Subscribe
Share
Share
Embed
Raw, unfiltered notes on pricing, monetization, and value of AI products. From pricing expert, Maciej Wilczynski, Ph.D., from Valueships. Perfect for product creators, software entrepreneurs, and everyone who needs to speed up their monetization game. Topics include: token economics, outcome-based pricing, migrations from subscription to usage-based - the mechanics nobody tells you about. Solo-engineered on my own, always keeping it below 20 minutes. New episodes every week.
Maciej (00:09.518)
You're building products with the prices of frustration. We're gonna show you how to build it right. Twenty minutes, you're ready to start. AI AI monetization. You're building products with the prices of frustration. We're gonna show you how to build it right. Twenty minutes, you're ready to start. Hello everyone, welcome to the new AI monetization. And for today, we're going to cover something that is super important. Something that is, you know, pretty much.
Like the essence of understanding where we're going with the overall AI pricing, especially in the SAS world, in the Gen AI products. And it will be a little bit of historic, but in a good sense. And it will be a little bit of where are we going? So we kind of covered two parts. One, a little bit history of the metrics, and second, a little bit of the
where this might go in the future and generally like this kind of summit that I'm calling like frog jumps of how metrics and how monetization of software models evolve. Okay, so stay with me for a moment and we're going deeper. Let's go.
So, first of all, a quick introduction to the history of software and generally the IT industry as we know it. you remember back in the days, like in the 50s-ish, there was a company called the IBM. It's they created like a very big mainframe computers, they sold them to other companies, and it was pretty much like a big calculator, right? And
what they did, they were selling it in units, in instances, right? So you pretty much pay per one computer. And what happened after is that
Maciej (02:11.104)
More or less in like 70s and 80s, where the personal computers emerged, obviously including companies like Apple, like software from Microsoft, we came up with this idea to sell software in licenses. In again, not people, not computers, I'm sorry, not computers, but actually people using the license. It literally meant
That one person equaled one unit of a product. It was burned on disks initially, later on CDs of course, but it created a whole entire ecosystem of how software is built and sold. Take any company like Adobe, like like Microsoft.
Any company creating a software gaming industry, it was always one license, one copy, you sold it, and here we go. So after a while, it's pretty much like 20, maybe 20 to 30 years of well-established business. Someone had this idea, and someone is Mark Benioff, among the others, of course, but decided: hey, maybe, maybe let's not install the
Software on the actual computers, but you know, put it on servers, and this is how cloud computing and putting things in cloud was created in a natural line, simplifying things. But this created a certain phenomenon. This created a certain increased phenomenon in first of all creating tiers of the product. So while we had
Like Office Pro and Office Office Office Home in the past. This allowed us to switch the licenses without actually going to the shop and buying new copy. So it all it took was to call or email, and we could upgrade the product or downgrade the product according to our needs. So it was a push of software companies to not only like maintain the applications in the cloud, but also it was a push to create
Maciej (04:35.734)
A new expansion revenue streams for the customers. So as you can see, we already covered like 40 to 50 years of software, but in the like early early millennium, but it literally is moving from a higher vendor control and higher value capture of the software versus the customer, right? So back in the days we had this like you install software, fine, do whatever you want with that.
And once we move towards subscription, we try to care more about the software. The reason for that is super, super simple. We want you to succeed. So this is why the customer sub or customer success units came to life. And it was in our best interest for you to extend the subscription to pay us in this kind of recurring licenses, not this, you know, lifelong perpetual perpetuals. And in other words,
Highing the customer to the subscription model made us care more about the customer itself. But you know, like markets are growing, models are expanding, people are trying to think, okay, well, how can we milk the customers more? What else we can do with them? And this is pretty much quite obvious. And you really want to have this as a, you know, all in all, you want to have this as a kind of method.
of building the of building the products and someone get this idea okay so we already had on-premise we already have t ring we have subscription hmm how about we do some more of a micro segmentation in the product so you know good better best repackage the solution add more value metrics so not charge only for license but maybe charge for some things that we or actually the product is doing so you know
Someone came up with this idea that, hey, back in the days there was something like jobs to be done. It worked pretty well. So again, let's do it, right? So it was an amazing idea, but it allowed to fragment the software, tailor it to the customer segments, and create a whole sub-niche of like software as a service application as you know it. So for instance, if I have an email in software, I will probably use a value metric that is more aligned with email sent.
Maciej (07:00.726)
If I have a CRM, it's probably tied with the sales seats. If I have a social media monitoring platform, obviously you're going to be charged for mentions, for keywords monitored, and things like this. This, however, again was quite problematic in order to measure because you didn't you didn't really want customers to be punished for usage. So you
Companies created like huge limits. Having said that, those companies who managed to well tune up and down these limits were more profitable than the others. And about, I remember seven years ago, maybe f maybe five years ago, just kind of after COVID era, the consumption-based, the usage-based pricing like evolved and emerged. I mean.
It was nothing new. We pay as you go, like on gas station, we pay as you go for pretty much most of the things we purchased. But in software, you need to understand it was a paradigm shift. It was a complete, complete, completely new, completely new frontier effectively. And here's the thing. Because of the fact that it was a
Completely you think it was not that widely adapted. Some trailblazers did, of course. Some companies managed to create great content on that. There were many pricing influencers. Some businesses were made around that. And some of these businesses were recently acquired. For instance, meter were recently acquired by the OpenII, I think, in order to fuel up the usage, usage-based pricing and usage-based billing.
kind of processes. But here's the thing. It was created but not widely adapted.
Maciej (09:10.434)
Why? Because market was not yet ready to fully abandon the previous seat licenses. Fast forward two years and we have AI, 2023, 2020, 2023, we have AI. And obviously it starts to hey, the discussion starts to be slightly more interesting. Slightly more about a this AI thing is burning tokens. You know, there are input and output tokens. Well
Probably it might not be the worst idea to you know charge for it. And again, if I'm charging for these tokens, then probably I can make a markup on that, right? So I'm buying tokens from Anthropic and I'm selling my AI wrapper on these tokens. The problem, the big problem with that is very, very simple. If you're charging for consumption,
On these tokens, you're literally making money as long as your main supplier doesn't increase prices. And guess what? Here is what is happening, right? We have table that was recently introduced and banned by the by the US government. But what we already knew is like that it would be super expensive. It means that
If you had a consumption based model, then probably it was a more risky approach. So smart people thought of okay, okay, okay, so consumption might not be enough, but let's think of something more value based, right? So outcome based and output based. The difference between the output output and outcome is like paying someone for hours done and someone created a job and someone created a decent job.
So imagine you're hiring a worker to create a pavement on your on your terrain, for instance, like here, like you want to have a nice pathway, and you pay them for doing it. They made a pretty shitty job, but you will pay them nevertheless in most circumstances. Obviously, you can argue, whatever, but you pay them. Outcome-based would be: please create me the best sidewalk ever created in the history of Bali Ubut. This is where I live. So
Maciej (11:36.488)
The best, it has to be the best. This is the outcome I want to. Obviously, the bigger risk, the bigger the reward. In most cases, the attribution the attribution is terrible and you can't really do it. But companies nevertheless try to achieve it. So, for instance, we have intercom with you know, hey, we will pay you for sucks, you will pay us for successful resolution, only for successful resolution. We have chargeful, you will pay us 25% of what we recover for you, right?
So this creates a pressure for companies to focus on the attribution, focus on not only the value, calculated value, but also on the fact of how much of this value are we actually taking all our own, taking from the customer versus whether it's created by the customer, right? So previously I owned a software as a service application, and it was kind of the success.
Despite the customer success teams and their best effort and you know tutorials on boarding and all this stuff kind of belonged to me. So I there were like you know these dead users that were not using an application, and it was pretty much understood as a market standard. I believe it's a it was a wrong market standard, but nevertheless, this happened. With a more outc outcome, output-based model, your
Renumeration is tied to the customer success. So again, you create another pressure on value creation. So the history of software, in a sense, is literally creating more value for the customer, but also capturing more of this value in time, right? On premise, okay, just buy it from the shelf, install it. I don't care whatever you do with that.
Subscription. Well, I have to care because you might churn you usage. Well, if you don't use, I don't get paid. And outcome, if you don't have success, I'm not getting paid. So this is kind of the way it's evolved in time. And in the future, we might see some models like ROA as a service. ROI as a service, which is, you know, we agree on a certain ROI, and the bigger it gets, the better. So kind of like an expansion of
Maciej (14:00.064)
outcome-based model, which is how the best consulting companies in the world do kind of success fee, fees at risk. This is how we at ValueShips also charge our customers. So again, if AI simulates the labor, it is doable. some research says that there will be like bidding model. So hey, I can do it cheaper as a model than so than than the other model. So let's bid together. Or otherwise you'll bid on outcome. So kind of like Google auctioning system. This is
This is the way where it might go, but you know, this is a distant future. This is the distant future. And I still think, I still think we might, to be completely honest, we need to focus on this important part that I want to say now, which is what is actually happening in this whole evolution. So the thing is, AI models are not like super new, they were already invented. And in software.
And there there is a kind of phenomenon of a adopting not what's the newest, but what adopting what is second newest. Let me give you an example. So note that far ago, not that long ago, we had companies moving from on-premise to cloud, like a few years ago. And it's still a thing. Most of the installations in software, most of the installations in software are
cloud are on premise based, not cloud based. Still, on top of that, on top of that, now we had the situ we have a situation where companies want to move from licenses, and they're already seeing that there is something like outcome based, but it's too new, too risky. So what is their next best alternative? Consumption based.
And this is critical insight. So the new license is consumption, usage based. It's not outcome-based. Outcome-based pricing will be done by maybe five, maybe ten, at best fifteen percent of the most innovative, bravest, and the best trailblazing companies. But these are like kind of like you know, this innovators, early, early innovators.
Maciej (16:28.342)
But the most of the market will move to the consumption, which will already be a substantial paradigm shift from the licenses. And current research proves it. Only 17% of enterprise level decision makers want outcome-based pricing, and 55% prefer consumption. So if I had to create like a Raman five minute noodle pricing, I would probably think of some subscription layer and consumption layer on top of that.
And obviously, there are like some details inside of it, but this is a fundamentally important thing. Our status quo is not licenses anymore. It's like not license-based pricing anymore. Our new status quo is consumption-based. So if you want to move from licenses to consumption, you're effectively
Taking a very old legacy model and moving it to somewhere slightly better and new, which is fine. And it allows to capture more value, and this is perfectly fine. But this is the reality. But if you really want to make a revolution, this is outcome-based pricing. I'm not necessarily saying that you should always charge for outputs or outcomes. I think that the attribution is hot. Getting to the right outcomes with the customers is super hard and mostly available in
Highly niche vertical use cases where you really know the overall process. So probably the niche, these niche targeted applications might do it. And this is the pattern that I also see. Like, look, lovable, very broad, very huge range of use cases. They're going with consumption credits, right? But a very niche application for, for instance, for C for CNC manufacturing, it can charge for every cut.
Every new unit of product created because they understand the product, they understand the process, and this is this is different. So, this is the conclusion that I wanted to get. Software, this is kind of the pattern that I observe, like analyzing the software history. Is like software pretty much always picks the second best model that is currently available. They're risking changing something, but not as much. And similarly,
Maciej (18:51.818)
When someone says you need to go full outcome-based pricing, I would say, yes, definitely. It's a good idea. But do you really think our use case supporters is that niche, it's that vertical, it's like that narrow that we can really get a proper attribution? And we're going to be discussing we will discuss the outcome-based models in like next next episodes and next chapters. But here this is the important insight. Consumption is the new black.
Remember it. Take care. See you next week on the AI monetization. Follow us on Spotify, Apple Music, Substack RSS, or whatever you're using. And there will be YouTube channels soon. So take care. Bye.