Guest
Tapish Kumar
Senior Director Pricing & Packaging at Freshworks
Episode description
In this episode of the Data for Subscriptions podcast, Tapish Kumar, Senior Director Pricing & Packaging at Freshworks shares his insights and learnings around monetizing AI services. From challenges of providing AI services to requirements on quote-to-cash with consumption models, and whether AI serviced through usage becomes an ARR killer, Tapish uncovers it all, including practical examples of what he is doing with AI at Freshworks.
Highlights
Challenges of providing AI services
Requirements on quote-to-cash when moving towards consumption models
Is usage-based AI monetization an ARR Killer?
According to Tapish, “the notion is starting to creep up on our ranks.” ARR is crucial for evaluating SaaS companies, particularly during an IPO, as it contributes significantly to the valuation of startups and the market capitalization of established firms. The impact on ARR extends to the compensation of sales teams and the incentives for partners, all of whom are traditionally aligned with ARR targets. With a consumption-based model, revenue is collected post-usage, challenging the predictability that comes from fixed ARR. To address this, purchasing programs are being considered to introduce a level of predictability for both customers, who need to forecast expenses, and vendors, who must budget for future costs and consider profit margins. The importance of accurate usage forecasting, leveraging AI to analyze customer data, is highlighted as part of the solution to improve usage analysis and supplement revenue predictability.
Transcript
Tackling AI Monetization
Behdad
Hello, and welcome to the Data for Subscriptions podcast, where we learn and explore how to run better subscription businesses. I’m your host, Behdad Banian, and today I have the pleasure of welcoming Tapish Kumar, Senior Product Pricing and Strategy Director from Freshworks to the show. Welcome, Tapish.
Tapish
Thank you so much Behdad. It’s great to be here.
Behdad
We have an exciting topic lined up. It’s AI monetization and that’s the biggest seems to be topic that’s running in the industry right now because the fact that it’s also the debate around usage based and what that means. But before we get to this really exciting topic, why don’t we start with a short introduction about yourself, Tapish. I mean, you’ve been in and around product pricing and business model evolution from the get go. Tell us what you what makes you so passionate about that space and also how it LED you to fresh works.
Tapish
Thank you so much, Vidal. I think it’s a great question. I think my passion really lies about how we have transformed business models and how companies have evolved their businesses over a period of time. I’ll start like I think I start post B school. I started my career with Adobe and I’m sure a lot of people know here, Adobe is actually a classic case study for how to transform your business. And I was part of the part of the journey, not as part of the pricing team, but essentially from the partner side. And you can think about, that’s when I started realizing that when you change any pricing, any business model, what kind of impacts that you have externally, customers, partners, and then internally, if you think about what impact does it have on how you enable sellers? How do sellers actually sell the value of the product? I mean, there are multiple things that I saw about this. And that’s how I started getting very intrigued about business model evolution. And my next job at Autodesk was all about business model evolution. And that’s how the passion for pricing started. And so as as I grew in my career, what happened as a result is now, you know, AI is very big thing. I’ve always started looking into big trends. It was at some point it was IoT and now it is AI. So I’m actually at Freshworks looking at AI monetization, you know, as any other companies investing in AI.
Behdad
And tell us a little bit more about Freshworks as well. We’ll get back to discuss more of the offerings and details later on. But what do you do at, I mean as a company at Freshworks?
Tapish
So Freshworks is actually a pretty young company. It’s one of it’s one of the first Indian startups. Now actually we have gone global. So with no more in Indian startup, we started in 2010 and with the first Indian company which actually went IPO and NASDAQ and we actually do a cloud based software SaaS company and we provide cloud based tools for cloud relationship management that is CRMITSM and then e-commerce marketing as well.
Behdad
All right, let’s jump right back into the topic then. So there’s been several announcements in the last couple of months. One of the more known ones is Salesforce that started to introduce his AI agents and shifting its business model to usage based as a starting point. Tapish, why don’t you give your view of what are we talking about when we’re speaking about AI monetization? Because it might mean something, but it might mean different things to different people. So you should level set what are we talking about here when we say AI monetization.
Tapish
So I think from AI monetization, one of the key things is that the I think based on our interaction with customers and partners and several other folks is that people are starting to look at AI as a value driver, a real value driver. I mean we talk about value based selling, value based pricing a lot, but value what is the, what value am I going to get with AI? And obviously there are a lot of hesitation and skepticism about AI as well amongst large customers, mid sized customers to around trust governance. But at the end of the day, what what value am I getting? So what is happening is that when companies started monetizing AI, they wanted to tip their toes in the water and see whether the customer is ready to evolve and buy more. So think about like Microsoft Copilot, right? Microsoft actually started doing with the seat-based model and then they started like, hey, am I seeing adoption, all customers buying, you know, AI and they are the covering for the entire user base. I mean there are a lot of questions that came along with it. So I think the seat-based model is where we started with AI monetization and now with more and more a value, which is being driven like with LLMS that are coming out and models that are coming out of open AI, I think people are starting to move towards customers are starting to ask about value. And that’s why consumption-based models are starting to pick up a little bit.
Behdad
Yeah, and there’s been many reasons for it. One of the topics that has been raised is that it’s simply not sustainable because we can take ChatGPT, which is arguably the most known example when it comes to AI applications. First it was tested for free and then came the monetization on top of it. Now they’re tracking the usage in terms of how many queries that you actually feed into the system. Similarly, other tools that we can just line up, you might have a number of credits, but again, here what you were doing is that you’re paying a fee, you get a number of credits and it starts to countdown based on how many queries that you ask it. So these are the examples and we’re still at the absolute infancy of what we’re seeing and in terms of the applications that are going to be embedded in every single situation that we are interacting with our, you know, IT tools, phones where it might be.
Tapish
If I have to add one more thing is generally what we are seeing is people thought AI is premium earlier. Now it is actually starting to become table stakes. So, and as a result, what you would see is a lot of customers, as a lot of vendors are actually thinking about, hey, should I actually include AI as part of my existing capable existing packaging pricing plans? And the reason is because they want more adoption and AI is actually essentially more augmentation of existing features as opposed to being a premium feature or a different use case altogether right now.
Behdad
That’s an excellent point. Let’s talk now about fresh works. And you’re offering because you’re basically offering and doing precisely what you’re describing for us. So walk us through again. I mean, you’ve laid the foundation of what your overall offering is, but let’s go down into this is how it works and how you’ve embedded AI into it.
Tapish
OK. Yeah. So I think with Freshworks, I think we were all, I mean, I think like any other company about AI monetization, we were nervous about whether the product will be used, how will it be used. I think we had some hypothesis about customer use cases, but I think we have obviously matured a lot in the last. So we started with obviously beta testing like any other company, like public beta, let’s have let’s have customers test it out, give us feedback about it. So private beta, public beta. And then finally we went GA last February with with at least a copilot offering. And then we just go, we’re just going beta with our AI agents capabilities too. And there’s some like very exciting AI capabilities coming or products coming very soon as well. But what has happened is obviously we, you know, when we are also tiptoeing, we also went into the agent based or I should say seat-based pricing for the product. And we also realized that a lot of the customers were like, Hey, I don’t want to, you know, to use AI for all my entire installed base. I just wanted to use it for a certain use case because or certain use case, certain number of seats because first, I don’t trust AI as much. Second, I don’t know how much value will it deliver. And the third is essentially, is there a cost aspect? So we also learned a lot about pricing as well as part of that exercise.
Behdad
Tell us a little bit more about the challenges that you’ve been facing or for that matter what you see as well other companies would be facing one, stepping into providing AI based services.
Tapish
So I think so that I would describe the challenges in few categories here. The first one is essentially at the time of selling itself value realization. So customers obviously, as I said, customers are not very customers feel I do not have the trust on what AI can do for that. And essentially there were it’s all about value realization. What use cases will I solve for you? And as we mature, as different companies mature, that was so. And 2nd is how do we enable sellers to go, you know, convince the customers that yes, this is value add. Now this is great for an SLG customer or sales like growth companies. But when you look at PLG companies, you don’t have sellers selling it. In such a case, what happens is you we need very strong demos or very strong presence out there to show that what value can be tried now? Is it people are now starting to compare what we can do versus what open AI can do for free or even for $20.00 a month, right? The for the, for their pro plan. So those, some of those challenges are coming up as well. One second is once they start, even if they purchase, it’s an add on right now. So the customers are like it’s another decision for them like or do I really need to purchase this, it’s another add to my budget, Should I do it, should I not do it? Those kind of questions. And as a result, our attach rates are low when it’s an add on. And I think add-ons is a typical low attach rates. Then once they start, the second part is when they start, once they get onboarded, they always think about the base capabilities. So let’s say if I buy an ITSMI would think about resolving my IT issues, not about using AI to resolve IT issues. So the adoption time, adoption curve is also longer in such a case. So by the time they come up for renewals, I mean they’re probably, I mean the half of the like time is already gone and they’re not started using AI. So when that is another challenge on adoption, how do we increase adoption one, how do we reduce adoption time as well?
Behdad
You mentioned a couple of times proving value to customers and just providing usage based in itself gives a lot of challenges because you need to be able to measure that usage meticulously. And we’re going to get to that in a few minutes as well. But already now I’m not only yourself, I’m hearing other people in the industry speaking about outcome based. Outcome based is arguably even harder because that’s about understanding precisely what value means for the customer. In your view, have you already started or, at Freshworks, have you already started to experiment with outcome based and try to figure out what value lies for customers?
Tapish
So in consumption based pricing there are or I should say outcome based, there is this outcome based pricing which is essentially what value that you get out of this. The second one is action based as well. So if you think you asked about Salesforce and sales force is actually, in my mind, an action based because in a conversation it could be a successful conversation or it could be not be a successful conversation. In such a case, it’s not an outcome, it is just an action that is happening as a result of it. So, so we are starting to see evolution of consumption based rising to action and outcome. Now to answer your question, are we exploring ideas about outcome? Yes, we are exploring our ideas about outcome, but then the project, there are several challenges around outcome, especially around the fact that the first big challenge with the outcome is definition. And how do you define what an outcome means for a customer? And what we have seen is when we actually start defining outcome definitions, customers do dispute it a lot because they also want transparency around every single conversation. So think about AI agents, right? AI agents, I, I do not have as a customer, I do not have control over what a agents are doing. I would not know what kind of outcome am I getting, but if I don’t as a vendor, if I give you the transparency on every single what conversation led to whether it was the right outcome or not, that’s when you can reduce the disputes and conflicts around it. So outcome based pricing is great, but it is extremely difficult at this point of time because of definition issues.
Behdad
Got it. Let’s flip a little bit to discussing more of the back-end requirements that are needed in order to kind of enable this usage based monetization that we have been discussing. So as a starting point, what requirements would it put on your or does it put on your quote to cash system?
Tapish
Great question. So I think AI and consumption is, is a great change or I should say transforming labor for quote to cash systems. If I have to say, I think there are, there are, I think there are three pieces to the quote to cash. One is I’ll explain when it comes to usage date. One is collection of usage data that is extremely important and that helps define us more strategy. You know, once we analyze it better, then we actually be able to create right strategies. That’s the whole fundamental framework around usage data. But one is usage data collection. Second is preparing and analyzing this data because that is extremely important. While you can collect all the data, but if you’re not analyzing it well, you cannot create, you know, you cannot create right strategies and you cannot to solve customer problems. And last but not the least, once you get the data is the billing, billing piece of it. Because what happens is when you look at billing, right, it’s especially with outcome-based pricing or consumption-based pricing, there is always conflict, always disputes with customer around how much did they use and how much are they supposed to pay. Because what happens is people end up going into over ages they overuse and they think there is a definition that they understood. And that is where we start seeing a lot of conflicts and that creates a lot more manual effort. We need more people to resolve this and CSAT gets a hit. I mean, there’s a lot of those things that happen too.
Behdad
Yeah. So from our own division point, most businesses, even even SaaS companies who’ve arguably been in a digitalized model from day one, when it comes to the automation maturity of the whole usage data management flow, it’s still fairly poor. So there’s a lot of improvement area. So on a more higher level, where would you say you are at freshwork in terms of having the automation level across the board in the entire process? So we can start with the collection, Where would you say you guys are?
Tapish
So I think with collection, we’re fairly automated because we’re a SAS company, right? So we do, I think we do a good job of the moment the customer starts on board is on boarded. Like how much, how many times did the customer login? How much you know, how much are they using, what features are they using, how many times are they using, who is using it? What are they using it for? And obviously we have some augmentation with our CSM, so customer success organization where we get some of this information too. But we have a very fairly good understanding by because of collection. Now I think there is a semi automated stage when it comes to preparing the data. So while we collect the data in the data link now how do you use this data is, is the next challenge and we are semi automated in that portion.
Behdad
Yeah, So we’re here, we’re speaking about the we can call it the analytics piece and you highlighted how important it is to learn off of the customers usage.
Tapish
Yeah, correct. So I think like we have certain tools, right, you know, and I think as I said, you know we’re fairly young company and we start up. So it was like, hey, we start with the very basic tools. And I think at this point of time there is a need to be more mature about some of the tools about collecting and analyzing data. So we’re not collecting mostly analyzing of data, right. I think those are things that we are working on, but we have a fair good understanding of what customers do.
Behdad
How about billing?
Tapish
Ha, so Billing is extremely complicated. We have some automation around it because if you think about again PLG versus SLG, right, PLG which we’re all self-serve models. So we generally have good billing aspect of it because customers know they’re buying by seed base and it’s pretty easy. But when it comes to some consumption based like you know there are smaller customers. So we don’t have a whole lot of lift and shift around billing or we don’t need a whole lot of manual effort. But when it comes to SLG or you know, larger customers, we have a lot of billing. I mean there’s some automation already, but the problem is arrayed again around disputes management, conflict management and those some of things are extremely challenging.
Behdad
Yeah, I can imagine because as you were elaborating in terms of way where you are with your offering around AI based services and how you’re kind of experimenting, you’re also testing your way in terms of where do we think the value lies. Also mentioned transparency, understanding. I can see that this is a foundation for a lot of debate and disputes. Now if you don’t understand exactly what you’ve been using, then you’re getting billed for a specific sum and you’re saying, but I don’t recognize why am I being billed for this sum. And then here starts the back and forth. I can imagine that the disputes will be starting to skyrocket, correct?
Tapish
And that’s what we think is right now I don’t have outcome based pricing out in the market. We have more like conversation based or we call it a session based pricing, which is more like actions as as as A and there’s less disputes there. But people when they see a huge bill at the end of the month just because they overshot their own estimates, that’s when the billing conflicts start. So I think what is very important, you know, for one of the learnings that I have is instead of, you know, how can I reduce disputes? That’s the big question that I have, right? And in my mind, I think as long as we do a good job of value selling and defining what does the customer, what does the customer, what should the customer expect? What is the definition of these pricing metrics? I think it’s extremely important and also give a sense on what does. Maybe it’s not just about just written documentation, but also in the demos as well here, what does a session or a conversation actually mean for the customer? So these are some of the things we should look at.
Behdad
So some of the best companies that we’ve had the pleasure of working with is that it’s one thing to kind of collect the data inbound for yourself, but it goes two ways. Transparency of the consumption data for the customer being AB to B or B to C is imperative. One of the huge upsides of usage base is that it is per usage. But one of the downsides is that if you don’t know where that lies and what triggers what, then you’re it creates a catch 22. So you also mentioned transparency previously and I would say that this is one of those areas where businesses need to really look out for transparency towards the actual customer here. So basically on a daily or whatever cadence that is relevant in that specific business. So you’d know where you’re at, how much have I consumed, how much is going to cost me? The more you can inform and educate the customer, of course, the happier the customer, less disputes. Then again this would expect a highly automated and sophisticated back end when it comes to usage data management, correct?
Tapish
Agreed. And transparency is actually of two, there are two levels of transparency. 1 is at the customer level. Obviously that is extremely important and I don’t in my mind when you release any particular offering or any business model, we should always have a transparent view of the usage to the customer. That is in my mind that is MLP which is minimum lovable product, right. We should not release any product or offering without that now. But the said there is a second part to it that is internal. You need transparency around the customer’s usage for customer success, account managers, sellers, that needs to be there as well. Why? Because what customer sees is what we want, they’re nurturing, nurture, nurturing team to see as well and more. There should be more detail on the internal aspect of it because they should be able to answer any questions of the customer. So I think those are some of that’s why I feel the transparency is 2 prong approach here.
Behdad
That’s a great point. That’s an absolutely great point. Can you also paint the picture for us in terms of how your system landscape looks like? Because I think in light of what we just discussed, you said you’re a fairly young company, you’ve been at a stage and now you need to mature. So walk us through what you have.
Tapish
So I think from system standpoint, it’s a very complicated and I think we as you know, if we are not part of the CIO organization, we always feel that AI this is, this is pretty easy. This is pretty, IT is pretty easy. Like one of the challenges that every pricing person has is hey, how does my billing system work? Can it support new business models or not? I want to launch something new on the consumption side or outcome-based side. Can I do it or not? And then generally the answer that I hear is no, right? We are still, we are, we are still some years before we can support these new models. And I think that is the case. I started to believe that the systems will always be behind what the new business models will look for. So, I think as a pricing person or a business model person, we, I’ve started developing a little bit of empathy for the IT organization. While we used to say, hey, change this billing system, change these back end systems, it’s fairly very difficult to do that even for a smaller company like us. I mean, my previous job where we were like what $12 billion company it was, I’m sure it was extremely hard there. It’s not as it’s not easy just because it’s a smaller company that I work for right now. So I think those are some of the things now. I mean, obviously we have third party tools that we go use. We had, I knew when we started, we started using a lot of open source, our systems as well, open source software. So we have to start getting rid of those and start getting into more enterprise centric systems and our softwares so that I can support our business model evolution in the future.
Behdad
Yeah. Can you maybe indicate approximately how many different kind of data sources we’re speaking about when we look at your system will ask in between the CRM and the ERP and the billing system and potentially the data warehouses that you have. When you when you collect all of that usage, what are we talking about in in rough numbers? Are we always speaking about the 10s and fifteens? Are we speaking about the thirty 40s here?
Tapish
I think just from if you look from billing and usage data collection, I feel I think we are at around easily around 20 plus for sure.
Behdad
Wow.
Tapish
So I think and again, as I said, 20 plus is mostly around the usage data collection, usage data simulation, then analysis. And then obviously you know, most, a lot of people end up using Excel to analyse the data. Then you have Tableau, then you have Google. I mean you just name it. It’s there is a lot of analysis. I think some of these need to be assimilated as well.
Behdad
Well, we can say that there is there’s a lot of also improvement to kind of unlock the speed and agility that you’re after because what you just described in terms of well, it’s great empathy. We all need to have empathy for all of our colleagues. But that aside, it is a huge challenge for businesses because on one side most of our discussion today topic has been about the potential of growing your business and the potential of basically doing that with AAI that then needs to be usage based. But then the reality is that you need to have a a system landscape that enables you to do that as opposed to holding you back.
Tapish
Yes, correct. Yeah. So I think we do need enablement or when I say enablement systems to be to enable our new business models and changing landscape. But I think some of this is also to the fact that a lot of systems that are around are also very nascent as well. So if you think about the legacy systems like SAP, Oracle and stuff, they’re also trying to evolve into the new age to support new age business models too. But when you look at these startups or a lot of AI tools or in fact AI, you know that that they’re embedding AI into this, to the, into the package, into their products. It’s actually very, very interesting. You know how they’re evolving as well and how they’re disrupting this old legacy business. You know, think about Zuoras and you know, the Arias, it used to be a company called, I mean, still is, there’s still 1. And I think I’m starting to see a lot of disruption by these new age startups, which is great.
Behdad
Tapish, we’re going to take two questions before we start to move towards wrapping up for today. The first question here, both are from Steve and the first question reads what challenges are being observed regarding the unpredictable cloud costs related to AI usage on for example, AWS and other platforms.
Tapish
I think the question if I read Stephen is about, you know, especially increasing AI cost, but if you end up using the public cloud and I think that is and that also depends on company to company a lot because some of the companies are like, let’s forget about AI cost usage, right? Cost right now, don’t talk about margins. What we want is AI adoption. So I’m seeing that at least at Freshworks and even in other places, adoption is key cost. Let’s keep it aside for us for a second, but as a pricing person, we do definitely look because we what we don’t want to be in a situation is we ignore cost from margin standpoint right now and then later on suffer once we have a very good. So we’re trying to keep a lot of balance around that and challenges is obviously, you know, because of consumption based even like AWS I showed the offer consumption based pricing or even if you go to open AI, the cost to increase quickly. Very quick example is if you have a query-based system or query-based copilots, customers are asking a lot of questions. So for example, if I don’t have, if I’m not a native English speaker, my English is not great, but the question, I’ll probably ping the system at least 10 times to get the answer versus more someone who’s more mature will probably do it two times. So the, the cost is increasing from that angle. And I think very important is the LLMS have to start evolving from that angle to say how can I make it more efficient for the vendors to provide services and reduce cost for them as well.
Behdad
And let’s take the second question while we’re at it. What are the most typical metrics being used when measuring usage of AI? Is that the queries or length up responses as an example?
Tapish
Yeah, I think it depends on the products or like AI agents is resolution time deflection of tickets. I think that’s another one. How many tickets were deflected or resolved by AI? That’s another one. How much time did it take to resolve? And especially when it comes to different languages, it’s also even more difficult because what happens is if translation aspect of it, right AI can AI do translation, live translation of what the customer is asking? I mean, your agent is sitting in in US, whereas your customer is asking in a different language, let’s say Japanese. How do you ensure that the AI agent is resolving, right? So the resolution time also increases too. So those those are some of the things that we are seeing from AI standpoint here.
Behdad
Great, thank you. So one of the topics we didn’t discuss so much today, I just want to quickly touch it is there when it comes to usage by usage based AI monetization, one of the biggest concerns and there’s been some writings about it that it’s going to be an ARR killer because but that necessarily has nothing to do with AI as much as it has to do with usage based. But of course, the sentiment here is that since we all expect AI usage going up and it needs to be usage based, therefore, we’re going to see an explosion of usage-based revenue and that in itself will then kill all predictability. Now I’m just going to go 1st and say maybe it’s not that bad. And for example, one of the previous episodes we had, we discussed usage forecasting. And there are ways, of course, granted you have the right system capabilities and the right type of data, you can actually start doing pretty interesting forecasting to guide yourself in terms of what you would expect in terms of usage and therefore, of course, revenues. And we’re doing some of those tests with some of our customers that are kind of leading the pack. But I wanted to also ask you, Tabish, is this something that you’re really concerned about in terms of it being an ARR killer?
Tapish
I think that the notion is starting to creep up on our ranks as well. And the reason is because you know, when you look at SAS companies, they’re primary metric that a Wall Street analyzes, especially if you’re an IPO is ARR and that helps with your evaluation for a startup or even for a market cap for our companies like us too. And what happens as a result is if ARR is, it’s an ARR killer there and you know, it also impacts that your sellers are compensated, your partners are incentivized by ARR as well. So we are starting to see that consumption base is, is, you know, is probably going to be that way because you end up collecting money after the fact. So one of the things that we are thinking is how do we supplement it by buying programs. And when you look at buying programs is obviously it brings in a little bit of predictability for the customer because when you look at this maturity of the customer, they are looking for predictability and transfers how much I am going to spend. And then from vendor perspective, it is how much it is going to cost me in the future and I should budget for the cost as well impacting my margins. So I think one thing that we, you know, when I said this usage forecasting, as I said, it’s very important that we collect the usage of the customer of the AI aspect and to Stephen’s part about how many metrics and stuff. And if I can have slice and dice and of all these AI capabilities by different metrics, I think we can do a little better job of the usage analysis. But yes, arr killer is some is a real problem that is starting to come up and we’re thinking from buying program standpoint. And I the last thing I feel that I think even the the the gap rules and all some of these accounting rules, if I have to say, will also change to account for some of these things because a lot of the OR the the way investors view some of the companies will might also end up evolving in the future.
Behdad
Got it. All right, Tapish, we’ll go for a wrap up and just a few final questions for you. But as a starting point, where do you see Freshworks when it comes to AI monetization, let’s say a year down the road and at the speed that we’re going, I think a year feels like a pretty long time still.
Tapish
Yeah. I think what we are, I think it’s all tied to what customer wants from our what kind of one, what kind of customer we want. Second is what does customer want from us, right. That is very important. And I think what will happen is I see the following changes happening moving from seat-based pricing to a different pricing model, more consumption based. Obviously ARR is 1 aspect, but let we’ll think about it later. Second is I’m looking at within consumption based, it is action based or outcome based is another one, so more value driven. And 3rd is I feel that we will actually start looking at AI as table stakes as opposed to a premium capability. So I feel that is going to happen very soon in and across the industry and we are not the loners here. And you’ll see like Microsoft just did that with their copilot too. So I think that’s what we’ll start seeing in the future from that angle. And but even if we include it as part of, you know, as part of there’ll be probably usage limits that we’ll have to apply to control our costs too. So that’s another pieces that we’ll end up doing.
Behdad
And Tapish, what is your advice for anybody’s who’s either contemplating dipping their toes in terms of usage based AI monetization or is already on the way, What should we be thinking about to make sure that we’re successful?
Tapish
Build a hypothesis, validate with customer first before launching anything. I think that is extremely important. And then obviously like we do in any startup tries to put up a product market fit, think about whatever AI is resolving for the customer. Is that really a solid solution that we are doing? Are we really resolving customer problems here? And then based upon that, have a hypothesis on the pricing metrics, what do what pricing that you want to do and think about a, a as augmentation as opposed to a previous feature.
Behdad
If I may add one more point, because I think that it was a great one you raised and it’s about speed and agility. And I think that this is a factor you could argue has always been the case. But let’s be frank, over the past 15 years, the 1st 10 to 12 years, if we look at SaaS and subscriptions, introduce the pricing point. You had the same pricing point for five years. Maybe you adjusted a little bit in the last two or three years because of we’ve seen churn, we’ve seen fatigue in terms of subscriptions. Now the speed and agility question has become much more important. Why? Because we need to be better at providing alternative packages. We need to be able to test. You said it yourself, if you want to quickly introduce new price points for the same product and see what the adoption and response is, I think you take that point. And if you look at the AI with everything that you’ve shared today, this is one point that I would really urge everybody to think long and hard about. How does the back end look like? How do you manage that usage data so you avoid finding yourself in just a conundrum of manual steps and mistakes?
Tapish
Yes, agreed. And to to your point without one, when you’re looking at back-office systems as well, when you’re devising your new business models, you should include your IT team as well. Get their feedback from that angle on how much time would it take with if it’s a company changing or business evolving model that you’re going to launch. So include them early. Do not wait until you define and then say this is what I want.
Behdad
Tapish, thank you very much for the conversation today. It was super interesting. And also thank you to everybody tuning in for anybody who wants to learn more about this drop case study in the comments field. And we’ll make sure to follow up with you. We have more to share with you and more learnings and terms of customers that are doing this and we’ll be happy to follow up. Again, thank you, everybody.
Tapish
Thank you so much everyone for hearing us.