Good morning, everybody. Before I get into the meat of the presentation, two things to remind you of. The first is that these accounts are 100% IFRS 16 compliant, as are all the comparisons for the last one and two years. The second is, throughout this presentation, unless we say otherwise, every number is given on a two-year comparison. I should start this presentation by saying that the numbers are an awful lot better than we were expecting they were in the first half, and they have continued that way into August and September.
Although the numbers are good, the one thing I'd like to stress is that they are probably not quite as good as they look, and that there are a number of factors that we feel are artificially boosting sales at the moment that could mislead you as to how strong the rest of the year will be. That basically comes down to pent-up demand for clothing, and we're seeing that very strong on items like suits and women's formal wear. Secondly is the amount of savings that consumers have saved over lockdown, which they are beginning to unwind. The third is the fact that in August and September, far fewer people went overseas. Not only does that mean that they saved the money they would have spent on travel, but they were here to spend their money in U.K. retailers.
Those factors, we believe, will slowly work their way out of the system as we go through to the rest of the year. That said, although things may not be as good as they currently appear, we do think that the outlook is much, much better than we thought it was going to be a year ago, and actually much better than we thought it was going to be two and a half years ago. The first half of this presentation is really going to be on the numbers and our forecast for this year. The second part is going to be on how we see the future going forward, so over the next five years, and why we feel the outlook is potentially a lot brighter than we thought it was maybe two or five years ago. Moving on to the numbers.
Sales were up 8% in the first half. Full price sales were up 9%. If we look at that between the periods that we were locked down and the periods that we weren't, in the lockdown period, online was up 70%. Obviously, retail zero. Once England came out of lockdown, looking at that period of time, online sales fell back to around 46% up on last year. That was much higher than we thought. We thought we would lose an awful lot more of that online trade than we did once the shops had opened. The shops did much better than we were expecting as well. Down 8% in total, but on a like-for-like basis, only down 4% on two years ago. We were expecting compound annual decline of 6% in our retail store sales. That would have given us around 12% declines.
That just gives you a measure of the difference between what happened and our expectations, and we think is evidence of this pent-up demand. Moving on to profit. The profit was up 3%. The difference between the growth in sales and the growth in profit is largely, we believe, down to the GBP 20 million or so we lost as a result of lockdown. In terms of interest, our interest was down significantly on two years ago. Two reasons for that. The real reason was the GBP 2 million drop in external interest, and that's all about the fact that we have less debt than we had two years ago. The other factor is the lease interest.
This is an IFRS 16 accounting measure that is the interest that we don't actually pay on the stores that we don't actually own with the money that we haven't actually borrowed to buy the stores that we don't own. Profit before tax up 6%. Tax charge down on two years ago. That's partly super-deductions and partly the revaluation of our deferred tax asset, which has increased in value, and the full amount of that increase in value is taken this half, and that's because future corporation taxes are going up. Profit after tax up 8%. earnings per share on two years ago up 11% as a result of the buybacks we did two and a bit years ago. Moving on to cash flow.
I'm not going to talk about capital expenditure because the outlook and history of capital expenditure is exactly the same as we set out in March. There's no new news there. There's quite a lot to say about working capital. Seen a big swing in working capital, a GBP 4 million inflow, as opposed to a GBP 35 million outflow two years ago. Lots of things going on there. First of all, less outflow because we'd paid for less stock, and that's all about the fact that two years ago we would have been increasing our stock as we went into the autumn and season. This year, as a result of stock delays, our stock is actually slightly down. We sold the land upon which our new Elmsall 3 warehouse is being built.
We have charged GBP 21 million in our accounts for the repayment of business rates for the period of time our shops have been and will be open. We haven't yet paid that money, that comes as cash inflow. Finally, credit has gone up, mainly VAT and staff incentives we plan to pay but haven't yet paid. The flip side of that, outflows. We're beginning to build back customer receivables. That cost around GBP 45 million, we invested GBP 43 million in Reiss. It looks like we've had a big increase in employee share option trust costs. This is all about this year. We've done this year's amount and last year's. That gives cash flow before distribution up on two years ago by about GBP 11 million.
You can see the difference between the GBP 20 million of profit and GBP 11 million of cash flow is all about the increase in CapEx. Once you move all the noise underlying cash consumption of the business has been marginally impacted by an increase in CapEx, which in turn is all about the increase in the new warehouse that we're building in Elmsall 3. in terms of the balance sheet, stock down 2%. This number actually flatters the reality of the situation. A lot of the stock that we have on our balance sheet at that time was actually in transit. Stock in the U.K. was down 12%, and actually NEXT stock was down 18%. Some of the difference is things like beauty, where we didn't have any stock two years ago.
The reduction in the NEXT stock levels is indicative of how short the business was of stock at that point. Since that time, our stock position has got worse. Actually, at one stage we were at -25% and has now rebuilt itself back up to the levels that it's currently at. We're currently about -18% on NEXT stock. Each week that we move on, we can see our stock position improving, and we expect the stock to end the year around flat. In terms of debtors, unusually, there is quite a big difference between the reduction in our debtors and the reduction in our customer receivables. That GBP 50 million difference is money that we're owed by partners like Zalando and money that we're owed by companies that we've invested in, like Reiss and Victoria's Secret, where we have lent those companies money.
In terms of the reduction in receivables are down 11%. That compares to credit sales in the last six months up 13%. The apparent contradiction between those two numbers is all about the amount of money that was paid down last year during the pandemic as customer savings increased and they used that money to reduce the amount that they owed us. That accounted for about 22% of the difference. A further 2% was the fact that we have got additional provisions that we took last year that further reduced the receivables. Moving on to dividends, there were no dividends in the first half. Since then we have paid GBP 140 million special dividend, in line with two years ago.
In terms of our right-of-use assets and lease debt, these are, again, the stores that we don't actually own and the debt that we haven't borrowed to buy those stores that we don't own. Both of those numbers have moved down. That is partially about the fact that as we're renegotiating our leases, the amount of rent we're paying at each store is reducing and the terms we're negotiating are shorter than they were two years ago. Moving on to online. Online's had a fantastic half, up 52% in total, 55% on a full price basis. What you can see here from this graph is that August and September have continued as strongly as May, June, and July. Looking at how those numbers break down between brand, LABEL, and overseas. U.K. up 46%, LABEL up 70%, overseas up 62%.
If we take the total amount that sales have gone up by online, the 55%, what you can see is that the story isn't quite as simple as it appears to be. It would be very easy for people to say, oh, well, what they lost in stores, they gained online. I had a blouse that someone didn't buy in a shop, they decided to buy on the internet. That didn't really happen. What really happened is that the money that people were saving on, for example, adult formal wear, they were spending on kids wear and home. There was an enormous change in mix of the product we were selling during lockdown. That had a profound impact on our returns rate. What the blue line shows on this graph is our returns rate in 2019 hovering at around 40%.
This year you can see we started the year with much lower returns rate. On average, during lockdown, down 16%. Of that difference, the vast majority of it, 11%, can be explained through just the change in product mix. If you take the low returns rate on home and kids before the pandemic and use those to calculate what the returns would've been during the pandemic, you'll get 11% drop as a result of mix. The balance was consumer behavior. This was phenomenal, where customers reduced the total size of their orders because they were being more careful about only ordering the things that they really wanted to keep. That gave a further benefit of about 5%. What you can see is that since the pandemic ended, our returns rate has come back to where it was broadly two years ago.
You can see that the customer behavior changed pretty much overnight, and the mix has more slowly moved back to a more normal level. I think that's indicative of the fact that we are expecting product mixes to return to pre-pandemic levels slowly as the rest of this year and next year progresses. Moving on to growth by customer type. What you can see is that all areas grew strongly. U.K. cash customers grew the most, mainly driven by new customers. If we look at the overseas number, that overseas number at 62% could be misleading taken on its own. If we look at the growth on sales on our own websites overseas, they were up 49%. However, we're getting an increasingly important contribution from third-party websites, people like Zalando in Europe, and those grew at 210% during the pandemic.
Looking at sales per customer, what you can see in the U.K. is that we saw a significant increase in both credit and cash sales per customer. That phenomena, we believe, is down to lockdown, and you can see that the average sales by month and how once lockdown finished in April, they returned to levels of two years ago. Moving on to profit. Profit up 74%. A significant improvement in achieved margin, just over 2.5%. Bought-in margin was down 1.4%. Two things going on there. The first is increased costs of freight. As the season progressed, freight prices went up, and we didn't have time, in many cases, to incorporate that into our prices. What that meant is that we took around an GBP 8 million hit to our P&L in the first half of absorbing those price increases.
In the second half, all but GBP 7 million of those have been incorporated into prices. The balance of the change was down to the fact that we sold a lot more LABEL, which has a lower bought-in gross margin, and within the mix, the home and kidswear has a lower bought-in gross margin than the high-fashion items in adult clothing. Surplus, we saw a significant improvement. This wasn't that we cleared the stock that we had any better than two years ago. It was the fact that our stock for surplus grew by only 12%, whereas our full price sales were up by over 50%. It was all about the amount of sales stock we had rather than the effectiveness with which we sold it. You can see that actually clearance rates were down very marginally on two years ago. Stock up 12%, marked down sales only up 10%.
In terms of other costs, the big saving here was the fact that we're no longer printing our catalog, and that is net of any increase in digital marketing costs. In terms of warehousing and distribution, 0.6% positive contribution from warehouses. Leverage over our fixed overheads and lower returns rate driving the gains there. On the earlier slide, you'll have seen that lower returns rates gave us a GBP 20 million benefit in the first half. That was partially offset by the fact that when we're delivering to customers overseas, the vast majority of those deliveries are done by air freight, and we're seeing big surcharges on air freight prices throughout the first half. We think those will continue, by the way, into the second half and to some extent into next year. You can see that cost is 0.8% of margin.
Leverage over systems and central costs gave us a further 0.6% advantage, and that gives this huge swing in the first half. In the second half, we're not expecting the benefit from returns to filter through into the accounts in the same way. You should expect in the order of 20% net margins online. That compares to around 19% two years ago. Moving on to the finance division. The finance accounts are relatively straightforward, with profits moving in line with receivables. The only marginal complication is the fact that the credit sales are up, as we mentioned before. The receivables are down. This graph explains why that happened. What you can see here is our average balances for the last two and a 1/2 years. The gray blocks are the blocks last year.
You can see that if I put monthly sales on this, you can see that where we saw the big drop in sales, just as the pandemic starts, and that's when we closed our warehouses and when consumers reined in their spending, you can see the balance is dropping. Conversely, where we saw sales rise at the back end of last year, credit sales rise, we didn't see anything like the same increase in balances. That's because for the whole of the last six months of that year and the beginning of this year, consumers were paying down their accounts faster than the previous years. Net profit down 13%, in line with receivables. Return on capital 12.9% this year compared to 13% two years ago.
As we look into next year, we are expecting our finance profits to continue to grow in line with the credit business. Moving on to the retail business. Retail had a tougher half. Sales down 38%. As we mentioned before, like-for-like down 4%. In terms of how the shape of sales when we reopened the shops, what you can see is in April and May, +2% and -1%, these are like-for-like sales. You can see that we really did get a significant bounce when the shops reopened, and we got a bit of pent-up demand, and then steadily that fell away to sort of more normal or expected levels of sales at -12% in July.
We believe that the sales in August and, to a certain extent, the front end of September, and these are estimates that we put on here, we believe that that is a result of people not going away on holiday. For the rest of the year, we're anticipating that retail sales will be more in line with July. In terms of the performance between the different types of stores that we own, we saw exactly the same pattern that we saw during the lockdowns last year, with retail parks bouncing back much faster, and in fact growing, as against city centers and regional shopping centers, both that fared much worse. Although that trend was the same as last year, if you look at the difference in performance last year, you can see that it was much greater.
We think that what will happen is that as the city centers begin to come back to life, we will see the difference in performance between city centers and retail parks beginning to narrow over the next six months to one year. As you can see, we were fortunate in that when we went into lockdown, 62% of our sales were already coming from retail parks, which helped sales when we came out of lockdown. Operating loss of GBP 18 million in the first half. That number is significantly flattered by the protocols of IFRS 16 accounting. If you add in the lease interest, which in our accounts has to appear in the interest section, the loss in retail would have been around GBP 309 million.
Going forward, when we talk about our retail profit, we will include, for the purposes of our management accounts, the lease interest within the retail numbers. A loss of GBP 39 million in the first half. We're expecting for the full year a profit on the same basis, including lease interest of around GBP 60 million. In terms of rent, and this is the rent payable rather than lease interests, what you can see is that against two years ago, we've made around a GBP 40 million saving. GBP 20 million of that is as a result of closures, and the other GBP 20 million is as a result of the renegotiation of lower rents with landlords. If we look at the leases that we've renegotiated this year and the ones we expect to complete this year, we will negotiate 73 shops this year.
Rents in those shops we think will come down around 52%. The weighted average term of the leases we're negotiating is three years. The annualized saving on this portfolio around GBP 11.5 million. Importantly, of those 73 stores, 28 of them are on flexible rents. This is where the rents come down or go up as sales vary. Of those 28, 15 of them are total occupancy costs.
This, in many ways, is the ultimate flexibility, where we pay landlords a percentage of sales to cover rent, rates, and service charge. Just to give you one example of that, this is a shop in Kent that we renegotiated recently that was turning over GBP 1.6 million. That shop, the rent was GBP 200, rates GBP 124, service charge of GBP 88. We've negotiated with the landlord to renew for three years on the basis that we pay 14% for the lot.
That would give us GBP 230,000 of rent with a 44% overall reduction in occupancy costs. One of the things that we're finding with landlords is, in essence, the more flexible they are prepared to be with the rent terms, the higher the amount we're prepared to pay in the short term and the longer lease we're prepared to sign. If the landlord will give us a total occupancy cost, we're happy to sign up to five years, and we're happy for that number that we pay today, as long as the store is profitable, to be slightly higher than it would be if we were negotiating a fixed rent. Just going to talk about the estimated cost of the lockdown at the beginning of the year in the context of the whole company.
Starting with retail, we estimate that during that 10-week period, we lost around GBP 250 million of sales. That estimate is based on the assumption that the stores would have been down 12% on two years previous, i.e. a compound annual decline of 6% on two years ago. We think we picked up GBP 20 million of those sales in the April, May bounce, giving us a net loss of GBP 230 million in retail. The marginal profit, because all of that loss was pretty much full price, was 54%. That cost us GBP 125 million of margin. With rates relief in the period we were closed of GBP 20 million and some other cost savings, electricity, maintenance, and recharges that went to online around GBP 19 million. The effect of lockdown on retail profit we think was around GBP 86 million.
If we look what we picked up online, we think we picked up in the order of GBP 155 million, around 67% of the sales we lost in retail. Just to quickly remind you that we don't think that was because people didn't buy blouses in store and bought it online. It's not as simple as that. It was more about the money that people were saving on buying clothing and restaurants and other things they spent on home and children's wear.
That gives us a slightly lower net margin, 39% of GBP 60 million. There were other costs of GBP 14 million, mainly the share of central overheads that retail wasn't paying. Return savings are very important, that will disappear in the second half of GBP 20 million. That gave us total win online of GBP 66 million. Total cost of lockdown to the current year of around GBP 20 million.
If you're surprised that that number isn't bigger than that, it was also a surprise to us. We have been through these numbers again and again and again, and we think that's about as much as we can get to. Mainly because we picked up so much business online in home and children's wear, and that wasn't necessarily business that we were only picking up from the closure of our stores. Moving on to the outlook for the full year. In the first half, we were up 8.8%. Our guidance for the second half was that full price sales will be up 6% across the whole business. We've revised that today, and we now estimate that full price sales in the second half will be up 12%. That's a combination of two things.
First of all, season to date, we are up 20%. We are forecasting for the rest of the season that we will be up 10%. That corresponds to an increase online of around 32% and a decrease in retail of around 13% on two years ago. That would give us 11% for the full year. Breaking that down between retail, online, and finance. Focusing first of all on the GBP 410 million that we anticipate losing in retail. We think that will cost us in the region of GBP 220 million of marginal profit at a rate of 54%. We'll win back some of that by NEXT branded sales, one online, which we think will deliver around GBP 156 million of profit. Slightly lower achieved margin than retail, around 48% marginal profit.
LABEL, again, lower marginal profit because we're working at lower margins, GBP 77 million. Overseas GBP 40 million. Total Platform we think will deliver GBP 10 million of profit in the current year. Of that, only GBP 3 million is the profit that we make on the commission on our partner sales. The balancing GBP 7 million is the anticipated share of profit that we will make in the equity investments that we've made in various partners. The lion's share of that profit we anticipate coming from the Victoria's Secret joint venture and some of it from Reiss. In terms of finance, GBP 17 million move backwards on finance, which we've explained. Then in terms of cost increases and cost savings, both of them equaling each other out, GBP 135 million down and GBP 140 million gain.
I'm not going to go through those in detail here, but they are detailed in your pack, which you can read at your pleasure. That will give us total profits, if things pan out as expected, around GBP 800 million, and an earnings per share growth of around 9.4%. In terms of that GBP 800 million, in terms of it translating into operational cash flow, we anticipate that it will translate into roughly GBP 690 million of operational cash flow. That's before CapEx of GBP 185 million, investment in customer receivables around GBP 116 million, and the investment that we've made in Reiss, GBP 10 million debt, GBP 33 million of equity. If we left it there, the company's gearing would drop to around GBP 400 million, which we think is lower than it needs to be or should be.
We anticipate pushing debt to the year end back up to around GBP 600 million. To do that, we will be able to distribute in the order of GBP 208 million special dividend. What we'll do is, as we approach the end of the year, when we have a more accurate picture of how much surplus cash we will generate, we will declare a special dividend and pay it in the current year to get us to around GBP 600 million of year-end debt. In terms of why we're comfortable with that GBP 600 million as a level of debt for the business, the main reason is because nearly twice as much is matched by customer receivables. If we were only a customer receivables business, actually that level of gearing would be low. In terms of the financing of that debt, it's very comfortable, GBP 1.3 billion of financing.
This is after accounting for the bond that we will repay in October. We think that'll give us around GBP 500 million of headroom over our peak cash requirements next year. Taking a slightly longer view of the company, at the beginning, I said that we felt that the outlook for the company was not just brighter than 18 months ago, but a lot brighter than it was two years and five years ago. It's worth just reflecting that five years ago, in 2017, we, at that point, stopped buying back our shares, not because the share price was too high, but because we weren't sure that there was a lot to invest the money in in the company. The best thing we could do with the levels of uncertainty we had at that time was to pay money out to shareholders.
Two years later, the situation had got significantly better. We ran our 15-year stress test. Our online business was beginning to motor. At that point, we established that the economics of the business were such that as long as our online business could go slightly faster than our retail business was declining, there was a way through to a profitable business, and that we were generating in the order of GBP 12 billion of cash over that 15-year stress test cycle. As we stand today, things feel significantly better than they did two years ago, really for two reasons. The first is that the threat to our finances from our retail business has declined significantly, and the second is that the opportunities we have online seem more numerous and bigger today than they did two years ago.
Just in terms of the retail threat, really, there's nothing clever about that. That is just pure maths. five years ago, retail took 60% of our trade. This year, we think it will take in the order of 30%, and pretty much the same again next year, certainly no more, as our online business continues to grow. Moving on to the opportunities online. There are four areas of focus. The first is within our own brand. Over the last five years, we've more than doubled the amount of choice within our ranges. In essence, what has happened is that our buyers have ceased to be constrained by the four walls of our stores. That's allowed them to experiment and push into new fabrics, price architectures, fits, sizes, and designs that they simply couldn't have fit into our store portfolio.
They've also pushed into new areas, everything from performance trainers through to garden furniture, extending the boundaries of the NEXT brand, with the one caveat being they have to be adding something to the product by way of design, because if they're not doing that, they're just producing copycats. What we said to our buying teams, look, is that as long as you can create value and give something our customers will appreciate, something that's new to them, then try it. Online allows us to do that. In addition to the additional products that we've got within our own ranges, we've also dramatically increased the amount of third-party branded stock that we're selling on our website to the extent the business this year we think will take over GBP 700 million. That is partly about the addition of new brands.
This year, our growth over two years ago, about a third of that will come from new brands. Much more importantly, we have deepened and broadened the products in the brands that we already partner. An important part of that has been our Platform Plus system, which I think we first talked to you about 2.5, three years ago. This has allowed us to have visibility of stock that is available in our partners' warehouses, that we don't stock, that we can sell online. What we do is that if the order is taken on a Monday, we'll pick it up from our partners on a Tuesday and deliver it to our customers on a Wednesday. We're offering Platform Plus stock on a 48-hour promise. Importantly, we have taken ownership of the service.
The moment that stock leaves our partner's warehouse, we have visibility all the way through to when it's delivered to our customers to when it's returned to our warehouses. What we've developed here is not a marketplace in the traditional sense of the word because our partners aren't delivering it to the customers, we are. It's additional choice without degradation of service. We've increased our customer base, and of course, the vast majority of the increase in our customer base over the last year was down to lockdown. Behind that, something else is going on. We have significantly improved, over the last three or four years, the software, techniques, and the skills of the people who are placing our online advertising, to the extent that we are getting much, much higher returns on the investments that we're making in all forms of digital advertising.
That means that in the current year, we'll spend around GBP 100 million on online marketing, and we expect to push that further next year, because I think in some ways, we're at the beginning of this journey, not the end of it. What we're also doing is that in the past, we have struggled to make a success or get the returns from adverts we place in third-party media to sell third-party brands on our website. For example, where in the past we've placed adverts for Adidas or Nike or any of the other partners that we work for in third-party media, it's worked, but not enough to pay for the investment. That's because we're only making half the profit or less than half the profit on the sale of those goods. The other half goes to our partners, quite rightly.
What we've experimented with recently is partnering with brands where they pay for half the advert and we pay for half the advert. We've had a very successful collaboration with Adidas in the past year, and that has proved extremely successful. A situation where it wasn't worth either of us advertising on other sites for their product on our website means that combined, both of us make a profit doing it. We think we spend about GBP 4 million on that this year, and we expect that budget to more than double in the year ahead. Moving on to Total Platform. Total Platform this time last year was literally a glint in our eyes. We didn't have a single operation live. Today, we have four clients live.
We expect Reiss to go live in February next year and Gap to go live in the summer, towards the end of the summer next year. The partners that we've got are up and running, are operational. The operations are working well. The sales that we're getting through those partners are overall in line with expectations, and the profitability is coming through in line with our target margin for Total Platform of between 5% and 8%. You can see this year we'll make around GBP 3 million profit on GBP 50 million of gross transaction value for the partners on their websites. I've mentioned earlier on, we'll also make GBP 7 million on our equity investments.
Just in case you're thinking that that GBP 7 million looks like too big a slug of the GBP 50 million gross transaction value, you're right, because a lot of that profit, particularly in Reiss and Victoria's Secret, relates to their retail turnover, in which we have a stake, but which doesn't go towards the gross transaction value on Total Platform. Those are the opportunities, and we are quite excited about what we can achieve over the next five years on these opportunities and more. I think there are three really tough questions that the business needs to answer if we're to make a success of the next few years. The first one is, are the customers that we've recruited through lockdown here to stay? Are they just people who came on our site for lockdown and will disappear, or will they behave more like normal customers?
The second is, can our warehouses cope? In particular before we open our new big automated warehouse at the end of 2023. The third is whether our technology, our IT systems, are really ready to deliver the sorts of applications we need to deliver to make a success of our online business. I'm going to start with an analysis of the customers. What I'll start by saying is that although the evidence we have looks good, it's by no means conclusive. What you can see, this chart just shows the customers we started the last three years with. You can see that the big growth in customers has come in new customers. These are customers who at the beginning of the year had not traded with us for more than 20 weeks and placed more than one order.
There is a question mark over those customers. What I'm going to share with you now is the evidence we've got of their behavior since we've recruited them. If we take the cohort that we recruited in the run up to January in 2019. Those were customers recruited November, December, January 2019. Compare them to the customers that we recruited in the same months in the run up to January 2021. If we look at their retention rate, the 806,000 customers, the retention rate there was around 18%. If we look at the 1.4 million recruited this year, despite the fact that it's a much bigger number, their retention rate over the last nine months is slightly better than it was on the two years ago.
There's nothing there that would suggest that these customers are going to leave us any more quickly than the customers recruited before lockdown. In terms of average spend, the average spend is up. That's what you'd expect, because remember earlier on we were saying that during lockdown, customers spent more. The amount that the sales have increased for these customers is broadly the same as the amount of the rest of our customer base, around 23%. We look at another cohort, and these are the customers that were recruited in February, March, April 2020, and traditionally the customers recruited outside the Christmas period retain much better. Those customers have retained at 23%. Looking back to before then, to the customers recruited in 2018, they retained at 18%. If anything, we're seeing better retention rates. The same in terms of average spend.
We're seeing an increase in average spend of the new customers versus those we were recruiting two years ago. It's too much to say that it looks positive, but it certainly doesn't look negative. It's early days. There's lots that still hasn't really worked its way through the system. There are a lot of people still working from home who may return to the office. Stores have only been open five months, and retention may have been significantly improved in August and September as a result of people not being away. Where we stand today, the stability of the customer base that we've recruited during lockdown looks encouraging.
In terms of whether our warehouses can cope or not, what this graph shows is the capacity, the weekly picking capacity of our main box warehouse, and that is where we do 80% of our picking, and it's where all the capacity constraints of the business are. two years ago, we could pick 3.4 million units a week. Over the last two years, we've increased capacity by around 15% to 3.9 million units a week. If we compare that 3.9 million units to the profile of demand over the last six months and project it forward, what you can see is that at several points already, we've bust that capacity. It's not as bad as it looks for two reasons.
The first is that what I'm going to now show on the graph are full price sales. You can see that where we've bust capacity, it's normally because of a sale event, in particular the mid-season sale in March, April earlier this year. The reason that's not a problem is because sale stock is not promised for next day delivery. We give ourselves two weeks to deliver most of our sale stock, if not longer. That's the promise the customer gets up front, which allows us to smooth that markdown sale picking into quieter times. It does beg the question, though, as you look at the end of the graph towards December, as to whether capacity will constrain our ability to deliver sales at Christmas.
It also begs the other question is that if we beat our demands target, will we have the capacity to serve that demand or will the constraints of our warehouse stop the business in its tracks? The answer to that question, we believe, is that the capacity won't stop us, and that we can deliver significantly more than the apparent capacity of our warehouses, but that will come at the expense of service level. I just want to spend a little bit of time explaining that. What this graph shows is the amount of stock we pick and pack by hour throughout a 24-hour period. What you can see is that at 2:00 A.M. there's a dramatic drop in packing. The reason for that is that 2:00 A.M.
-is the last time we can get stock out of our warehouses in order to get it to the customer that day. If we stop taking orders at 11:00 P.M. for next day delivery, we've then got three hours to get the order we take out at one minute to 11, we've got three hours to get it out of the warehouse at 2:00 A.M. That gives us a lot of spare capacity the following day. If we hit a problem where we can't fulfill, where we bust that capacity, what we can do is pull forward the cutoff. In this example, we pulled the cutoff for next day delivery forward to 8:00 P.M. That gives us six hours to pick and pack the last item promised for next day delivery.
Much more importantly, it allows us to take the picking that we would have done before 2:00 A.M. on the orders taken after 8:00, 9:00, 10:00, 11:00 o'clock at night and move those into the following day. We can smooth our orders into times of the day where we have the capacity to pick and pack. That does come at a cost because it's more expensive for people to be working through the early hours of the morning than it is during daytime. Obviously there is a degradation in service. Some of the customers ordering stock at 9:00 o'clock might think, well, if I can't have it tomorrow, I won't order it at all. Our experience is, particularly at the busiest times of the year, that most customers will tolerate a two-day delivery rather than a one-day delivery after 8:00 at night.
The worst that can happen is that we will lose some of the demand that will be way in excess of our forecasts anyway. We're not overly concerned about this, and we believe that we do have the capacity to get through Christmas, and to beat our targets if necessary, albeit that that may come at the expense of the service we offer our customers. Just looking slightly further ahead into next year and the following year, in the run-up to the opening of our big warehouse. Next year, we think we can deliver another 15% capacity in around about July. This is partly the use of picking space in the shell building of the new warehouse, and partly the addition of a new automated packing sorter, which will come online around February, March next year.
Once we're into 2023, in October, we'll begin to commission the new automated Elmsall 3 warehouse. That, as we go into the following year, should give us a 45% increase in capacity. The pinch point is going to come next year, where we'll be up against this 15% increase. On our forecast to date, we think that we'll get through next year okay. Again, we've still got that same ultimate safety valve that we can pull forward the cutoff in order to give ourselves more capacity. Just to put the 45% capacity in context, the graphic that you're about to see is of our Elmsall 3 warehouse. The buildings behind are our existing warehouses. They're outlined in red. This new building in the front, half of it will give us the 45% we mentioned on the graph.
Within 18 months of us deciding to mechanize the second half, we can give the company a further 45% in capacity to take that increase to 90. Once we've got Elmsall 3 up and running, we think not only do we have more than enough capacity to cope with what we can reasonably expect by way of online growth in our own business, but also the capacity to take on a significant number, and amount of business on Total Platform. In case you're wondering about what we're planning to do once the 90% is burst, we have acquired land next to the warehouse, and we'll be looking to get planning permission on that to build an Elmsall to cope with the growth that we may get in four or five, 10 years' time. Final question is whether our technology, our IT systems are ready.
The answer to that is we think they are ready, but getting them in the state that they need to be in order to move our business forward is going to be expensive and take a lot of hard work. Just to remind you where we're up to, we've talked about modernizing our systems before. At the moment, our systems are pretty much all developed in-house. They are function rich, they're resilient and secure. The code is written as a monolith, and what that means is that each application, say the login application, will reference lots of other applications within its code as it's being written.
What that means is that because we have a lot of interconnected code within all these different functions, when we develop one part of the website, let's say login, it can have an adverse effect on any of the other parts. It could knock over our delivery screens. That's not a problem as far as the operation of our technology is concerned, and it's not a risk to our current operation. What it does mean is that developing our software takes a long time, because we have to spend almost as much time testing it as we do developing it. The process we started two years ago takes all of these functions, divides them into discrete containerized applications, and that each piece of code, each application only communicates with the others by passing data between them through a communication layer. That process is well underway.
It means that all of these applications will be able to be developed alongside each other. The progress we've made so far is mainly on our e-commerce site, where we have modernized the header navigation, footer services, cloud infrastructure, and the test and release system. Over the next nine months, we'll deliver search and product testing, customer login, product display page, and content personalization. What that will do is that will modernize the parts of our website that we change most frequently. Once we've done that, we should see a significant acceleration in our ability to develop the website. We're modernizing the back end of our website over the following two years. At the same time, we've kicked off modernization process for all of our other major technology applications.
Whilst everyone can see the need to modernize our systems, what I wouldn't want you to think is that because we're modernizing them, we're not developing them. Actually, we have to continue to develop our systems as fast as we would have done were we not modernizing them. The analogy that our IT director uses is that it's like running a hotel where you've got to redecorate, rewire, replumb every room in a hotel whilst continuing to trade and only ever shutting two rooms at any one point in time. We recognize that is a big challenge, but it's what we've been doing for the last 18 months. It also means that every so often we're going to have to write code in our legacy system and accept that we may have to rewrite that code when we modernize that particular application nine months to a year later.
We've taken the view that we'd much rather move our systems forward and duplicate some of the efforts than we would to stand still. Because at this point in time, really, our technology can't stand still. What that means is we're going to spend a lot more money. These costs, these capital costs, are already in the projections that we've given you. We anticipate that in the current year, we'll spend at least GBP 36 million on technology capital. The vast majority of all of this expenditure is on software. At least 75% of it will be on software, and that's really people costs. The balance will be on infrastructure, a lot of which will actually be code in the cloud.
Putting that in the context of our revenue expenditure, what we anticipate is that over the next few years, our total spend on technology will get up to around GBP 170 million a year. What is interesting about this number, apart from the size of it, is the fact that where we are today, we employ pretty much the same number of people in our technology teams as we do in our buying, design, merchandising, and sourcing teams across the business. What this shows is not only the amount we're prepared to invest in technology, but also the importance it will have in moving the business forward. In order to do that, as I've already mentioned, we're going to have to take on a lot of people.
We have reorganized our entire technology department, and if you're interested in this sort of thing, there are, I think, three or four pages in our annual report that explain exactly what we're doing. We also regraded and recalibrated our wage rates within the business to make sure that we're competitive in terms of recruitment. Those are the three sort of challenges and uncertainties. I think the three biggest challenges and uncertainties facing the business. What I think it would be very easy to do is to look at the success we've had over the last year, and in the last few months, and kid ourselves and our investors that it's plain sailing from here on in. It is not going to be plain sailing. It's going to be really hard work. We think all of these things are doable, and we are on with them.
They need to be taken in context of the opportunities that the group is now presented with, which I think are significantly more numerous and bigger in scale than we have had for the last five or six years. The way that NEXT feels at the moment is it feels like a very different business from the one that we were managing five years ago. For the last five years, a lot of what we've done has been a rearguard action against the problems caused by the decline of our retail business. Where we stand today, those problems are far smaller in size and number, and the opportunities far bigger. That's all I've got to say today. I'm finishing on an unexpectedly optimistic note, and hopefully in a time period that is slightly less than our normal presentation.