One97 Communications Limited (PAYTM) Earnings Call Transcript & Summary

So the operations of PCTL are there, but not very significant. PCTL is an entity through which we make investments in some foreign ventures. So for example, we had launched -- we had announced the setting up of a business in UAE. We have announced the subsidiary in Luxembourg. So a lot of that is done through PCTL. So I would think of it as largely -- it is an operating company, but largely -- use of proceeds in that company would largely be funding other overseas venture, including -- including funding needs of existing overseas ventures. So that's the plan, but we don't have a firm number on it, which is why we have taken an enabling up to INR 500 crores, and we have taken it for up to 12 months that we don't have to keep kind of going back to the Board for the same type of approval. Of course, each investment, if any, will be approved by the respective Boards and will be disclosed if it's material.

Sure. So Madhur, I think if I'm not wrong, you also used to have one Singapore -- I mean, One97 Communications Singapore entity, which used to be your umbrella entity for international business earlier. Now I think you are doing this international business through this PCTL. What is the difference? I mean, because I think you've got that PayPay cap in that entity, right?

Yes. We can take you through some of the nuances offline, Suraj, if you don't mind. But basically, One97 Singapore has been doing international business of some of our sort of old business, which is the marketing cloud business, where we service telecom companies and enterprises around the world. So basically the One97 business as it was called, whereas PCTL is doing largely the merchant payment and merchant lending partnerships, what we have described in previous press releases as partner-operated and Paytm operated, merchant payments and upsell to lending type of businesses.

Sure. And last 2 questions very quickly. On this AI platform that you have started offering to merchants, once you start monetizing this whole business, in which line item of the P&L the monetization part...

Cloud and marketing, marketing cloud...

Marketing services. So amongst the 3 lines, it will come in marketing services. Of course, the nature of that will evolve over time, but vast majority, if not all, of this will be in marketing services.

Perfect. And the last one is on FLDG. So while I think we have enough discussion on FLDG, I understand it is P&L neutral. But just from business point of view, why these lenders do flip flop. So for example, I think you're saying from next quarter onwards, one of your large lender would be doing FLDG again. I think 3, 4 quarters back, they started doing FLDG, then they stop, then they again coming back. So just from, let us say, business decision point of view, I mean, why lenders choose FLDG or not? I understand it is probably best for you. But because it's also a recurring thing, right? FLDG works like a current account or maybe, I don't know, something like margin trading funding from an operational point of view. So when somebody does flip flop every quarter, it looks like there is some decision factor. Can you throw some light that what is that decision factor?

Yes. So this lender used to do FLDGs back in FY '25. At the beginning of FY '26, we had a discussion that they said, it's not really that important for us to take FLDGs. So we requested them not to. So they said, "Hey, for 2, 3, 4 quarters, we won't." They were nice enough to not take FLDGs for 6 quarters while scaling up business with us and being one of our longest-standing and deepest relationships. So we are very appreciative of the fact for 6 quarters. What was meant to be maybe a temporary pause ended up being 6 quarters, right? So it's not a huge surprise to us that at some point, they would want FLDGs again. Like I said, we don't -- we are largely indifferent. So we don't like challenge them to -- on all the details of it. But our sense is it may have also to do with certain RBI guidelines with respect to they can offset that against credit losses and so on. So then it is more useful to them in the current construct than it may have been in previous constructs, right? But like Vijay said, like this is not a top thing that we negotiate with lenders, right? In fact, I tell the team, this is not a top 5 factor for me. Do more business, do it with the right partners, do it obviously at the right credit quality, make sure we are investing in collections, make sure we are building funnel for more and more lending down the road, don't be focused on 6 months, focus on 2 years, 3 years, 5 years, build great technology. Those are things that we focus on when we are thinking about which partner will scale up over time. DLG is way lower on the list. Of course, because we have conservative accounting around this, it all kind of comes upfront. So in these conversations, it's something that we have to highlight. And this time, we have taken sort of opportunity to flag it not at the end of the quarter, but at the beginning of the quarter, right? Of course, the quarter that we are discussing right now, there is no impact of that, but we have taken the opportunity to just kind of flag it ahead of time. So there are no surprises at the end of the quarter. So in these conversations, it's quite relevant. In vendor conversations, we don't pay as much priority to it.

We will take the next question from Pranav followed by Jayant Kharote from Axis Capital. [Operator Instructions]

Two questions. One is on the sales team additions. You had a [indiscernible] addition to the sales strength. But your device additions are largely flattish on a year-on-year basis. So is this sales additions in a different area? Or is it largely towards the same Soundbox additions? That's question one. The second question is on the lending business. Do you have a sweet spot in terms of GMV of the merchants where you would like to lend either based on EMI to GMV or EDI to GMV or any other such ratios. And if you have such a range, how many of the merchants -- device merchants do actually fall in that range because you have 16 million, but how many would fall into a lendable category? So those would be my 2 questions.

Yes, awesome. The first question of [ sales additions and executive ] are in the same category. Pranav, they take time to get mature. We sort of train them and we first give them retention jobs, revisits in the markets, if you will, and then we give them deployment jobs. So it's in sequencing. So I mean, we've been able to recruit because we've been pending to recruit for a long enough time and quick commerce has been taking up a large number of people of these kinds. So that's why you will see hopefully that we are adding more number of Soundbox merchants. And it is also that we are rejuvenating the relationship like you asked for. One thing that we are more bothered than the GMV is about aging on our platform because a lot of time, the shopkeeper may be taking cash, may be taking another way of payments, but the fact that he's retained on our platform gives him more probability to get a loan from our lender partners than the dollar value, which is large. Even if you look at the average ticket size of INR 2 lakh we're talking about even INR 1,000 a day means INR 500, which is our threshold that the person should be able to pay every day. So INR 500 [indiscernible] INR 365, that is the kind of number the ticket sizes are. So the threshold is not based on GMV, but rather based on aging and GMV, some element will definitely be there. If you're earning, let's say, INR 100 a day, you're not going to give you credit there. By the way, we could give you postpaid there. You remember that. So not the term loan merchant cash advance. How many merchants are creditworthy? That's a very interesting one. Because if you notice, we have not dramatically added the count of merchants because I've told my team, it's a business model of, like I showed you in the beginning, payment and credit -- payment and credit. And that means that we need more merchants who are creditworthy than the merchants that are just signed up. Again, the quality, quality, quality of merchant, merchant. Quality equal to bottom line, bottom line equal to financial services is a bigger bottom line than the MDR or subscription, and then that means retained merchant is better valued and so on. So while we may have acquired a lot, we just removed the merchant whom we don't see active or transacting. So while you're seeing the net active, the gross actives are bigger and that is why this number looks that it is less or not. But at the same point in time, all said and done, the net active that you are seeing net active monthly, which is what we report every quarter, will grow up, definitely will grow up, I mean I personally see on a base of 15 million, we believe that something in the plus of 10 million are definitely a good quality merchant.

Okay. So at least 2/3 you think would be customers you would be happy to -- merchants you would be happy to lend?

It is all based on the relationship and aging. And I do believe that I want even more than 2/3 to go on than to acquire new customers. That was the pitch to my sales team.

Okay. So would it be fair to say at least 2/3 have INR 1,000 per day or more in terms of payment?

I don't think that is a number that I will often tell, but I do track literally the same number that you are quoting lendable merchant, if you will, and 2/3.

Okay. 2/3. That's good to hear. Last one minor question, if I can squeeze in. On your indirect sales cost growth, there's a big spike in the software and cloud expenses. Is that a one-off? Or is this a new normal for your expense rate?

Well, we have -- sorry, go ahead.

We have commented on it. As a part of some of the migration of merchant payments business that we did to our subsidiary, there is a little bit of platform redo that we wanted to do in terms of consolidating certain platforms, which does have some costs while you're doing that project. So that was largely in this quarter. There's a little bit of that in the next quarter. So there is that sort of onetime exercise. I should also point out that we are spending more on AI, right? We are seeing huge amount of gains and huge amount of ROI, largely in indirect expense and sales growth because of the spends that we are doing on AI. So given the nature of our platform, so for example, if I am retaining merchants better, it will show up in revenue, right? If I'm doing better collections, it will show up in revenue. And there's also a tendency that, hey, if you reduce the cost of doing something, your one instinct is to try to do a little bit more of that activity, right? So you may not necessarily see that as a cost reduction, you may see that in revenue growth. So the combination of revenue growth, which is, like I mentioned, like-for-like now in the 30s and indirect expense overall being very much under control despite the sales and service investments we are making despite marketing costs and despite AI costs, we are seeing the goodness of that, right? So we monitor it very, very closely, but we are actually okay to spend more on AI costs as well. So there is some element of that. Once that -- I think maybe by end of next quarter, we'll be able to kind of start talking about a slightly more normalized number, both for increase in AI costs as well as putting this onetime project behind us.

I want to specifically tell you, Pranav, very interesting fact. I personally look at the cost and the output of AI at a different level than, let's say, Madhur himself as a cost control will look at it. I look at it yield of what shipped. So not just what got created, what shipped, what earned revenue and so on. And I'm very happy to tell you that what we have made, obviously, slowly, slowly, revenue doesn't show up immediately. Products are loved and liked. You're not spending so much, you're growing. All those advantages are very visible. In this quarter, when I looked at the cost of cloud, majority was because online business was in an entity, offline business was in an entity. We brought into single entity. The platforms were to merge, so there were duplication. And like Madhur said, a little bit of next quarter also, it sorts out. So it -- actually majority cost was this migration -- majority incremental was migration.

Understood. So essentially, there is no need for indirect cost to grow at an accelerated pace for the existing business. So obviously, if you have ways of earning revenue through AI spends or cloud spend, then that's obviously great.

Very well put. Very well put. It was very clearly because majority of cost of incremental cost on the cloud was attributable very completely towards this migration of online, offline. And now that brings us omnichannel payments just in case. Same machine learns the route.

We will take the next question from Jayant, followed by Kapil Sharma from Jefferies.

Congrats, Vijay and congrats, Madhur, for another great quarter. I'll kick off on the MDR on UPI. This 2.5 basis points, if you could give a broad split between consumer and merchant. And where it's coming from is clearly, merchant, you guys are going to enjoy the fruits. On the consumer side, there is a worry of this pool getting competed out, especially for the top, let's say, 2% to 5% spenders who will do more than this 2,000 spends. Are we going to throw our hat in the ring and willing to sacrifice a little bit of that margin? Or given what we have done this quarter, I think rightfully focusing more on the profitable GMV? I mean that...

I would say, first of all, this is not implemented yet. And number one, so there's -- one can get into many, many different types of scenarios. I think what we have put forward is a number that we are confident of delivering. And like Vijay said in the response to an earlier question that there are certain scenarios under which hopefully, we'll do better than this, that obviously, when we're putting out a number, we don't put the high end of our number. We sort of haircut it for various scenarios. But specific questions about, hey, whether we should be assuming that there's more chances of being competed away on consumer side versus merchant side, what intensity would there be? I think we'll only find out in quarter 2, quarter 3 after implementation.

And important thing, Jayant, is that let's be very open and transparent about it that consumer business is ours to win. There have been, I would say, non-internal, not just led by us, issues that has made us have so less UPI market share of consumer, if you will. Our product and capabilities are continuously moving up. And internally, we are far more, I would say, expecting, hopeful, sure of our market share win than we were last year and last to last year. So it's not something that the result is out and we are there. We rather want to tell you that we will change the result with our hard work and perseverance and pushing as much possible even on the consumer side. You said it already that merchant side, we will get the win, but consumer side is for sure us to win.

You want to highlight, Vijay, what is the external thing? Because the numbers are very strong that you've seen, or maybe I can take it offline.

We take it offline.

See Jayant, yes, my intention is to say that there are a lot of factors of -- because we are part of a network that we have to align and work and get all the network pieces all together. And the more you dig, the more you find optimizations possible. So I just want to say that erstwhile we were not so much focused on the product and technology as much we are and the results are showing up.

Okay. Second is in the MDR GMV that you have assessed, which is, let's call it the eligible GMV, is that growing slower or faster than the overall GMV? And what I'm trying to understand is once we set a new take rate in the new world, does that take rate expand from there or stabilizes or maybe comes off?

I think the mix is not changing dramatically because obviously, where Paytm has unique strengths is in the small merchant business, which is growing very, very fast. And for a different set of reasons of enterprise business and the online business is also growing quite fast, right? Online, as you know, we were under embargo until August of last year. Starting beginning of this year, we are seeing good hunting goodness. And in enterprise, we are improving because of EMI offerings, DCC offerings, greater distribution, et cetera, et cetera. So we're gaining share across the board. So I don't think there's a huge secular trend that I can point to that we are at X percent eligible GMV, and I expect that to go up by 10 points or go down by 10 points over the next couple of years. I also don't think that will be a major factor. The major factor would be just executing as well as we can for our merchants.

What would have happened last 12 months?

To be honest, like I've literally not even bothered to check that data. I know what the data is as of this quarter. And I know kind of -- I know that as we roll forward that data, this is -- the mix effect is not going to be a major factor.

Understood. One last I could squeeze in on the DLG accounting. While Madhur you've explained this multiple times just because we want the models to be clean, contribution margin shrinks, then recovers as we reach the tail end of the portfolio and the lumpy payments come out. All this while EBITDA margin keeps expanding at a normal pace. Is that a correct understanding?

No. When contribution margin goes down, there is a drag on EBITDA as well, of course, in that quarter, right? So when you go from -- let's say for a second, I'm giving 2% to a partner. if I'm going from 0% to 2%, that 2% shows up in direct expenses in that quarter. And there's a little bit of additional revenue in that quarter, but most of the additional revenue comes over the duration of that loan. So that impact on contribution margin does slow down on EBITDA as well. What we have said is there's a huge amount of tailwinds in the businesses overall, including in the merchant lending business that we don't think that we're looking at EBITDA going down next quarter. We do think EBITDA will continue to go up this quarter despite this drag.

EBITDA absolute going down or EBITDA margin going down?

I'm saying the isolated effect of giving DLG is a drag on EBITDA. But on an overall basis, given the huge momentum we have as a company, quarter-on-quarter, we'll still expand absolute EBITDA.

Jayant, EBITDA margin [Foreign Language].

Great. Sorry, Madhu, just because I will use this last. EBIT margin, depreciation going up, CapEx going up, how should we think about that?

I think depreciation year-on-year hasn't gone up very much. But like I said, we have done more CapEx in the first half than we did previously. So with the drag in FY '28 and so on, there will be more depreciation. But it is all being done for a specific purpose, which is higher engagement, higher retention, higher financial services revenue. And we are only doing this because we are seeing great payback of putting in that investment, whether it is in service cost, whether it is in CapEx on the eventual financial services and other monetization funnels.

Perfect. Congrats once again for a great set of numbers.

Thank you very much.

We will take the next question from Prakhar, followed by Gaurav Rateria from Morgan Stanley. Prakhar, you may go ahead.

Sure. Great set of numbers. Good to see high-growth companies. Just quick questions. One thing which really impressed -- surprised me, your monthly transacting users. On a quarterly basis, you used to add 1 million to 2 million customers a quarter. This quarter, it's 5 million. Fabulous numbers to see. But when I tie it in the fact that you are far more focused on quality customers, of course, this is total users. One, how does this tie up? And is this a run rate that can sustain over -- for -- at least for some quarters and then it normalizes? That's the first one.

I wish this sustains for sure for a few more quarters and more quarters, number one. And like I answered in previous question, and in the beginning of the call itself that it is one of the most critical KPIs that I personally carry, the number of new users, and users and so on and so forth. So by the way, we've hit internally even 100 million sometimes, even the users on our platform. So I'm going to say that our target is to take 1/3 market share in this business. There is nothing less than that we will settle for.

And if I can just ask you, if there are 2 things that are working very well for Paytm, what are those? Because this run rate going up at such high level...

Good product and technology. The industry has not seen good product and technology, straightforward. It is good product and technology, cleaner, simpler, good product and technology. And like I said a few minutes back, the more we are digging in, more we are finding out, oh my God, why was this blocked? Oh my God, why was this blocked? And then we go back and solve for it, solve for it, solve for it. Prakhar, let me say this. I'm very clear about it. Paytm has top of mind recall. Paytm has a product. Paytm has a technology. Paytm has rather superior technology. I'm going to say that also very clearly now that I know of what everybody else is doing. I think the amount of attention that we are able to do it is something that we believe many others have to monetize. We have understood what the monetization is. So there is a lot of advertisement and a lot of cross-selling that is attempted. So other products have also started to become, I would say, a reason for our customers to move to us. So in a way, a lot of positives in both ways.

Good. Got it. Second question is, sorry, going back to the MDR on UPI. Just wanted to know like there are a couple of lobbies of merchants who are also trying to negotiate for a lower MDR, which includes the petrol pump lobbies and maybe there are others, et cetera. So when you have talked about the 2.5 bps as additional revenue pool that you might get, have you factored those negotiations or it is as is where is sort of calculation?

Prakharji, we have taken the framework from NPCI as is because again, you end up going into many, many scenarios. I will point out that it was postponed maybe a few days, maybe a couple of days before the implementation go live. So our understanding is whatever those things are all sorted out, had they not postponed it by 1 or 2 days, keeping in mind Diwali and festive season coming up, then it would have gone live on October 15 in the format that they have put out there. And there's no mention that they are revisiting the framework or anything like that, right? So again, given we are doing an earnings call and given this development has happened, we thought it was important. It is unusual for us to comment on something which is not implemented yet. But given it is a major factor and there's a lot of analysts have asked about it, investors have asked about it, we did want to put the estimate out there. My best guess is that the framework will not change right now, but I could be wrong, obviously, I don't control it. But we are reacting to the framework as it exists and is "final."

Got it. And the last part is that on the lending side, you've had a great growth. But when you look at the customer base, if at all, now or even in later, can you give us some insight on how good the portfolio or these customers are? And what are the -- any of the risks that you are monitoring, it could be regulatory in terms of pricing, et cetera, or even any of the cross-sell like insurance, there is this huge paper that is there, if that is of a relevant outcome for you from an insurance attachment that goes with it. So any clarity on that because it's been your fastest-growing revenue pool right now?

So Prakhar, the good news is that we did not have any insurance revenue. So all upside, downside is not belonging to us potentially there. Neither we tagged or we were topping up some margin by cross-selling, tagging along insurance or something. So nothing of insurance. Quality of customers, I think the core answer is that our disbursements are growing per lender count and the volume are growing. And in my opinion, lenders grow once they are happy with the kind of customers they are getting. I mean book responsibility belongs to them and our responsibility is of disbursement and collection. So we ourselves continue to keep an eye that we are collecting good and collecting nice. I think in a bigger macro sentence, I can say that our collections level of details tell us that there is a huge opportunity and festivities are showing up. And we believe that the kind of -- I mean it's an AI language recipe, by the way. It's not a chef language. The recipe that we have built of financial services distribution, credit distribution is something that is enough unique, and I'm pushing it as much into AI so that it is not even humanly dependent internally and so on. So we'll see a lot more. I don't think there are something that we are looking and we are not talking about it. We rather are always discounting in advance, if you know in the past. So right now, there is not much that I don't see anything. Madhur, anything that you want to add?

Just to clarify. So Prakhar, on merchant loan, we do give ECL data. It is largely unchanged in terms of the range that we give. PL, of course, is distribution business only. I do think -- I just wanted to add that we take huge responsibility of being cautious on this business, right? So while on the one hand, we keep trying to improve the funnel and the lendable base, like Vijay mentioned earlier, that is what we count on for the growth of the business, right? We don't count on, hey, we are going to be reckless and loosen the policies and those sorts of things. We try -- and there's a reason why we try to -- want to be very cautious and very responsible about this because now we have over a dozen lending partners. Many of them have been with us for 3, 4, 5 years now. And we also think that we can create something which is a huge agent for financial inclusion if this continues to scale up linearly and at a fast pace for the next 5, 7, 10 years. We think this is a product that can create huge impact to small merchants in India where financial inclusion, employment, investments, so many great things can happen. And like Vijay showed at the beginning of the meeting in the presentation, a model which is also -- we feel like we are pioneers of, right? And we could take to other countries as well because this is a problem which obviously, we're going to prioritize solving in India first, but there's also a problem which is unsolved in many other parts of the world, right? So -- but the starting point of that, you have to be cautious and responsible because it is a credit business. You have to recognize it for what it is. And you cannot afford to have 6 good quarters and then 2 bad quarters, right? That kind of sets you back not 2 quarters, it sets you back many more quarters. So you have to kind of recognize it for what it is and be as cautious as you can be.

And Prakhar, like Madhur said it -- we do not want it to look like there is a future time bomb sitting in our revenue. I mean one thing is very clear. That's why conservative accounting, not letting revenue come in, which could be discounted obligation in future. The best thing I can tell you is we want to be a distributor technology company. We don't want to be the book owner company. We've announced it to the public a number of times. I like financial services business, but not so much that we end up taking the weight of the book on our shoulders. And we will remain a technology platform and a platform business that compounds by consumer and more usage and customers include merchants and consumers, both. So it's a technology distribution business, and I'm very happy to say that SKU that we distribute, if you were to look at it in a slang way, credit is very highly in demand, and it requires logistics equivalent of, if you will, collection.

We will take the next question from Gaurav followed by the last question of the day from Rahul Jain from Dolat Capital.

Congrats, Vijay. Congrats, Madhur. A couple of questions. First, the MTU acceleration that we have seen, is this led by better gross adds or improving churn? It appears that your CAC has fallen substantially. How sustainable is this?

Gaurav, idea is that -- I mean, we've been using these words, but I'm going to repeat the same words that are written. When you start to look at whom do you want to acquire, you spend less money on spreading in a wider customer base. So per customer who is a valuable customer, it shows up as a lower CAC. We have large enough budgets. We want to expand and expand the customer base for sure. We will spend more money on customer acquisition. But like I always say, the valuable customers, not like spray and pray get the large number of customer and hope that there is a UPI GTV, which is converting someday in revenue. So that being the case, how has the CAC improved just by focusing on whom to acquire or not acquire. As much precise that we can become. And who allows it? The AI models allow this to happen. Our models are made by us, our models are not third party. This is a very important sentence. We are not renting intelligence from outside. We make our own models. We make our own system and capabilities. And that is why we've been able to nuancingly give these outputs. And the elusive dream of AI that it will behave N equal to 1, it will behave and deliver impactful outcomes on cost of operating and revenue upside, that's what we stand for.

Got it. My second question is on the investments that is going in the AI part. Of course, right now, it's kind of dragging on the -- as a drag on the P&L, when it starts generating revenue, it will be accretive from an earnings point of view. So how big is that drag right now? What would be the operating leverage that we can think about it?

I wouldn't say it's a drag...

It's an uncomfortable. I don't know how the migration of cloud from other payment system to the new payment system and sense it, absolutely no drag. I mean every dollar that we spend on AI has more number of dollars made than ever. I mean, absolutely no drag. I mean it is not that we are -- either it has accounted for saving more cost than it has spent or it has earned more revenue than it has spent. So Gaurav, absolutely no drag. We are not bulls******* on random KPI call, how many tokens do you spend? We don't even talk about that. We rather are one of the most efficient users. Madhur was invited by Cursor to become part of the global CFO council they made only because they said you are the world's fourth most optimum coding model user in the world. I mean -- so I'm sorry to say this, absolutely no drag, absolutely no confusion. Cloud cost increase is due to our migration of 2 machines running into the -- becoming one consolidated machine and all 3 switched on for some time, and they optimize.

Sorry, let me just clarify that cloud expenses have gone up. There are 2 main reasons for this. One reason is this consolidation that Vijay mentioned, which has nothing to do with AI. We had our online business payment stack in a subsidiary. Then the off-line business moved to that subsidiary end of last year. So we were running 2 different stacks, right, because there was one payment stack for online, one for offline. So beginning of this quarter -- beginning of last quarter rather, we started a project to consolidate those, which gives us like higher TPS and better efficiency. That project requires certain onetime cloud costs, right, because of the way the data queries work and so on. We -- and most of that was done last quarter. Some of this will be done this quarter. So I'm not saying that this quarter suddenly the cost will go down. So that was the major component. The second component, yes, we are spending more on AI. Gaurav, in our case, it is not, hey, you spend on some AI projects now and you'll get the benefit 6 to 12 months from now. And maybe it is the case for other companies, I cannot comment on that. I think if it is different, I would say it is probably because we already work in a very data-rich environment, right? So if AI changes the decision, for example, that put do more sales effort here, do more service effort here or visit this merchant versus another merchant, that is how we are running our business anyway. Now AI is giving better inputs, for example, right? So there's -- if AI is helping us deploy something faster, it is, hey, instead of deploying it in 3 or 4 weeks, we're deploying it in 2 or 3 days, right? And you're getting that uplift as a result. So because we are very data-rich, very data-led in our decisions, and it is end-to-end technology anyway, the impact of a lot of these AI things that we do starts to show up almost immediately. Now of course, there are greater long-term benefits because models are getting better. They are getting smarter about looking at our context and so on. So yes, but more than 1x for sure, multiple X benefit we think we get right away in that quarter. And then over time, hopefully, we get better and better.

All right. Last question, Madhur. How to look at the sustainable operating leverage in the business, like every incremental dollar that you make in CM, how much of that will flow into the EBITDA? And similarly, how much of that EBITDA will flow into the free cash flow on a sustainable basis?

So we don't try to manage it quarter-on-quarter, but at least the target that -- when I take a step back, the way I look at it is, our indirect expenses, including higher expenses in marketing and sales and service and cloud expenses, et cetera, should over a 3-, 4-quarter period, grow at less than half of revenue growth, right? Revenue growth/CM growth, assuming CM through the cycle is flattish, should grow half of CM growth, let's say CM growth to be a little bit more precise. And so that does 2 things. One is there's structural operating leverage. Second is if you can accelerate revenue growth, then you get accelerated operating leverage.

And similar thing on the free cash flow?

Free cash flow also but there is this component of CapEx, which we'll give more guidance on next quarter. But yes, broadly, free cash flow will work with -- in EBITDA terms, right. And just to be clear, we do EBITDA, we don't do adjusted EBITDA. So ESOP cost. ESOP cost already after that, yes. You want to take Rahul? I think we lost. Rahul, go ahead.

So just one last question from my side. I just wanted to understand how the maturity of some of your lending partners has evolved over a period of time because they've been now using a platform like us. And we're clearly seeing that they are scaling the capital that they are deploying -- but in terms of getting respect for the data that we are trying to provide because one thing is using it as a digital channel and one thing is respecting the data that we have. So is it being seen in terms of the approval rate of their collaboration on our data? Or is it still more restricted to their own ways of underwriting? That is what I wanted to understand.

We will -- I think scaling up is a measure, Rahul, of the fact that they have more confidence on our insights. I think there are other aspects of it, though, which is also they have more confidence over time on our collection capability because remember, we help them with distribution and collection. So collection capability. So as they scale up, they spend a lot more time with the collections team as well to understand how we monitor collection, how do we do early warning signals. So it is not just about, hey, somebody went delinquent now can you collect the loan. Actually, we have the ability to look at early warning signals and make sure the sales and service team expansion, part of it is in this area that do early intervention to make sure that the merchant actually does not drop below a certain GMV because we see daily GMV, of course, right? So they have more confidence in collection. And then they also come on board some of the journey that we would have had with other lending partners, which is to say how to think about repeat business, how to think about top-up business, how to think about newer segments, newer pin codes, et cetera. So that is a constant dialogue. It's not like -- so there is a lot of collaboration and dialogue between us and lending partners. Of course, they start with something that they understand already quite well. But over time, many of them, I wouldn't say all of them exactly at the same pace, but many of them are curious to say, hey, what are the factors for scale-up of this, right? And part of that is obviously -- sorry, I forgot to mention technology as well. So they will work with us to say, hey, what are the bottlenecks in this? Is there like a KYC failure that we need to address? Is there something else that we need to address? So they'll work with us on many of those things as well.

Sure, sure. And on the marketing business, have we mentioned that what is the driver in this particular quarter? And how soon we see a meaningful number on the AI revenue, let's say, any milestone number, INR 500 crores or 5% on our revenue, any benchmark, how soon we could expect that?

I think we probably will take at least 2, 3 quarters more, at least because it's a brewing thing. It's a small revenue per merchant and sizability required size enough. And in the large revenue buckets that are swinging the revenue, it will, for some time, show small, but then it can grow. So maybe, I mean, by same quarter next year, we should look at what it has done.

With that, we have come to an end of this call. Thank you so much for joining, everyone. Have a good day.

Happy Diwali.