I don't have that, but I'm also not here. I'm here, okay?

I've got grids and bins and robots. I've got on-grid roboting pick.

I've got early prototypes working of dispatch ports. I've got the software that runs the robots around.

I've got the software that does the store pick. I've got software that does consolidation.

So we've got and can illustrate most of the components. We can show small sites that we built small sites, but they were 10,000 square foot.

They weren't 4,000 square foot. We've got and built -- we've got robots now running around in freezers, right?

So we've got robots in chill, robots in -- ambient robots in freezers. We've got and tested robots moving between temperature zones, which is an important part of this.

And to come back to your other question, so we're kind of -- we're here. We want to be here before we -- I don't want to sell thousands of them into [indiscernible], right?

Because I don't want to take the responsibility of delivering thousands of them that we might not be able to hit the price targets and therefore, we'll need more capital or we might have the returns or whatever it might be. And we want to clear up those process flows when we're dealing with a couple of dozen sites, not when we're dealing with 1,000 live sites, right?

We don't want to be deploying 3 sites a day at the same time as trying to make the first one work, right? Your first question is why not a 1 module site?

And there are people who are looking at sites that are not probably 2/3 of a module. There's a point where there's some things that you can do when you miniaturize that will only expand to a certain size.

And then when they expand to a certain size, some of the costs double. So for example, if you can run a single grid with a single maintenance area and you can have a part of those robots running into a chilled area, you can eliminate a whole maintenance area.

That works to a certain size and a certain size, it just isn't feasible anymore, and I need to have 2 grids. At that exact moment, I need 2 maintenance areas.

I need 2 wireless controllers controlling my bot fleets. I need more grid barriers and stuff like that, right?

There's a cost uplift. And so there is this kind of area between the biggest of the micro store-based automations and the other sites where you kind of get into an area where you can see an increase in cost, but it's not really worth it for the increase in volume.

You'd rather have 2 of these than 1 there. And then you get to a point where, no, I'm going to take that.

Does that make sense? It's kind of -- so we model an enormous amount of data around what is physically feasible to be built.

And we are talking to retailers for stand-alone sites, attached sites, ranges from 10,000 to 50,000 in the grids, throughputs from $8 million to $50 million, sizes from 4,500 square feet to 17,000 square feet, different use cases, right, different peak hours, different amounts of customer interaction, courier interaction. We've designed lots of different examples, and we're very good at simulating them and understanding what throughput should be available out of them.

We just hope to build a few that are good examples. One of the challenges at the moment is actually saying no to a few people who want something that we could build, but we don't think that's the thing that we need to have thousands of, and we'd like to build a few of the ones that ultimately we believe there will be thousands of to show that concept to the world in a real-live 100% operating environment.

[indiscernible] Thanks very much.