Name: The data-driven path to consumption-based pricing
Uploaded: 2026-06-24T16:37:52.057Z
Duration: 46 min 56 s
Description: The data-driven path to consumption-based pricing

Transcript for "The data-driven path to consumption-based pricing": Hello, and welcome to the data driven path to consumption based pricing. My name is Chris Cahill, and today I'm joined by senior product manager Ravi Trivedi and senior solutions engineer at Urge Coates. We're all part of the Rebonero team, and Rebonero specializes in monetization technology, and consumption based pricing is a very hot topic. Many people are calling it the year of consumption. We've seen major brands such as SAP move to usage based pricing for their AI offerings. And in the bottom right there, you can see the Google Trends data. These are searches related to software consumption, And you can see that towards the end of twenty twenty five, they really started shooting up, and it's just continued to go that way this year. Every year at Rebonera, we produce our monetization monitor report. So this is a survey of 500 product leaders from all around the world. And the results here really talk to this rise of consumption. So 70% of, AI producers say that their delivery costs are eroding their margins. Okay. This is prompting, 52% to move to, usage based pricing to offset those rising cloud costs. And 56% of those companies expect usage based revenue to grow by 2027. Also, in addition to to that, we have, consumption becoming more mainstream across the board for different deployment models. So here we have, SaaS products, whether public cloud or private cloud on prem embedded. A year ago, the the most popular monetization models were a mix of subscription and and perpetual, a little bit of consumption there. But what's really surprising is that today, the most popular monetization model for each of these deployment models has moved to usage based, pricing. So that that shifts towards usage based or consumption based monetization. It's really, really skyrocketed over the past year. And that's what we're here to talk about because it's causing lots of lots of, you know, questions in the marketplace, lots of companies debating, like, the the fastpocalypse or or clawed shock, the the notion that AI agents are coming in and disrupting traditional seat based pricing models. We have seen this with, AI pricing models. So, you know, there's still a lot of experimentation in the market when it comes to AI. Going back, a year ago, the the most common, monetization model for AI was to to bundle AI with your, subscriptions. However, we are seeing that that subscriptions are projected to fall, by 2027, and the hybrid subscription plus usage based approaches are are really, on the rise. So Revenera's customers have have been, you know, talking about this issue a lot. They're they're coming to Ravi and Urge asking these questions, like, why is this happening? And more importantly, what can they do about it? How can they get on board, with the the market trend moving towards consumption? So, Ravi, why is this happening? Yeah. Interesting. The short answer is consumption pricing has moved from being a niche experiment to becoming a mainstream strategic conversation. Everyone's talking about it. More than half of our customers are already implementing it. And the broader market has been paying attention to this high profile AI pricing announcements that continue to reshape the expectations from our, customers. Right? But the shift is not just about the market buzz. It's also being driven by real structural change in software economics. The costs are becoming more variable, especially in the AI case, as the agent pick workflows expand. And at the same time, many vendors are creating significantly more customer value without necessarily capturing that value in form of revenue. So what we are really seeing is a dual pressure of rising costs to serve on one side and growing urgency to capture value, being, delivered to the customers on the other side. Right? So let's break this down into three three or four forces, that's causing this. Right? The first force is, the customers, the buyers are extracting exponentially more value out of the same software and the same subscription that you've been giving to them. The buyers are deploying AI and agentic workflows, and exchanging how they interact with your products. The value your buyers receive grows much faster than it used to as shown in this graph. It's no longer a simple linear relationship where one more user means one more unit of value. Instead, the product mixed with AI are now driving a broader workflow, more automation, and more outcomes. And that matters because if the customer value is accelerating at this pace, pricing models can't keep up. These models were designed for simpler era, and they're starting to break down. The value curve is moving up sharply, and the monetization models need to keep up with that reality. Okay? So that's that's force one. The second force is on the supplier side, the vendor like you. Right? The cost to serve your customers is becoming far more variable and more expensive because you're rolling out AI capabilities. Right? If you have low AI footprint, you're not going to see this much. But as you, provide more AI capabilities, your cost to sell, the compute that you incur, the high performance computing cost that you incur to provide this to your customers is, again, growing up. In traditional models, the delivery costs were often stable and consistent and predictable. But in AI enabled data driven offering, that's no longer true. The cost now increase with the compute or the data processing. And as the usage of your AI capability grows, those costs just don't sit in the background. They become, the topic of conversation within your c suite, executives discussion, and you're often being questioned on the business model that you have in place. And this is why consumption pricing is getting so much attention. If your delivery economics are variable, but your pricing model is still fixed, your margin profiles can become increasingly fragile over the time. K? And the the I mean, this is the real tension. Okay? Look at the revenue captured by you as a seller. Even while customers is extracting the far more value of your product by deploying agents, and the cost to serve them is rising up like we saw, the revenue captured often is not keeping pace. Why? Because many existing pricing models that you have in place are no longer aligned with how the value is actually being consumed. Right? You might have heard issues like, c clause, clot shock, or SaaS apocalypse. Right? And you are already seeing discussions about the your subscription being compressed. There's a poor alignment between the pricing metrics that you have and the value that your customers are deriving out of, your product. Right? I mean, you you might have also heard the newspaper headlines where the company valuations are being sent south because they're experiencing a c plus, phenomenon, in the industry right now, mainly because of the way they have, modeled their subscriptions around the users or concurrent users. Right? So when we put all these three forces together, we get this big picture, the the value capture gap as we call it at Ravenara. The vendors are delivering far more value than ever, but capturing less of what they should. Right? The green curve, here is rising sharply. The the the value that your customers extract, the cost to serve is also rising, but the revenue captured lags behind, creating what is called a value leakage gap. The value that is delivered but not monetized by you. And that gap is exactly why companies are moving beyond the legacy monetization models like perpetual subs and subscription. Right? They were built for a different environment. As the software becomes more intelligent, connected, and usage driven, consumption and a mix of subscription and consumption become much more relevant because they are often, with them, there is a better chance to align the monetization to actual value that, you're creating as a seller. Right? So so then the next question is, I mean, what what are we seeing in the market? Right? So, what what I mean, what are the sellers doing in response to this phenomenon. Right? We see three main monetization models emerging if you have this problem. Right? The first one is the tiered sub subscription. This gives your customers a predictable base with the usage guardrails that allow them to move up the, tiers as the consumption grows. Right? The the second model, is a subscription, plus usage. Right? What we refer to as a hybrid. This combines the base subscription with a metered add on so your pricing can better align to incremental value and usage that, your customers are deriving out of your offerings. Right? And the third is a pure credit or token based model. Right? And some producers in certain market segments are liking this because it creates a flexible currency for consumption and your buyers can kind of enjoy the benefit of on demand access, a, a a lower entry barrier or experience across product usage as well, as the newer AI driven workflows, are being put on your offerings and customers interacting with your software using AI. And some we know are actually doing, mix of one or more of these three models that, are described here. Right? So these are the three monetization models that are emerging in the market. Okay? Now can you simply jump on to a, consumption based pricing? Right? The answer is no. The biggest mistakes that companies are making is really starting by finalizing their consumption rate card. Right? Or putting a price on each meter that they want to price I mean, charge their customers on. Our advice is don't jump straight into the token tables or price points before you've done the foundational work. And the foundational work starts with understanding what value the customer derives out of each capability that they are using in your product. Right? And then validate your packaging, fairness, and willingness of your customers to pay, and then think about the migration strategy with a pricing experts in the field like Simon Kusher or Deloitte, who whoever those pricing experts are. And we want you to collect the usage evidence that you plan to monetize. Right? And that's where Ravanara plays a key role in collecting a product and, license usage telemetry. And when you have that, run a pilot with your customers to test your assumptions against the real buyer behavior. Right? And only then should you set token rates or detailed usage pricing as, described here. Right? The message is simple. Assess the value before you set your rates. If you skip that sequence, you risk building a model that looks logically, internally correct, but does not work in the market when people buy them from you. Right? Now we I mean, I I said that collecting usage, assessing a value, is the first step. But is it easy? The answer is no. When you partnered with Ravenara, you can make this work. But, according to a recent survey that we did with our customers, right, the biggest blockers for aligning price to value, is is like 55 percentage of people actually told us, they they're not sure, if they understand the customer value drivers. This is the biggest blocker from them. And without having usage, you can't make a reasonable guess of this value delivered. Data availability and insights is the next biggest hurdle right in, in aligning price to value. 36% flagged up uncertainty around the price that they fixed on AI features, and another 36% cited quantifying and communicating the value to their customers is a biggest, blocker, for them. Right? And 35% of them said the current monetization models don't align with how customers want to buy their offering. Right? So, I mean, what this tells us is that the challenge is not just technical. It is strategic, analytical, and collective organizational wide, thing. Right? The companies are struggling to define value clearly and, are not able to measure it consistently and translate it into a model that customers will accept. And that's exactly why a data driven approach matters so much. A a a before you launch your consumption based, pricing, having this plan of data driven path is very critical to your success. Right? So the the big question is, I mean, how do you launch a consumption based pricing? And, we're joined with Urge, who I work with closely, and he's going to take us through how do you implement and launch such a consumption based pricing. Thank you, Ravi. So in my experience with working with ISVs from small to extremely large enterprises, I'm I'm here seeing a consistent pattern on product management and go to market teams approach into dipping their toe into the world of AI monetization. And what I've done is kind of break it down to a 10 step framework that I, see. And, of course, every organization is different and you might approach these, segments of these, of this framework at different times. But at least this gets you thinking about, practically, how can I introduce some kind of AI feature or product market offering which has the best chance of being sustainable in terms of revenue capture for my business? And I think the theme of today, as you've been hearing, is predictability is key for when you first start off this journey. And this is important because your CFO is going to demand it because he needs to know how much money he's going to be making in one year's time, two year's time, three year's time, which is kind of counteractive to how consumption based model work. You've got your CRO, demanding this predictability because he needs to know what money he is going to forecast to be able to make, going forward, you know, depending on how much, you know, resource, is delivered to pipeline generation, etcetera. But also your customer demand and predictability cost predictability of your solution. So it's reason why Chris at the beginning of this, presentation and Ravi, in the middle segment said that most of our customers and most of the software vendors that we have spoke with has started this AI monetization journey with a strap on of these capabilities onto an existing subscription offering, if you like. So it allows them to make a beachhead in these accounts, get the some market share or and show real, you know, return on investment, possibility of these features, while doing it in a way that their customers are used to and predictably, but in buying your solution. So you the foundation there is also, is clear. Make sure your subscriptions, of your core functionality and whatever additional offerings you're going to offer are solid because that is your safety net when you're experimenting with consumption based, pricing, which brings with it flexibility but volatility. Once you've got there, that's when the real work, starts. And as Ravi said, the foundational thing to do here is to study study and understand how value is derived by your AI feature or your software in general, which best correlates with, like, finding a metric which best correlates with that value creation, in Ravi's, graph earlier. Now this is not easy, and it's unique to every software vendor. Every software vendor in the world is really targeting their solution at a different business problem. Right? So everyone's metric associated to alleviating that business pain is going to be different. And, unfortunately, if it was easy and I could just tell you exactly what to measure and how to pin this, then I think, it would be a lot easier. However, these are some common things that I see customers, that are, you know, partnered with us really look into. And it's analyzing end user behavior of their software, but more importantly, using their existing monetization metrics or their licensing metrics. If there's user based proceeds, the, you know, measuring how many users at peak load, are actually using the software, the average time they're using it, and what actual, set of features or capabilities in their software are they using it are. You know, 90% of the is the 90 percentile only using 10% of features. That is incredibly important information if you ever want to attempt to monetize those 10 of features, in a different way, namely more consumptive way. Next up, once you've kind of analyzed it and you've got a guess guesstimate of the correct metrics to go after, it's important that you layer in this consumption approach, not necessarily directly by monetizing it in a consumptive manner, but more of a practical approach of using it to perhaps put guardrails around, different monetization packages that you're offering. So an example here is you bundled your AI capabilities as part of a subscription. Maybe that subscription is a kid subscription because you've got a, bronze, silver, and gold, and only the gold customers, are you allowing to use these AI features that you're, that you're, creating. Thinking back to Ravi's graph where that margin was eroding, you need to protect your business, against that total, profit margin erosion. And sometimes, we didn't, highlight this earlier in the graph, but that profit margin cost of operationalizing that feature could reach the subscription you're actually charging for your customers. So in order to really run a sustainable, kind of go to market offering here, you need to add some kind of control layer of a fair use policy. That's what I really mean by introduce credits as a control layer. You don't have to start saying to the customers to the customer, each credit costs 10¢ and every time you use it, it's gonna be kinda pay me 10¢, or I'm gonna take 10¢ off your balance that I'm holding for you. It's more like buy the gold subscription, and I'm giving you 500 tokens per month for each user, and that protects my business in terms of a fair usage policy and protects you by giving you the predictability that you you love so much in a subscription by us. Next up, we have the fact that most people, when they think about consumption based pricing or paper use pricing, that's pay per use is a dangerous term in my head. Right? Pay per use, when you say someone, when you when you say that to someone, I always think, right, there's something I am using, and every time I use that specific thing, I'm going to pay. This the whole notion of credits, the whole to notion of tokens should be beyond just the use of a specific thing, a specific technology, a specific capability. It should be portfolio wide access to anything I want to just consume my tokens on. So, yes, most people start their journey with a specific things, thing they, introduce as an AI feature, and those tokens, should be used on, like, an AI Copilot style chatbot. But, eventually, there should be a strategy where those tokens that could be used on a chatbot can also get you additional seats when you need it. Or temporary access to an additional module for for instance. You know, this this virtual currency shouldn't be gate kept into specific features. You should try and drive the adoption of your technology via the vehicle of tokens as far as possible. As far as that what makes sense to your business. Now this one, I I often hear overlooked, and I bring this up because, often in my, customer conversations, customers complain, especially in certain industries such as engineering software. I hear it quite a lot where there's a lot of trial abuse. And what I mean by trial abuse is at the moment, let's say with subscription trials, you enable a trial for your customer for thirty days where everything is allowed to be used at an all you can meet buffet rate because you're trying to entice the customer in to use those features and love your software. But what if they needed to do only a specific thing? What if they only needed to do one thing that they've collected all of the things they need to do and smash it out in the thirty days and no longer need to buy your product because this is a real, situation that, ISB is fine. And not only have they given you an all you could eat buffet to, use the software and how you need, and you've done you've gorged on it and finished your task, the ISB doesn't actually have any telemetry on what you have done in that software. So you don't actually know what you've used it, what you've used the trial on. Well, layering on tokens really combats this issue well because instead of an all you can eat buffet for your trials, what you're offering is a set amount of tokens to your, potential prospect. But what it's doing is, a, limiting the amount they can gorge at that bound buffet, but more importantly, giving you telemetry on what they are deciding to eat so that your salespeople can be armed with that information, and have better value driven conversations around what they found interesting in your software. Plus, this is releasing a monetization model where there isn't actually risk to revenue. Right? So when you're kind of pushing this out into the world, you're at least trying it with something that isn't, you know, directly tied to revenue at this period of time. I know it's indirectly linked to revenue, but at least you're releasing this model of how to purchase your software and getting your organization used to dealing with the operational aspects of, of a token based monetization model, but without the risk to your c suite and your CFO, saying, you know, this may, impact revenues in x y zed banner. Once you've kind of played about with, everything we've spoken about so far, this is where the real crunch time happens in terms of adopting a consumption based monetization model. And, of course, as you've heard, predictability is so important, in this first part of your journey that we we at Ravenara see the most common approach being a hybrid monetization model, AKA two scenarios. Either you'll have a subscription for your core, capabilities of your software and then additional advanced capabilities such as AI capabilities or, you know, Copilot chatbot, functionalities is modified separately from the base subscription in some token, manner, or there's a situation where your customers are reaching a ceiling of productivity based on their subscription capacity, and your customer base is reaching that capacity seasonally. So only two months of the year, three months of the year are they slamming into their, you know, 10 user, limit for their, subscriptions, but they're they're being prevented from being as productive as possible because of that limit, and they temporarily want to raise those limits. And, typically, they don't want to go through the whole procurement process of trying to buy extra seats for a limited amount of time. Plus, your organization hates trying to sell seats less than one year, let's say. Your three year deals, but, you know, your sales team is trying to force a sale to one year, but your customer is saying, I'm only gonna need it for one month, and I want that flexibility. So scenario b is all about allowing that flexibility for seasonal use of your software, giving flexibility to your customers for short periods of time, by introducing a way of accessing that productivity via tokens. Once you've kind of gone ahead, experimented with some beta customers that are interested in these models, this is where the snowball effect starts to take place. As you gather data from these customers, from how they're using it to approaching, you know, seeing how their usage is, moving throughout, time where they're perhaps reaching the peak, where they are approaching their limits so that your sales team is armed with this information to go and say, look, three months you've hit the limit for the last three months. We think by buying more base subscription, you're gonna get more productive, and you can actually buy these subscriptions in a cheaper manner than using those, peak capacity loads via tokens, which are more expensive way to access the technology. You know, you're gonna drive upsells in that, manner, and your tokens are basically a marketing pipeline lead generator at that point for your, for your, sales teams. Conversely, you're going to see situations where customers are just sitting on their licenses and not, actually using, the software as they originally predicted. This gives your sales team, your customer success team, your, consulting teams chance to identify those customers early, see where why certain users are using certain parts of the functionality and the other parts of the organization that are just neglecting that. Understand usage patterns of that power user and divide the plan of emulating that power user's, behavior into other better, stakeholders into the account and therefore reducing that risk around, you know, renewal churn, and, you know, protecting that bottom line. Another one that I see overlooked, when I speak to customers is the fact that tokens give you yet another monetization lever here. Right? I mean, all too often do, do you hear your sales team, having to discount because of perceived value or during renewal time, having to provide, you know, discounts because they haven't seen the expected value that, they they first assume they would get during the first engagement with your with your organization and you having to discount to keep the customer happy. Well, tokens give you another lever that's separate from that discount lever. So instead of giving that 20% discount that their procurement team is demanding, you give 10% more tokens. Therefore, keeping them happy that they're getting more value out of their investment while defending price. Right? So, it really does add that extra, you know, bit of ammunition to your sales or renewal teams around, you know, keeping a customer happy, but keeping that revenue that you're having to discount at least in your organization. If you give a if you give a general discount, that cash can be used somewhere else. If you give an upsell of token capacity, yes, it's that perceived value discount for procurement and their metrics, but at least it's keeping the spend within your platform and not someone else's. And then this is where a lot of people dramatically overlook with these consumption based, token based monetization models is the endless demands that end users will put on you. Because, you know, usage based licensing, is very high in risk compared to subscription, which they've, you know, been used to managing to up to this point. And, you know, as we keep, using the term Claude shop, it's quite apparent in new news media at the moment. Right? People want control over the spend of these consumption based models, and they are going to demand it. And so your technology layer of supporting all of this not only has to operationally fulfill those tokens and count them, if that's one thing, but also give the power to your end users to manage how the the rate of consumption of those tokens. So we typically hear end user license administrators, demand controls like a burn velocity. How much could each user burn through, per day, per week, per month to prevent that hungry power user from consuming all of the tokens that you have budgeted for your entire organization for a year? And we've all heard we've said it before, Claude Shoppe, there's many organizations actually facing that today with the general purpose, LLM providers, out there. So giving customers control and and having a technology stack that allows that flexibility either for yourself to mandate or handing that self-service power over to your customers is super important. There's an endless string of requests when it comes to customer controls. And then finally, this is an iterative process. And I promise you, when you first do this, you're not gonna get it right, nor are you gonna get it perfect the second time or the third time because it's large organizations like Salesforce and GitHub and, even Claude and OpenAI, how often are they having to change these, pricing models? They're making it up as they go along, which means the market currently is making it up as they go along. So which means you also have to have this period of experimentation, and you need a technology stack which allows you to do that experimentation without relying on heavy engineering of development processes to make those agile changes and make those changes in a small enough scope where you can experiment with one customer without affecting another customer as well. That's that is super important. So where are we at? I think, you've heard some key takeaways here. You know, we will start back up to the talk about predictive, revenue and making sure your subscription base is solid before you go ahead and introduce volatility to your revenue streams. You need to measure every metric you can think of that is attributable to the value that you're creating for your customers. Because if you have a hope of the monetization metric, you need to have a baseline. Right? You need to have the metrics already recorded so that you can model and forecast over them. If I make this metric a consumption metric, what does that do to my revenue? What about this one? What about this one? And that's that's where your pricing teams really come into their own, but they're only as good as the data that you can, provide to them. You introduce consumption in a controlled way. So not necessarily, releasing it as a true go to market offering where each bit of use is, 10¢, 20¢, but you bundle it into some packages, but with controlled, fair use policy guardrails around. So, there is predictability given to your customers. But then once you have real world data, your snowball effect starts to take shape, and then you go for a hybrid monetization model, a subscription plus either overages, sold by tokens or, a subscription based, like, yeah, subscription based with advanced capabilities and monetize buying tokens. And then you can see how this is gonna be iterative. Let that data come through. Let me tweak and tailor your rates, your metrics, your packaging of your tokens, so that, you know, you're influencing those upsells, renewals, and making sure you're gathering that data to make proactive decisions around, you know, making more money for customers, but also just making sure customers do not churn and then keep going. Because this is probably an effort which is not going to take you months. It will probably take you a year or so to actually be in a stable position to call it a true success. Great. Yeah. I mean, this this is good. I mean, being a product manager myself, I I I just love the way you put these 10 steps, to launch, a consumption based pricing. Right? A data driven approach. And and I think I mean, one thing I've learned over years is that, a lot of vendors, have put their license checks, right, at the key key monetization events. They they do want to make sure that the the license is checked when when somebody is using a a a part of the application that means a lot to their customers. And and bringing all that, monetization, data or the license check data or or, I mean, or or even product telemetry in addition to monetization telemetry, is is a great start, right, and and necessary step, to assess a value. So yeah. I mean, it it it was interesting how you, broke this down into these steps. Yeah. I I found it interesting myself. Yeah. Awesome. Yes. Yeah. Key nature is key here, and we can't, reiterate it, enough. And where do you get that data from? You need to get that data from product usage, telemetry tool tools, so understanding the user behavior of your software in general. So this might not be just, you know, licensing data per se at this stage, but, you know, user behavior within your software, I always say, you know, what 90% of your end user pays, what do they do when they first open the application? Right? 90% of them put file. Of that 90%, 60%, the next thing they do is click open, and the most popular file type type there to open is PDF. That kind of telemetry is the macro, and that's where you should start probably because you need to understand how people interact with your software. Because if you start changing things like, consumptivizing the, the fact that you open PDFs, that's going to change that behavior. So you need the benchmark before you start implementing consumer behavior to see if it's a positive impact or a negative impact. Then you do need to start experimenting with metrics that you've identified could be attributed to, that, ratio of value creation to, you know, revenue capture by instrumenting licensing technologies, which allow you to see that, you know, monetization, licensable, factor coming into play. How long was the user using that particular feature? You know, how long did that workflow take? How long did that simulation take? If I put a a limit on this, on this simulation run or report generation. What would be average amount of reports generated for each of my different tiered accounts? So if I put a 500, report limit here, am I really cutting the cap, too low for certain segment of customers? That that's where all of your licensing telemetry really needs to be coming. You need a good licensing technology to be able to, feed you that information. That's when the fun starts in terms of modeling and forecasting. I love it. Thank you so much, Ravi. So it's great to hear your insights on these topics because I know you're both talking to product leaders and business executives all day, every day. So, everything you've discussed right here is based on real world conversations that are happening right now and evolving as we speak. So, yeah, I hope everybody in the audience found this a a valuable session. If you would like more advice on on how to move to consumption based monetization, please get in touch with us at revonero dot com, or reach out to Erjun and Ravi on LinkedIn. We also have an ebook, AI monetization done right, a practical guide for product and pricing leaders, which we'll put a link in the in the chat you can download. So, yeah, that just leaves me to say thank you very much to you both, and, we'll see you again soon. Thank you.