done in the past, which was amiss, and I apologize, we should have done this from the beginning, is to pay our respects to the elders, past, present, and emerging, and acknowledge the traditional custodians of the country on which we operate. Our head office and where we're sitting today is on Gadigal land, but we have staff, customers, and partners across all of Australia. I won't spend too much time on the summary. I will get into the detail of this. I will turn, just as a refresher of what Beforepay Group is and what we do. There are two parts, as many of you know, to our business. Beforepay is our Australian domestic lending business. It has two key products.
It's got the Pay Advance, that's our flagship, short duration, a small dollar, mission-driven, ethical lending product. The Personal Loan product, a newer, larger dollar, longer duration product that we've been in the process of rolling out and are starting to scale. That's the lending business Beforepay. We also have Carrington Labs. That's our business-to-business offering, SaaS analytics capability that we offer to lenders all around the world. Our primary market for that is the United States, although we work with lenders anywhere. What that business does, is that provides us a lot of the IP and capability that we've developed in terms of how do you assess borrowers, especially those with non-traditional credit profiles, in a more effective and efficient way.
That's part of the business that obviously we'll be talking about over the course of this webinar as well. Laavanya will go through these numbers in more detail on the following pages. The only one I want to call out, and the real standout of the result is, of course, that AUD 4.2 million NPAT number. That's up 50% from the same period last year. That, to me, just, I think, summarizes where we're at as a group right now. I think the group is firing on all cylinders. You can see the other metrics here moving in the right direction.
I think what we continue to demonstrate is that we have a sustainable, attractive business in Australia with that lending capability, and I think we really both have a great core with Pay Advance product, as well as some really exciting growth starting to emerge from that Personal Loan business. Of course, the exciting option of Carrington Labs, kind of build off that same capability, technology, and IP as the Beforepay business. So I think that very strong profit result from the half just shows that the group continues to go from strength to strength. Laavanya will now go through some of those numbers in a bit more detail.
Jumping into the detail, you can see that advances have grown from AUD 397 million in H1 FY25 to AUD 467 million in H1 FY26, that was an 18% growth. This was driven by the average advance size increasing, which you can see on the right-hand side. It's gone up from AUD 393 to AUD 458, which is a 17% growth. We've continued to focus on higher value customers, driving that increase in the average advance size through our performance marketing, as well as our credit risk analytical work. We've got a slight increase in our loan defaults from 1.1% in H1 FY25 to 1.3% in H1 FY26.
This small increase, as we've talked about in the past, we haven't targeted a specific net default percentage. Instead, we're focusing on the overall profitability of the business. This slight increase resulted in the average advances in the previous slide increasing.
Yeah. I think actually it's worth noting two other things on this page. One is there is a natural half-on-half seasonality, as you can see across the past several halves between the first and second half. I don't think we're surprised by the 1.3% number. As Laavanya noted, we do think very carefully about the trade-off with limit management and the default rate in order to maximize that marginal contribution, what we call net transaction margin, and I think we did that successfully this half. I think as many of you who have been on the journey for a while, will know, in recent quarters, I've actually been worried about the low default rate. It suggests that we haven't been optimizing those limits fully, I feel very comfortable with that 1.3%.
We have a small increase in our direct service costs from AUD 0.77 per advance to AUD 0.80 per advance when you compare FY25 to FY26. As we have increased our average advance size, that's required additional deductions to be able to collect on that amount, resulting in that small increase. What you can see is the percentage of advances has actually decreased. As a dollar value, we've actually improved on our profitability. We've gone from 0.2% of advances in H1 FY25 to 0.17% of advances in H1 FY26.
All of that comes together on this slide, the improvement in the revenue, plus the optimization from a cost perspective, resulting in the net transaction margin increasing from AUD 11.8 million in H1 FY25 to AUD 14.3 million in H1 FY26. That's an increase of 21%. We have had an increase in our operating expenses from AUD 8.8 million to AUD 9.7 million. As we've spoken about in previous quarters, this is as a result of our investment in our people side from a Personal Loans and a Carrington Labs perspective.
Yeah, I think many of you have heard us say over the years, that we don't generally target individual components of the overall PNL for as a management tool. These are the ones that we do target. We focus very much on optimizing that overall net transaction margin, as we were saying a moment ago, with that trade-off between the individual elements of the number of users, the average loan size, the defaults, and how those trade off. We have a pretty analytically rigorous way of doing that. As this number continues to move up, we feel good about that trade-off. If you do the math carefully, you'll see that generally, we like to see NPN grow faster than underlying components. That means we're getting that trade-off right, and here you can see that it is up 21%.
Coming down to our profit before income tax, you can see we've brought in AUD 3.7 million for the half year. We've also recognized a further AUD 0.5 million worth of unutilized tax losses on the income tax benefit line, giving us an overall NPAT of AUD 4.2 million, which is a 50% increase on the same time last year. A big improvement. Here on the balance sheet, we've got a bit more details. The thing to focus on is we've got a very healthy cash balance at AUD 9.1 million. We continued to be able to fund our customer advances, which has been growing, as I pointed out in the previous slide, through our cash position, as well as our growing loan book.
That's resulted in that great cash position. You can see that our drawn facilities remains flat at AUD 30.9 million, and we still have undrawn facilities of AUD 24.1 million remaining. Our equity position is also very healthy, it's grown because of the profitability up to the AUD 44.4 million.
I think that point on cash is a really important one. Cash is down about AUD 5 million over half. The reason for that is the loan book grew about AUD 5 million. The way we've structured our debt facilities, we have the ability to fund up to 80% of that through our third-party financing facility. Because we have excess cash on the balance sheet, we've chosen not to do that. As a result, we've funded those loans directly out of cash. We have the same resilience and strength, and would be able to draw down that 80% if we drew down the facility, should we choose to do so. We're simply trying to sweat the assets a little bit more.
You do see that flow through on the previous slide, where you see the financing cost actually has been dropping as we're funding these things through funding loan growth through the cash reserves.
We've given you a bit more information on the cash flow statement here. Again, reiterating that you can see the receipts from repayment of customer advances growing from AUD 388.3 million to AUD 452.2 million. That increasing advances, combined with our ability to be able to collect on those, contributing to that increased number. Again, pointing out the really healthy AUD 9.1 million we continue to hold in cash.
Finally, as we look forward, and again, we have a pretty consistent and steady strategy that we've been executing on for the past couple of years. It evolves slowly, but it doesn't, we haven't made any radical shifts, 'cause we believe we're on the right course, and I think our results back that up as well. In the core Pay Advance business, on the Beforepay side, that business is working well. We continue to optimize it, you know, again, managing limits carefully, continuing to drive growth through getting new customers, and particularly, the customers that provide more value to the group.
As many of you know, we have really been focused on that question of optimizing that customer base and getting the right ones in. There are some lower value ones that are, have been less of a priority to attract and retain. You know, that top line growth, the optimization of the margins, and of course, being very disciplined on cost. It's an enormously efficient and automated business. We're doing tens of thousands of these each week with a very small number of people. The Pay Advance business working well, keep getting better, keep doing the same thing. On the Personal Loan side, at the full year results, some of you will recall that I said 2026 will be the year of Personal Loan. I stand by that.
We've started to significantly increase the origination volumes there. We're still being cautious in terms of how broad we go and being thoughtful about credit, and we're still collecting data and experimenting. I feel confident about the future direction of that business and the ability to scale it over time. Some of you have asked in the past, "Will that flow through to the balance sheet?" The answer is, of course, in due course, yes. We are looking at how we can extend and expand our current debt facilities to accommodate that product's growth over time. Finally, in Carrington Labs. We do continue to invest in this business. We've increased the size of the team on the ground in the U.S.
You know, we had a flurry of wins and announcements over the half that probably many of you saw and are aware of. We also do continue to invest heavily in making that product better. It's a good capability, and so we continue to both improve that raw analytic core of the Carrington Labs offering, as well as the tech stack in which it sits and the way that we deliver that solution to our clients. The strategy continues to be very similar. We continue to just move forward methodically across all three elements of the strategy, and I think just to finish where I started, the strongest half the group has ever had, clearly by a pretty significant margin.
I do come back to that 50% year-on-year profit improvement, while continuing to invest in the business and deliver on the strategy. We're very pleased with where we are, and we always like to, you know, underpromise and overdeliver. I think we've done that for certainly this half and hopefully continuing to do so into the future. With that, very happy to move on to your questions.
Thank you, Jamie. Thank you, Laavanya. We will now move to the Q&A session. Once again, if you have any questions, please type them into the box on the screen. We do have several questions already. Do you think the relative slow rollout of the Personal Loan product impinges on the assertion of the superiority of the Beforepay credit risk model?
No. No, I don't think that's the case. I think, Sorry, I was about to go down a very deep kind of credit data, statistical, rabbit hole, and I pulled myself out in time. I think, with the Personal Loan, I think we have the luxury of getting that credit really right, and the way to do that is to issue a handful and then get the data, and then issue more, and get the data, and so on. Find the model as you go. That's what we've been doing over time, and also getting make sure the operations and the tech stack are robust. I feel very confident in the quality of the credit risk that sits behind both the Pay Advance and Personal Loan product as well.
Thank you. Your next question: Can you provide some commentary on refinancing of your facilities, whether you would expect to achieve interest rate savings, and whether your intention is to maintain the current size of those facilities?
We currently have a debt facility of AUD 55 million, which expires in October. We've started discussions to get a new facility. Our expectation is with the Personal Loan product continuing to grow, and with higher dollar values and longer duration, that we'll need a higher debt facility in order to be able to service it. Of course, we aim to do that with reduced interest rates and improvement from a PNL perspective.
Currently, funding costs in the half were about 50 basis points, 0.5% the amount advanced. The interest savings, assuming they occur, will be very welcome. They won't dramatically shift the economics of any of our products specifically.
Thank you. Your next question: Would your net transaction margin look the same if you were funding loans through your facilities instead of cash? If not, how would it have changed the picture if you had funded it through your facilities?
Yes, if we borrowed from the debt facility, obviously we would have had the additional interest expense off the back of that. In terms of the fully loaded cost of the facility, again, there's a cash cost, and there's this amortization. With the amortization, the range is between about 14.4%-15.4%. On a cash basis is 12.25%-13.25%, depending on some profitability metrics that we would hit. It would depend on the funding rate, overall, if we fully drew every bit of the facility at the highest cost, including an amortization, the cost of funding advanced through the debt facility is about 87 basis points based on the average duration.
When we fund it through cash, it's essentially zero. It'd be a tiny opportunity cost because we wouldn't then use that cash in an interest-bearing account of some kind, but it's not very material. If you look across the... That's for an individual advance. If you look across the entire book, probably the best way to think about this, if you looked at our interest costs over time as a percentage of funding, they were 0.7%, and now they're about 0.5%. That gap, there are some different moving parts in it, but predominantly, it reflects the change in the drawdown patterns across the whole of the book. Other than that.
Thank you. Next question. Given the heightened investment in Carrington Labs and the personal loans division, are you forecasting revenue expansion over the short to medium-term horizon, sufficient to sustain continued profit growth?
We don't put out forecasts. I think that's that's always been the way we've done that. I think that served us well. We are certainly rational managers of the business, and we don't ever let a dollar leave this building without expecting that it's gonna more than pay its own way and be a sound investment over time. I don't think we necessarily have a kind of very, like, specific point of view on what we want to see now versus three months versus six months, but we are very thoughtful about any expense and are very keen to make sure that we get a good return.
Thank you. Your next question: Can you provide a little more detail on the progression of Carrington Labs' sales pipeline, what timeframe do you provide for meaningful revenue generation before you would revisit the resources dedicated to this?
Yeah, good question. I think. Actually, let me start with kind of the timeline part of it. The honest answer is that the clients of Carrington Labs, especially as we engage with kind of larger institutions, those can be significant sales cycles. If you're talking to a, you know, a major bank in the U.S. or elsewhere, that can be a lengthy process, especially if it's gonna be something fairly material. It's difficult to say, and again, we like to kind of, we like to talk, do things and then talk about them instead of the other way around. It's difficult to say, here's where we'll be in three months, six months, and nine months.
Again, I think, if we didn't feel that we were making good progress and that it was the incremental resources were justified, then, of course, we wouldn't be doing it. We do feel very confident about that business, but I don't have a specific timeline to promise anybody.
Thank you. Your next question: Can you give us a bit more color on how you're working with AI?
Right. I apologize, it's gonna be a bit of a technician's answer. There's not a single standard definition of AI. Some people would include some kind of machine learning and other sort of modern kind of big data techniques, and that, and some would just focus on large language models, generative AI. On that first category, I mean, that is what this business is built on, right? The ability to take very large data sets and come up with sharp outputs that couldn't be created by a human. I mean, that's very much what the entire business is built on, where we use quite a bit of the, that narrower definition of AI, kind of generative AI, large language model, AI.
We obviously use it quite carefully because as all of you know, in the financial services business, explainability, compliance, ensuring you don't. You know, reproducibility are all enormously important, and improperly used AI can be quite problematic in that front. The way we think about it is, you know, a core of what we do is creating risk models both for ourselves and others, and then applying additional analytic infrastructure to translate those raw risk outputs into very concrete, commercial, practical decisions, and to do that in a fully automated way. We have this theory of what we call the control point. The control point is the point at which humans fully observe, understand, and approve every single thing that is happening, every piece of logic.
That usually comes as you are finalizing a raw model, either to use ourselves or to ship to a Carrington Labs client. From that control point onwards, we only use deterministic techniques. We'll use reproducible, machine learning, you know, algorithms and, you know, code and, you know, other statistical techniques. Upstream of that control point, you can use anything you want, right? You can use as much AI creativity, for the feature generation, for any kind of like, ideation, fast cycle typing, prototyping, all that sort of stuff. Whatever is happening up here, which could be unreliable, crazy things could happen, who knows?
It will all come down to that control point, at which point we will be 100% confident that it is robust and reliable, and it won't go through unless it meets that as well. The AI does live, generative AI lives upstream of that control point as well. In terms of practical benefits, again, I think our core ability to do what we do at a fraction of the default rates of many others sits on that, those twin pillars of, you know, machine learning and big data modeling, as well as of that upstream capability in AI. That flows through into much faster cycle times for developing a model, and particularly around feature generation.
I literally have code running right now, Databricks, which is building and testing a new model while I sit here talking to all of you. So much faster cycle times, but you also just do get sharper results. If you can generate, you know, 400 feature kind of new, manually coded features overnight with AI and test them, you will get better stuff than if you're doing it by hand at 1% throughput. Both from an efficiency and effectiveness point of view, it's possible.
Thank you. Your next question: The half year results states, "The group has begun charging interest on a small subset of Pay Advances." Could you please elaborate?
In line with that comment, there are some pay advances which will now pay a small interest charge, as well as the 5% origination fee. As you can see from the results of the half year, when you look at volumes with revenue, it is not material at this point to the overall Group and the economics of the business. Yes, we are charging interest on some advances.
Thank you.
Perhaps just something else to note there is that the Pay Advance product is issued under the short-term credit exemption. Under that exemption, the cost of the product can go up to 5% in a flat fee and up to a 24% APR. Our average duration is just under a month.
Thank you. Your next question: People who work, often take their retained earnings and invest in term deposits at 4%, so equity seeks a return of 4%. Why must Personal Loans be funded with debt? Equity is not automatically more expensive than credit. It sits on the same risk-return curve.
That's a very sharp point. Do you want to talk about equity-debt trade-off, or I can?
At this stage of our business, as we lean more into Personal Loans, as I spoke about before, we were looking to increase that debt facility size. Of course, we'll look at different options over time and do the ones that make the most sense for the overall business. At this stage, we haven't ruled or not ruled anything out in terms of those different options. Roy, you want.
No, I think what I'd say is, first of all, I really like the question. It's a very smart question. As you're probably all well aware, we're highly analytic people, and so we always think about these things from an analytic point of view. Right now, we are actually funding our loans through equity. That's one of the reasons why the drawdown in top cash. If you look at the total borrowings right now, AUD 30.7 million, obviously, if we were to fund that out of equity, that would be a substantially different approach to the balance sheet, and it's one that I don't think pencils out, given where we're from right now. Yeah, we always, we always do think about these things by a very similar way, and conceptually, I don't disagree with anything you're saying.
There is always a break, a crossover point in these things.
Thank you. Once again, if you have any questions, please type them into the box on the screen. Looks like we don't have any more questions coming through. That concludes the Q&A session. I will now hand back to Jamie for any closing remarks.
Well, just, thank you all again for being on this journey with us. I think, very pleased where the group is, and I think this half was a very strong result, and particularly the ability to show that significant profit expansion while investing in the business and progressing our strategic objectives. Really happy with where we're at. I think we feel very confident about the future. We appreciate all of you and look forward to seeing you again soon. Thank you.
Thank you to everyone. You may now disconnect.