Profit, operating profit in the quarter as the cost program's delivered. Our operating expenses decreased about 11% to AUD 8.9 million, AUD 1.2 million lower than the prior corresponding period, with employee expenses down 17.5%, and hosting costs down 9.6%. Our GMV for the group increased 13% to AUD 244.6 million. In U.S. dollars it was up 22%, so there was a strong currency effect in the quarter. It was led by Escrow.com and Loadshift.
The group revenue, however, declined to AUD 11.1 million, which is down 10.8% in U.S. dollars, which is the main operating currency of the group. Group gross profit declined 16% to AUD 9.6 million. Lower cost offset more than half the decline in gross profit. EBITDA was positive AUD 0.7 million. The group recorded an operating profit of slightly positive compared with a slight loss in the first half of 2026, and a positive net profit before tax.
Cash was AUD 16.5 million as of the end of the quarter, and the group remains debt-free. We achieved an all-time record quarter for Loadshift, and Loadshift is also profitable at this point. On a run rate, that's in the seven digits. Escrow.com grew GMV 18.6% or 28% in U.S. dollar terms, with a rising new star category being wholesale electronics, which is now 12.8% of GMV. That looks like it will continue and end up being our second-biggest category. Freelancer revenue declined 26%. I'll talk about the segment performance. This continuation of the last quarter has issues, and we are working through those issues as we speak and I'll address them in a second. There are some positive signs on that section. Skipping down to the group, there's the segment performance.
As I mentioned, the GMV declined 27.6%, which is about 21.7% in U.S. dollar terms, and revenue declined to about AUD 6.8 million, about 20.5%. This was a continuation of what happened in the first half, a slight slowing. Look, it hasn't been a great quarter for Freelancer, or it hasn't been a great year for Freelancer. We've been knocked around by really three to four issues that have been related, but we are slowly getting grips on each of them. The first issue I wanted probably to highlight is the second issue here, which is really on platform stability and organic search. Through the course of 2024, 2025, 2026, we've seen ever-increasing search traffic, primarily from AI scrapers, that has increased about 1,300% over the period. That caused, particularly in end of 2025 going into 2026, quite a number of stability issues.
We know that the customer acquisition of the business is very sensitive to bugs and platform stability. I'm pleased to report that we have largely solved the platform's stability issues in the last quarter. Uptime has been very good. At the same time, the deindexing that we suffered, as a result of both this as well as remediation efforts to stop the scraping, has rebounded. Google is now indexing about 5 million of our pages, up from 1.2 million at the start of the quarter. The traffic quality is not as good as it was before. This does continue to rise, and we are one of the most heavily linked-to sites on the Internet. So, fingers crossed, if we continue the remediation work that we're doing, and we're also enhancing so our logged-out surface, this will continue to rise.
We're also, as I'll explain a little bit later on, expanding into quite a number of new, highly targeted, high-value, sophisticated, and very interesting verticals, and that's also increasing our SEO surface. That issue I think has been stabilized, addressed, and remediated. The second big issue I should probably talk about is payments. As I flagged, in the first quarter of this year, we had a fairly disastrous migration on one of our credit card gateways. Which basically was a migration that one of the gateways asked us to do that resulted in a number of our merchant IDs being consolidated, where some of those merchant IDs had paying existing customers and didn't migrate across properly. We were actually unaware that this migration was actually going on.
As I mentioned in the last quarter, as soon as we got wind of it, we jumped on it and we've been working very aggressively with the payments team to peel off various flows from those gateways. I will say that when you look at our main U.S. credit card gateway, the business of Freelancer looks very healthy. There's a certain amount of initial payments coming in, and there's a growing recurring payment volume coming from that gateway. It looks like a very healthy business. Customers are returning, and the customers are spending more. This particular gateway looks fairly catastrophic, that I'm talking about here. We've since pulled off pretty much the vast majority of the volume from that gateway, and we're deploying new payment gateways, new merchant IDs, new alternative payment methods. For example, in the quarter, we've launched Google Pay towards the end of the quarter.
That's currently in the final stages of the A/B test and looking reasonably positive. We will, really as our top priority from here, be focusing on flow by flow by flow, really getting these payments to look like our main U.S. credit card gateway, which I do say looks very healthy and actually would show a growing business in Freelancer. This is something that we've really got to put our heads down and work pretty hard on. As we speak, we're in the process of commissioning Indian payments. We have a fresh merchant ID for our U.K. payments, which we're in the final stages of a migration of APIs across to before we'll go live with that. We have other flows currently in the works. We are in the middle of this.
We do think we've got a strong handle on what went wrong here, and we are in the midst of rectifying all of this, and I think we've got some stability in the payments and certainly with the new payment methods that are being deployed, we see some growth. This I think is well underway and we just have to put our heads down and continue working on that. The last major issue is in terms of platform integrity controls. We tightened up the authentication of the site, with two-factor authentication, over the course of the latter of 2025. With increasing authentication, you obviously have friction issues. We've been progressively tweaking that friction away. I think we've optimized it to the best we can have done. We have lifted quite significantly the security of the platform in the process of doing so.
But, this combined with all the other things did lead to some churn which we're now trying to fight to bring back. I am not happy about this. It has not been a great really 12 months. I do think we have got some stability. I think there is some light at the end of the tunnel. I will again, point out that, when we look at our main credit card gateway for the U.S., it does look very healthy. Customers are returning and it is in a position of growth, and we just need to get that working across all flows and just continue to optimize the core infrastructure as we launch new product, new features, et cetera, and continue to grow the business. While I think we have provided stabilization to the platform, we have got to now work to bring the revenue into positive growth.
I think we are heads down about it. The quality of the marketplace does continue to grow, in the underlying features and the improvements that we make. We have quite a number of A/B tests that we have been running over the last number of months, which do show some underlying improvements in things like award rate, funnel conversion, and so forth. Average project size increased about 25% to $498. Projects receive about 63 bids on average, so liquidity is very strong. AI-related work has grown to 7.7% of GMV, up from 5.7%. Sorry, project GMV, and at 7% in U.S. dollar terms. Employer-rated freelancers are up. Freelancer quality is up to 4.83 from 4.78, and we do maintain the highest rating on all the major sites like Trustpilot and Sitejabber and so forth, compared to our peers.
So, the customers do love us despite some of the core infrastructure problems we have had. In the enterprise division, we are doing more work with NASA. We are doing something in robotics this time. Robotics specialists supporting digital twin simulations of multi-agent robotic systems with applications for lunar and Martian exploration. Some of you might remember we put the fifth astronaut on the Artemis mission, being the Rise, zero gravity indicator for children's education. We are continuing to work on the human genomic editing challenge, which is a multi-year, I think it is about $7.5 million now, challenge, with four or five reagent tests complete. The final testing is scheduled for later this year. Our Data S-Index Innovation event was held in July, where the four winners advanced, and with our boom trajectory unknown challenge for the distribution of particles and explosions also was resolved.
And we also worked with the United Nations Development Programme to mark underwater explosive ordnance. So, I think you can see by the quality of the work we do there that we do very high-end sophisticated work on Freelancer that is very hard to do anywhere else on the internet. In field services, things are starting to move a little bit. We have got a field sales pilot for a very large trillion-dollar global payments provider. It began in July. The sales team have really boosted their cadence in terms of sales funnel activity. In September, we are onboarding quite a number of merchants for this particular payments provider. In the U.S., we signed a service agreement with a medical device manufacturer for on-site work in medical care, and we are starting work there as we speak.
We've also built two products, in the process of doing this, for managing and dispatching field workers. Something that would've taken us years to do in the past, we managed to do in the quarter. Also another product to run the field sales team, for the big payments company. That's starting to get some real progress happening. The enterprise also did a number of work for other companies. It's a very large trillion-dollar e-commerce company. We do research and localization and video production for, et cetera. There's some work we did there and other BPOs and so forth.
Look, I'll talk about in the outlook in a second, the future, but really this has been just working to get stabilization, work on the core infrastructure of stability, which is done, search, which is lifting, payments, which I think is somewhat stabilized and growing now as we deploy new payment gateways and the platform integrity controls where we had churn. We've optimized as much as possible and we now just focus on growth. Press grow. We're up 28% in terms of GMV. This is over the summer quarter, so this did pretty well. Revenue was basically flat, because we were working on larger value transactions rather than smaller value. We think that will uplift in the fourth quarter. It's just you do have a bit of a slowdown over the summer holiday period with activity in terms of count.
But we think that will lift in the fourth quarter. Certainly the GMV was quite up as we power into a very strong and I think very large new vertical of wholesale electronics. That's about 12.8% of our GMV in the quarter, and it's our fastest-growing vertical. We've had quite a number of meetings in this vertical, which is new and very, very large. We've got a whole dedicated workflow that we're planning on deploying. There are also some other interesting verticals that are popping up, such as data brokering and so forth, for the AI companies. On product, we migrated to a new platform, which we are expanding. We've activated JPMorgan Chase for domestic processing and rolling out international U.S. dollar processing. We have quite a number of innovations there to try and get the mantra of fast, secure payments very fast.
I think with a number of initiatives in the real-time payment sector, which we now support on our accounts, and we're rolling out progressively over the fourth quarter. I think that will provide a phenomenal experience. Wholesale electronics and domain names will be our main drivers of growth in the fourth quarter, and we'll be attending and sponsoring one of the largest conferences in that sector. Loadshift delivered the best quarter on record. Revenue increased 18% to AUD 1.1 million. GMV increased 16.9% to AUD 8.9 million. September was the strongest month to date, doing about AUD 380,000, up 13.7%. GMV of AUD 3.1 million, up 9%. Growth was driven by higher marketplace activity and conversion, and loads awarded increased by 14.8%, and the award rate also lifted and I think will continue to lift because we finally deployed full PSTN and in-app calling.
We now have, across all the modalities, calling going through our system. We lifted the call connection rate as a result of that to 93% from 21% on estimates only. We have some optimization to go on that. As you can see, the award rate, as I've always been saying, started to lift as a result of this. I think we can continue to lift the award rate, which would lift the monetization of existing flows through the site, in addition to growing more loads on the platform. We've also introduced a new way of loads being found and published by carriers, which is basically a big map where truck drivers can basically put their return legs or put empty legs up on the site. We're in the process of making this automated, which allows a huge amount of liquidity to be surfaced.
In the early days of FedEx, the innovation that happened was the planes used to only fly when they were booked. When FedEx came along, the planes flew empty or not, and then they auctioned off the spare capacity, and that opened up a whole new sector in low-cost air freight. We're basically doing that for trucking in addition to how the platform works today. I think that's pretty exciting and something to look forward to in the future. The monetization continues to improve. Paid memberships increased, and we've got Enterprise membership and so forth. We're also awarding larger and larger jobs. There's a 300,000 job, which hasn't been reflected in the third-quarter results, but will be reflected in the fourth-quarter results. We've got a number of conferences we're also attending, et cetera.
I want to talk a little bit about outlook before I open up to questions. Of course, Dylan's on the call if you want to ask him more about financials, and I've also got Andrew here about product. The main thing we've got to do for Freelancer is we really need to focus on payments. We've really got to, as I said before, our U.S. credit card processing looks very healthy. When you look at it's a growing business. Customers are returning. That's growing. We just need to get all the flows looking like that. Google Pay literally is being commissioned as we speak. India is in progress. It's been in progress for a while because it's quite complex. We have implemented the eMandate, which is a way in which cards can be charged when someone's not present for things like memberships and so forth.
We have an integration with one gateway, and we have an an alternate payment provider which is being integrated now for things like RuPay and UPI. There's quite a bit of bureaucracy in that particular space, but I think that that will deliver quite a strong benefit because we have quite a large Indian user base. We have the U.K. payments being peeled off. We have in-app payments coming for Apple Pay and for Google Pay, et cetera. Really, we're just laser-focused stream by stream by stream at getting those payments up and running, because that's a big thing. There's probably going to be some questions about AI, because I know everyone, it's a hot topic. My thoughts on AI are well-published.
There's a number of essays on Medium and a number of guest spots on Macro Voices, which is the number one macroeconomics podcast in the world, which, if you go to macrovoices.com, you can listen to it to hear my thoughts on it. I think we sit in a fairly unique place in the world. On Freelancer, while there are quite a number of types of work that AI can do, and AI is very good for getting things started and for imagining and so forth, what we increasingly find is AI is great for that, getting things going, but you very often need humans to make it real and to get the work done. We have the largest workforce of humans in the world, about 90 million. The other main benefit of Freelancer is we're pay per outcome, not pay per token.
When you get a job done on Freelancer, you're paying fixed price most of the time. So you know what you're going to get, and you know what you're going to pay for it, and you don't release the funds because they're held in milestone payments until the job is done. It provides quite an attractive way of getting work done. Of course, all the freelancers are AI-powered as well, so they have access to all the tooling. One of the big things we need to do is we need to surface which tooling the freelancers have available better to the clients, which we're working on now. The skills of all the freelancers now obviously being enhanced by AI has really lifted the quality and the value of the marketplace.
Now you've got a lot of highly entrepreneurial, highly skilled professionals that are already priced in the best model you could think of, which is fixed price, pay per outcome, but also with the benefit of being able to access, not just human skills, but also AI skills. Not that everyone wants AI. There's a lot of people that actually specifically don't want AI-related work and make a very big point of that for various reasons. Either they don't want AI work, or they don't want AI to be sucking their content into the training data of the models, or they want a human to talk to, et cetera. A contest model, I think is also. We've got a number of optimizations coming into the contest model, which I think is going to be great and very AI-proof.
Effectively, it's a competitive marketplace for getting work done, whether or not it's human-human augmented with AI or AI-related. When that gets deployed towards the end of this year, I think people will see some pretty exciting things and pretty interesting outcomes as a result of that. It's very key for us that we should build the bridge to AI. I think many times people get stuck in Claude or what have you, and go in loops trying to get things done, and it's great for punching something in and saying, "Oh, I want to build a website." You get a nice-looking page, and then you realize to actually turn that into a fully-featured site that goes into production, et cetera, that actually works and has ongoing tweaks and ongoing maintenance, et cetera.
Either you are sitting down there trying to hammer away at the keyboard and paying your $200 subscription or whatever it is, having trying to figure it out, and not everyone is capable of doing that. Or you might end up in a bit of a loop. We have got to build that bridge to AI. Whether it is by API or MCP, we have got both efforts underway already. We already have a beta MCP server that we are testing at the moment. And we really think that we could be the operator please for AI.
When people get stuck in Claude or Figma or whatever it may be, or ChatGPT or any of the software packages that are out there, whether it is a CAD package or whatever, and you want to get to a human operator, you hit the button, operator please, and you go to a human professional on demand, fixed price or pay by the hour. And I think that is a great marriage and I think a big customer acquisition channel in the future for us that we are aggressively working towards. It is also key that we work on. The great thing also about technology is that while there can be temporary dislocation, over time, whether it is the mechanization of agriculture or the mechanization of the factory, the information economy and the transistor and so forth, and software and internet and mobile apps and so on.
Ultimately, what we have seen is that more jobs get created than get destroyed over time. But what happens is with these new jobs is the productivity gains are so large that what you can get done, they are highly deflationary, these advances. What you can get done in terms of bang for buck leaps by orders of magnitude. And so it is key for us that we enable more sophisticated and more complex offerings to be done on the site. There are also quite a number of jobs that cannot be done by AI. Obviously, a lot of the physical jobs, et cetera. There is just a lot of things that literally that I think AI struggles with.
I think for all the things that people say design work may be at risk with AI and there is flashy things popping up on social media all the time about, oh my god, motion is dead or design is dead or what have you. I do not think AI has really cracked that yet. I think one thing that humans have is taste and judgment about trends and so forth, that I think is actually quite hard to supplant. Internally, when we look at automating a lot of the work that we do internally to make the benefit of AI. Where AI is most applicable is things like queue processing, whether it is answering support tickets or processing requests that come from customers in various form factors, for manipulation or processing or reporting or what have you. AI is very good for that.
But we have seen it really struggle in trying to be at the bleeding edge of, for example, design quality work and so on. The key thing is we've got to work on enabling more sophisticated and more complex offerings. We're going to be launching something literally in the next couple of weeks, as the first step towards doing that, where you'll be able to get something built that you would never be able to get done before online, through quite a neat product offering that will continue to get enhanced. I won't take the thunder away from it, but that literally is in its very final stages of testing and in with customers now in the beta and about to go live. We're also rolling out quite a number of vertical offerings, to really build focused, value add, interesting, contemporary, sophisticated, high-end vertical niches.
Whether it's a bit like eBay did with cameras, cars, collectibles, computers and so forth. They had category managers and really focused on those categories. We're doing the same thing and we're launching a range of categories, whether it's humanoid robotics or electric vehicles or anime and so forth. So, there's quite a number of things that progressively come out there, and we really have a factory set up to do that and rolling out. We have quite a lot of things we're doing around collaboration and sharing. We've been incrementally releasing some of that over time, which has been a little bit masked by some of the greater issues we've been dealing with. But I do think that's going to lead to some big improvements on the customer acquisition front as well as the retention front with all the teams' work and so forth we're doing.
As I mentioned before, with the contest offering, I think we have something that's very future-proof and very exciting and works well with or without AI on the contest front. We've also been automating a lot of the back end, and we have been doing obviously a lot of work on the cost control. We actually have a fair way to go still on the cost control. We've got quite a number of areas which we're chipping away at and cost control will continue to flow through. Then really, probably the last thing I want to talk about is on the acquisition front is really just on the channels, which I mean by the customer acquisition channels. I do think AI is going to be a great new customer acquisition channel for us, as I mentioned before, as the bridge to humans from AI.
I think we offer a fairly unique proposition with all the largest workforce in the world of humans, that we have ratings and reputation and so forth too, and bridging to help really make it real, once AI gets things going. On the channel front, we did get knocked around. I think we're seeing some stabilization. On the SEO front, I mentioned before, we kind of had a bit of a mass de-indexing event where we kind of rate limited the crawling of our site by some of the AI scrapers, which were also using the same IP addresses to do search indexation. As I've mentioned here, that's bounced back quite significantly. The quality of the traffic isn't as good, Amelia, as it was before, but it does continue to lift. That will continue. On the search engine marketing front, we've made quite a number of optimizations.
There's been a little bit of cost control while improving the return on investment there. I think we have quite a lot of new avenues to explore there over the next quarter or two, which I think will deliver some value to the site. Overall, it's really, two out of three businesses are going okay. One business we just need to work at. I think we've identified the issues and we're just aggressively in there this quarter. We're giving a bit of an update, but hopefully in the next quarter we'll show you some real progress being made there. I will now open it up to Q&A with the audience. As I mentioned before, Dylan Carter, the Chief Financial Officer, is on the call, calling in from the U.S. East Coast. So thanks, Dylan, for making yourself available. I have Andrew Bateman here from Product.
I'll open it up. I think, Dylan, you're going to be doing the hosting for this.
Yep. If there's any questions, please feel free to put it in the chat or—
We can't hear Dylan, so I think there's audio issues. Maybe, Kurt, you'll have to do the hosting then. Any questions in the audience? Not currently. Don't be afraid to ask a question. If you want as well, you can always email myself at matt, M-A-T-T, @freelancer.com, Dylan at D-Y-L-A-N@freelancer.com or investor@freelancer.com. Or you can post your questions in the chat. No questions? Okay, I've given us 60 seconds. If there's any questions. No? Okay then. Well, thank you very much for calling into the third quarter results. I'll see you in the full year results. Thank you very much.