What Is AI Actually Worth?

Pricing Is Not Just Math

Speaker 1 0:00

I used to think the pricing was a math problem. There's often those disagreements in the senior executive room that I hinted at earlier of it's not that nobody has any idea what they're doing, but everyone has an idea of what they should do and none of them overlap. It's not the solution to a math problem that you need to solve in that situation. I don't want to know like the architecture of your back end. And so we want to be very cautious about putting that into our commercial approach, as you already stated. You're putting sand in the gas tank of your go-to-market because now instead of your salespeople talking about the problems you solve and the value you create, all of a sudden you're talking about the innards of your implementation and cost model. And that's not your customer's problem. I think about the individual customer conversation, helping them understand that pricing metric, in that case, the hour, is not tied to the value. In fact, they would very much appreciate if you could give them the answer or the marketing campaign in five minutes while you're on the discovery call with them, instead of, well, it's going to take us two months, three months to build all the material. But if you give them five minutes, I'm sure they'd pay you a ton more.

Speaker 1:15

We have a special episode today on a topic that may sound mundane, but actually is really exciting, and that's pricing. Pricing for software, especially in this world of AI, is absolutely critical because you're trying to understand why people want to use what you have to offer and what they're willing to pay for it. And so what Dan breaks down is that whole notion of behavior of the ones that are buying from you, how they see value, how the previous notions of seat pricing or usage pricing is completely changing. And he gets into the topic of credits and tokens, and which is something that's an easy way of doing something, but actually doesn't connect to anything that truly adds value. He says that it's like showing your underpants. And so that's what I think is really useful to break this down to first principles and how it drives commitment from people to buy, to use, to grow with. How do you take something that is intellectual property and get someone on the other side to pay for that property and see value in that property and see value into what in what they're offering? And he really breaks it down to his own life experience where it's about people's underpinnings, their beliefs, their structures, the way they communicate in society. So it goes way beyond that. So please listen in.

Meet Dan Balcauski

Speaker 2:54

Welcome to the Spark of Ages podcast. We're joined by Dan Balcauski, the founder and chief pricing officer at Product Tranquility. He's also a program leader in product strategy at Northwestern's Kellogg School of Management. Throughout his 20-year software career, he has focused on bridging the gap between technical value creation and commercial value capture, helping B2B SaaS leaders dispel the pricing and packaging illusions that hold their products back. An Austin-based pricing strategist and former software engineer, Dan's prior leadership spans scaling product and go-to-market strategies as head of product at Lawn Starter and product strategy principal at SolarWinds, as well as hosting the SAS Scaling Secrets podcast. Dan holds an MBA from Northwestern's Kellogg School of Management and earned his BS in computer engineering from Iowa State. Some of the key takeaways you can expect from this episode the seat-based trap, pricing outcomes overhead count. The hybrid dilemma, pricing software seat speed with agency quality. And finally, beyond objective buying measures, how psychology impacts what buyers value. Dan, welcome to the Spark of Ages.

Seats Versus Tokens In AI

Speaker 4:18

And even in the B2C world, how do you price your products and services given what's happened with AI? So it's a great subject for us to dig into. So, Dan, you frequently advocate for value and outcome-based pricing over traditional seat-based models. And you warn CEOs against token or credit-based pricing by calling it pouring sand in the gas tank of your go-to-market engine and showing the customer your underpants. That's a great quote. But if AI agents reduce the need for human seats and buyers hate token math, what's another pricing path that B2B SaaS companies aren't seeing yet?

Speaker 1 4:56

You have hit on, I think, one of the core challenges that companies are dealing with. And not only companies, but their investors as they look at what AI is doing to white-collar work. And so I am not a person who says with absolute ban that seats are a bad thing. They often, not all the time, but often work quite well. For example, in a CRM space, your number of salespeople is usually a really good proxy for the amount of revenue a company generates because each of those sales people is a quota-carrying rep and it works really well if you're saying, hey, our CRM is going to help you make more money, that works. I think the opposite end of that view is if AI agents replace all of our white-collar workers, who are we going to sell seats to? I think that is one extreme and then two not helpful, in that if that happens and all white-collar employees are out of jobs, we have worse problems than meeting your quarterly revenue targets because you didn't sell enough seats. We'll probably have pitchforks and torches because we're going to have 30% unemployment in the world and in the US. That's going to be a bigger problem than seat-based pricing. Something else that you also mentioned is this thing I've talked about elsewhere, which is getting away from this idea of tokens. Tokens make sense if you're anthropic or open AI, because you're at that level selling infrastructure instead of AWS selling compute, storage, and networking capacity. That is the bare metric by which input-output is judged. But again, I'll just use the CRM example. If I'm talking to a VP of sales or CRO, trying to engage them in what their problems are and their challenges and how they want their sales teams to be more efficient and performant and drive more pipeline, close that pipeline. Tokens has nothing to do with their world. It is the farthest thing from their world. And the conversation about seats, how many salespeople do you have? Rajiv, Mr. CRO, you can answer that in a second. If I then turn to how many tokens are each of those salespeople going to use every month, you're going to be like, first of all, what's a token? And two, I have no idea. So what do we want to think about instead? Which is what is the problem that we're actually solving for customers? And the example I think everyone in my space has used for a while has been what Intercom did with FinAI. They were so successful with it, they rebranded the entire company and then quickly got bought by Salesforce. So congrats to them. But when they came out with FinAI, which was their AI agent for helping to resolve customer support tickets, they didn't try to make a justification behind it's this many credits to solve a complex support case and this many credits to solve a password reset simple case. They just said, hey, we've done our research. We have from all of our conversations with VPs of customer support or chief customer officers, we know that generally you guys track cost per ticket resolution very closely. Overall, in the market, that's around $20 per ticket. We're going to charge you 99 cents per resolve ticket. And so that allowed them to tell a very simple and compelling story. Now, even with that, I've read their knowledge bases, which go on for five, six pages of even in that situation, what qualifies as a resolve ticket. So even in that case, the execution is not as simple as the marketing pitch, but it's still much more relatable to that end customer than, hey, it's this many credits you're going to pay us, and then we're going to charge you based upon the complexity and how many you use. No, 99 cents per resolution. We're going to have a definition built into the contract of what that resolution is so we can both agree on it, and then stamp that as this is what you're going to be billed against.

Speaker 9:18

So I think what you're saying is it's easy to say token math if your input for a firm is tokens, right? That that's the cost for that software, right? For the use of that infrastructure. But that's not the value someone's receiving, right? So you still have to find a way to bridge to the value that they're receiving.

Speaker 1 9:38

Yeah, it would be like in B2B software pre-AI, if you were making your buyer decide, well, do you want an Oracle database to back up your data, or do you want us to use a MySQL database? And do we need to have disaster recover and failover across regions or single region? You're like, you don't make your customers make those decisions. Customers wants to know if my data's secure, right? And you guys have uptime SLAs. I don't want to know like the architecture of your back end. That's not my problem. And so we want to be very cautious about putting that into our commercial approach, as you already stated. You're putting sand in the gas tank of your go-to-market because now instead of your salespeople talking about the problems you solve and the value you create, all of a sudden you're talking about the innards of your implementation and cost model, and that's not your customer's problem.

Speaker 10:33

Oh, it makes a lot of sense.

Board Signals Beyond Seat Counts

Speaker 10:35

Dan, you've said that zero price complaints or zero loss deals is actually a warning sign that you are underpricing your product and that pricing decisions expose weak positioning faster than anything else. So when a B2B company transitions from legacy per seat pricing to an AI-native go-to-market engine, traditional board metrics like seat utilization break down. So what are two or three pricing adjacent strategy signals a CEO should present to their board to prove that their pricing model is accurately capturing the value created by AI?

Speaker 1 11:11

Wow, that's an intense question. So I think there's a few things that are pricing signals that we should keep in mind, whatever are either enabling technology like AI is or our overall pricing model. Number one would be things like overall net revenue retention. That was probably the number one signal as a business of whether your pricing model is delivering value to the business. Very talented salespeople out there, they could sell ketchup popsicles to men and women wearing white gloves. The question is: do those folks renew and then do they expand with you over time? And that exists in an AI world or not. I think other things I would suggest folks continue to keep an eye on, and they probably are already, would be things like overall sales cycles. It's very bad when you have to have, I've run into this before, companies have five to ten additional conversations at the end of a sales process, going round and round trying to explain how their pricing works and whether they feel ticky-tacked on the bill of materials, and you have a conversation, you think you're gonna get bill for one thing, that the invoice, and all of a sudden it's got 30 things on it, each with a line item, and now you're having to explain it. Again, these are things that hopefully, if we design and put the work in up front, we can mitigate. Obviously, there are different markets and approaches, right? Enterprise, for example, is notorious for just having more complex pricing and packaging because every enterprise needs something different, and so that's just how those products grow. Going to be very different from a PLG or volume and velocity type go-to-market model. Net revenue retention and your sales cycle length.

Speaker 13:03

NR, GRR, those sorts of metrics, churn metrics are really important to learn. But those are trailing indicators, right? So is there a lot of things? Is the way to get a sense of what it looks like? I'm sure folks, when they create that pricing model or implement it, they're trying to see what's the uptake, is it working, or is it causing lots of issues?

Speaker 1 13:22

Yeah, any transactional metric that you'd be able to divine from either your general ledger or your CRM system is going to be a trailing metric, right? And so there's a revenue or a PL in general is always going to be like driving your business looking in the rearview mirror, like it tells you what happened. What I've generally found is that those trailing metrics don't really tell you what to do next, they just tell you what happened. So what we do, or the I guess the mindset that I would flip people towards is we could do a lot of research up front. There's many research methods that have been devised over time. One of the fights that I'm constantly trying to shake my fist against is that somebody says, we can't get any good data because people will lie to you, they won't reveal their willingness to pay. I'll first point out there's a big gap between we can't get perfect data and we don't have any data, so we're gonna go with my opinion. That's a gap that you could cross significantly. But then also that's a problem that academic and industry researchers have known about for at least 60 or 70 years to figure out better and worse ways to ask questions in research context to understand those up front.

Speaker 14:43

Do you have a favorite method for discerning?

Speaker 1 14:46

We were talking before about like seats versus tokens, right? Yeah. And this is often a crucial, or I would say the highest leverage part of pricing. It's not the number, it's how you charge. So your determination of pricing metric is really important. And so if I say, hey, you're gonna buy this customer support resolution product, you do you have a preference between if we charge you per resolve ticket, per number of questions asked to our AI chatbot, or are you indifferent between those? That is a really powerful question because if I structure it correctly, and usually before I had asked that, I would say assume that no matter what you answer, the bill at the end of the year, what you're gonna pay us, is gonna be the same. But how we determine that bill is gonna be on either A, B, or you're indifferent between A and B. The Daniel Kahneman thinking fast and slow, system one, system two, folks, they would say that rationally you should always be indifferent. But that's not what you'll hear. You'll hear, I think it should be per resolve ticket because we could track how many tickets we have. We already measure that, and so that feels like I'm more in control, or well, you should maybe more by questions, because our most of our things are pretty simple to resolve, and I don't trust your system can handle the really complex questions anyway. So you're gonna end up billing us for a bunch of where the AI gets lost, and the person just sends us a bunch of messages and is rage typing into the chat bot on our website. Whatever they say, that's valuable information, and you don't need to wait until you launch to ask that question. That is information that you can gather upstream. And again, I'm not gonna take a strategic direction off of any one conversation. We're gonna talk to a bunch of prospects in your environment and be like, here, these are the common patterns of why people prefer A versus B. And then you're gonna map that against your strategic priorities, your constraints, and then determine the best approach.

Speaker 16:47

And I find that most people who do want to buy your product or are interested in deploying it, they are putting their job on the line in a way. They're putting their they're putting themselves out there, and they do generally want you to be successful too, so they'll tell you.

Choosing A Pricing Metric With Research

Speaker 17:01

We at position squared. If we are marketing the agency that now has released a software product, right, an AI product that takes all of our knowledge and puts it into a single system and then offers that to folks. But the experts still direct the strategy while the agents handle the execution, right? So now we find that while everybody wants to do outcome-based pricing, which is what you were talking about, a lot of that outcome-based pricing can be just aspirational because the the enterprise buyer many times just wants to pay for hours of expert advice, right? Hours of expert capabilities. So when an enterprise deliverable is produced by a hybrid combination of AI agents that save a tremendous amount of time by doing the execution, with human experts providing the governance, how do you prevent buyers from anchoring on traditional agency hourly rates or on just software seat prices?

Speaker 1 18:00

This is a great question. Rajiv, are you are you on Dan's couch now? Is this some some self-help?

Speaker 18:06

I figure since I have you here, I might as well ask.

Speaker 1 18:08

Perfect, perfect.

Speaker 18:09

Now remember, we have the additional issue that in marketing there's not as many salespeople. I can't just say, oh, for every million dollars, let me hire a salesperson, right? A team of five can handle a $10 million company or it can handle a hundred million dollar company.

Speaker 1 18:26

Yeah, yeah. I run into the same problem. So I'm a consulting firm. I happen to specialize in pricing, so I do value-based pricing for my engagements. I do want to flag that we've stepped out of B2B software and now into tech-enabled services, so there are some nuances. One of the key nuances is that outcome-based pricing, one is hard wherever you are, even when you can customize a sale. I would say that in a services business, that is generally, if you're selling a what's called a high-value services business, if I go get a haircut, I expect them to have the prices posted. But if I'm selling consulting or agency services, usually there's not posted prices. Obviously, there are exceptions, and therefore that allows me to price the client, not the service or the product directly. There are still challenges, and even then, it's really hard to do outcome-based pricing. It's really hard in product world. Even when you're doing individualized services, though, it's really difficult. Let me use an example. I go in and do a pricing transformation for a company. We do some really good work. Obviously, I expect them to improve net revenue retention, like we mentioned, probably overall ARR as well, probably lower CAC. But we're going to launch that. Usually that's tied to a new product that they're developing, or there's a whole bunch of new marketing and sales enablement material. And so the attribution of what I did to that process versus what the rest of the company had to input, right? All the meetings with the executives and all the people is quite difficult. However, that doesn't prevent you from understanding sort of what the overall business case is, what are their expectations, and then what is a reasonable value based upon the value you're going to provide that you could approach. But you mentioned something else there, which was how do you prevent them from anchoring? This is, I would say, maybe the primary challenge folks have to deal with in a services business, where it's not that they want to buy hours, it's that they've been trained to buy hours and they don't think about it any differently. There are many ways to approach this. One way that I found is helping to explain to folks little stories or anecdotes. If you buy something online e-commerce, do you ever pay for expedited shipping?

Speaker 20:49

Do I? Sometimes.

Speaker 1 20:50

So in that case, you'll pay more to get it faster, right? But if I hire you to do a marketing campaign for me, you're saying I'm going to pay you more if you go slower because it took you more hours. So if you could walk them through the logic, they realize that they don't actually want to buy the hours. But that's just how they've been trained, and they haven't really thought that, okay, that's actually not connected to the way that we're deriving value. Now, how do you untangle that when you have an existing business versus how do you handle that if there was a net new company? I think companies who are just fully embrace it and just rip the band-aid off and say, we're not selling hours anymore. There's many consulting companies. That's how they build their cost for a project and how people get promotions is how many hours they build that year, and et cetera. And so there's a lot of organizational constraints you need to rip off. But as I think about the individual customer conversation, helping them understand that pricing metric, in that case, the hour, is not tied to the value. In fact, they would very much appreciate if you could give them the answer or the marketing campaign in five minutes while you're on the discovery. Call with them instead of well, it's going to take us two months, three months to build all the material. But if you give them five minutes, I'm sure they'd pay you a ton more. And so it's about helping them understand that for themselves, and you just have to do that through some creative storytelling and some leading them through that process.

Speaker 22:18

That makes a lot of sense. I mean it is a very difficult thing. So we have to basically have to figure out a way to anchor them into something they believe. But I like what you're saying. People have been trained to do the wrong behaviors, so then you have to untrain them.

Speaker 1 22:31

That's not true.

Hybrid AI Services And Hourly Anchors

Speaker 22:33

So now you preach that free proof of concepts attract bad prospects and dilute brand authority. So in an enterprise market where there's a lot of AI pilots and maybe some AI pilot fatigue, how does a company price and package a 30-day proof of sprint? So it serves as a paid high commitment diagnostic while still de-risking the deal for the cautious buyer.

Speaker 1 22:59

Yeah. It's a good question.

Speaker 23:01

By the way, I'm still running into this with our software too, that I'm willing to build some agents with you, I'm willing to let you use some of my agents, but then at some point you have to pre-qualify them so they don't waste your time.

Speaker 1 23:12

Yeah. Yeah. Any pricing decision is going to be based off of four factors: your cost, competitive context, customer value, and customer willingness to pay. I think two of those are particularly relevant to your question. So the first is the value component. So in that situation, we can think about that transaction of a POC. Who is getting the value? Because in a POC, it's much less one-sided than in like selling them an existing product, right? Because that the hey, there's look at our logo garden, we've helped all these customers, it'll help you too, right? Buy into this vision. Here, you also need something in between that's not just revenue. You need, hey, we need someone to work with to understand all the things we just don't, the unknown unknowns, as it were, of where is our product not ready yet? And so that's gonna be a a slider or a balance, right? Where is that value accrue that's gonna impinge on that decision? The other thing that I would also be thinking about there is is the costs. The costs to you may be hey, we're gonna go in and spend a bunch of time with either our engineers or our professional services team, however, that deployment works in that scenario. And look, those people, opportunity costs, they could be building other products or they could be working with other customers, right? We have to understand like how much of that are we going to either eat ourselves, take a negative margin, just cover our costs, or positive margin, right? And that's also going to be colored by that first factor of value. And I would say the third thing which is relevant here, it ties into customer value, which is how much buy-in from that particular prospect do you need to invest your time because there's something magical that happens. In pricing, we'll call it the penny gap, right? There's an infinite delta in price between $0 and $10. When you get somebody to actually sign on the dotted line where money has to come out of their account, there is not only a level of sort of deliberation and decision making of oh, money has to flow out, there's a whole procurement process, right? But then there's also an emotional going back to I think Danny Kahneman and Amos Diversky, like commitment and consistency, right? Like I am putting forward my name on this project that now has dollars associated with it. And I think this could be where startups who've never dealt with enterprise maybe are like, oh, if we charge them $10,000 for this POC, it's the thing that doesn't even work. It's like they pay more per day to water their plants and their offices than $10,000. That's like a rounding error. But for that sponsor to say, oh, I paid $10,000, right? They're gonna show up to the meetings, they're gonna provide the feedback, right? If you don't, you're in a much, I think, weaker position from that regard to hey, we're investing all this time and we just feel like we're constantly chasing. And that's gonna be a judgment call depending upon the size of the business and then those other factors that I mentioned.

Speaker 26:37

You have to basically gauge that if I get them to pay for something, then they're truly committed. And it's worth my time to create to commit myself. And

Paid Pilots And The Penny Gap

Speaker 26:47

I think so. If you look at today's AI packages, right, on the consumer side, it's free for a certain level, and then it's paid for a certain certain other level. And now that's that that consumer experience is also translating over to the enterprise experience. There's a certain level you have to get to before you want to now commit your enterprise level. Are you seeing some change there? You're also dealing with software that can do just about anything, as opposed to software that was very task-dependent.

Speaker 1 27:14

Yeah, yeah. The fact that it can do anything, I think, is and I'm not doubting that as a as a truth, but rather I think that also puts up barriers, right? You're talking to someone who maxed out two $200 Claude Max plans last week. So I am definitely a buyer into what these things can do. But someone who didn't have a computer engineering undergrad and hasn't been in tech for the last couple decades, you show them an empty chat box and say, you could do anything. And they're like, I don't know what to do with this. So that doesn't necessarily translate to value, and that's often a gap. It's often a gap in the pricing conversation because people, the CFOs, like, hey, we're gonna put this thing out there, and if it's used, it's gonna cost us all this money, but just because it costs you money doesn't mean it is valuable to your end clients. The consumerization of IT that's been going on for a couple decades at this point, right? We saw that with everyone who wanted to bring their iPhones and hook it up to their Microsoft 365 on their phones, right? What I see day in and day out though is maybe the differentiation that folks have as they're bringing AI products to market, where saying, hey, we're Acme SaaS tool, right, for your business. We've added an AI chat agent to it, like everybody else. And I've been in research conversations for these type of products, and people are like, Yeah, you want to charge me X, but I could also just put this data into ChatGPT, and it'll give you the answer. And so there's there is an anchoring effect there, and then it's up to you and the business to understand okay, why is this better? Why is it better for them to interact with our chatbot than ChatGPT? Look, there's no one answer for every single product that's out there, but it is going to be an anchoring effect of folks saying, look, we already have a ChatGPT Enterprise account or with Anthropic or Google Gemini. Why am I going to pay extra for your add-on? And so I think that's something to keep in mind.

Speaker 29:05

It's also the sense of the roadmap as well, right? They've moved so quickly to add pretty astounding features, then you run into the roadmap barrier as well. You could use my capability and I will take you on a roadmap, but I see so much leapfrog with them. Why don't I just stay here? And even if you're ahead of them in this or that area, they'll catch up.

Speaker 1 29:27

Yeah, for better for better or worse. I think, I mean, I'm certainly not the only one to talk about this, but if we look at any company doing domain-specific work, right, Harvey in the legal AI space, right? And this happened multiple times over the last year. I think Anthropic released some security plug-in, and we saw the entire cybersecurity sector of the SP just get tanked 20% in one day. I don't know if that had a net impact on those companies' revenues because there's a broad-based product. I could write a book in Microsoft Word. Is that the best software if I want to write a book? There's probably better software to use, right? Or I could use Microsoft Paint versus Photoshop, but does it have exactly what I need in those use cases? And then the question is, are you niched down enough where you're not at risk of a general-purpose model or model company wanting to understand the in-depth workflows and nuances and caveats of your particular market? Such that, yeah, if Opus 7 got really good at understanding legal briefs, I would still use Harvey because it it doesn't include all of the workflows that our customers need and expect, and security, et cetera. I mentioned security, not saying anything negative about security of the foundation companies, but just whatever it is that is valuable to them.

Speaker 30:52

In the case of like a Harvey or something, that they can sell it. You can trust us because this is all that we do. And we're focused on your workflow. We've downloaded every case from every every courthouse and we've got specific workflows that are germane to you. Why not go this way? There's so much time savings. And you can be sure that when you talk to your client, that they're going to get good answers because they can't be wrong.

Speaker 1 31:15

Yeah, and even all the things that maybe we don't even think of, which is in that the legal team's not sitting on an island just doing legal analysis, right? They're getting a contract from sales and they're marking it up and they're sending it back, right, and giving the rationale, right? There's that whole collaboration workflow and review process. There's all of those things, right, that maybe don't even have to do with the uh very specific workflow of the analyst or the paralegal or the chief counsel, but all the interaction with the rest of the organization and how documents go through their systems, et cetera. We always have to think about sort of that that broader where it's like, oh yeah, I could throw this document into ChatGPT and have it give me its read, but that only solves 10% of my actual

Procurement Psychology And Buying Committees

Speaker 1 31:58

problem.

Speaker 31:58

And actually that leads us to the question about enterprise procurement teams. You would think that software purchasing, especially B2B software purchasing, is a cold, rational process, right? Yet consumer psychology shows that humans still buy based on emotion and primal triggers. So where does that notion of the primal trigger, the lizard brain buy button? Think fast, think slow, right? The thing that you do to help people think fast. How does it actually intersect with a 10-person enterprise software buying committee?

Speaker 1 32:30

This is a great question. In pricing, we have uh behavioral psychology branch of economics, right? I've got the guy who wrote Nudge on my on my shelf back there. Those things are important, but I don't think approaching economic actors as either fully rational or fully emotional is ever the approach, right? It's gonna be a blend and an understanding between them. I think one of the most common elements of that would be most B2B software has good, better, best pricing. There's a reason that tends to be the standard. It's not the only option. There are others that are important depending upon the situation, but in there we have the we call the magic of the middle effect, right? Most buyers will end up selecting the middle option because it feels like they're making the less risky decision. They're not overinvesting and getting all the bells and whistles, they're not cheaping out and maybe regretting it on the low end. I'll go with I'll go with the middle. I think the core of your question is we take a bunch of partly rational, maybe mostly emotional individuals and we put them in a room together and then assume that then we have one rational actor. That's never been my experience. If it's been yours, I'm curious to know. But I find that people still argue based upon what is going to do the best for their career, for their organization, their particular pet project or interest or pre-existing bias. And so those will always be important. This starts to get out of the realm of pricing more into sales, but I think one of the best pieces of advice that I followed is there's going to exist a buying committee, but oftentimes we're not going to be in the room with that buying committee. And what we need to do around our positioning is make sure that we're getting the full buy-in of our champion and then understanding that most of the rest of that buying committee can't say yes, but they can all say no. And so then, therefore, our job is to make sure we're enabling our champion with the right ways to handle the objections from the rest of those people who could say no. Because those conversations are, for better or worse, going to be happening when we're not in the rooms. We could say, well, if I was in the room with the chief legal counsel when they said this, I would say this. That that's nice, but you're not going to be in that conversation. It's going to be the champion going to be like, hey, I got this contract, did you just sign? And then the chief legal counsel is going to say XYZ. We need to make sure that we've prepared our champion and given him the materials to answer those questions effectively.

Speaker 35:05

So I think going back to what you were saying earlier, you really do then need to make sure that those who could potentially say no feel that your pricing is anchored in something that they can attach themselves onto as to the benefit that they're getting.

Speaker 1 35:21

Yes. And depending upon your product in the market that you serve, there's ways we could think about dividing the wallet such that maybe your primary champion only has budget or authority for a certain piece. And this all ties into overall company strategy and go to market, right? Maybe you you land with them and then worry about expanding later. Or if we thought intentionally about our pricing, you make those other pieces. Hey, you need this additional sort of compliance piece because your head of IT won't sign off, and then he knows what that part costs. If you look at the whole bill, you're gonna sticker shock, but also you said you needed this, and this is what this part costs. So if we can think strategically up front about how those are gonna break down, and a lot of this stuff early stage, you're not gonna necessarily know up front. This is gonna be things you have to figure out with rest.

Speaker 36:11

You have to understand if they look at things holistically or if they don't. So your champion has to help explain to you what's gonna happen in the room.

A Contrarian Model Without Subscriptions

Speaker 36:18

So now you've predicted that as AI feature velocity accelerates, most software products will actually get worse due to feature inflation and rush releases. What's the most contrarian pricing or packaging stance that a SaaS founder or AI founder could take right now that sounds crazy to the market today, but will look brilliant in two years?

Speaker 1 36:40

Oh man. I look at and I come across a lot of contrarian takes. I'm skeptical of many of them. One of the things that I've been thinking of a lot recently is are we in a worse position as a SaaS industry because we went from subscription to now having real cogs that we have to manage? And so people are layering on either these credit systems or other on top of like the base subscription, these other usage-based pricing. If I could put you in a time machine back to 2005, we're in a perpetual software world, would we be better prepared for this transition if we'd gone back there? I'll use an analogy. I go to the Toyota dealership but I buy a Toyota Camry. I don't expect Toyota to fill up my gas tank for the lifetime I own that car. But that's the situation we found ourselves in now with software. But if instead I had sold you the car, the original perpetual license for the software, and said, hey, you can also do all these other things, right? Just get these data sources and do this other analysis, but you got to top it up. Would that be more appropriate? Because we're juggling this intermediate state where people are like, I really love this subscription revenue. I don't want to add on with this AI. I think maybe the contrarian take would be the folks who completely wipe away the subscription component. And an idea would be instead of saying for your CRM, yeah, you pay per seat, and then for every account summary you generate, it's this many credits, and every email your reps generate, it's this many credits, and sales enablement, PowerPoint decks, it's this many credits. You just say, we're just gonna remove this seat thing as well, and maybe just say, if you want to spend credits for a human seat, like that has a credit price as well. You get a hundred thousand credits a month, you can spend them on humans, you can spend them on agents. I think we found over time, this is one of the reasons it never really works to micromanage your employees because your employees are like in the weeds, they see all the what you want to do is give them context. If they know the goal, they can make the better decision. And so if we take that same approach with vendor versus customer, right? The customer knows what they're trying to do with their business. Here, you just pay us, and then you do what you need to do with your business. And here's our model that allows you to do it. Hey, maybe in six months you're gonna fire all your human salespeople out. We don't care, you're but you're gonna have agents do that work instead. I don't care if you're still gonna use our software to do it. So that would be my contrarian take. We'll see if that happens in the future.

Speaker 39:20

And I get the software and I run it myself. Or I create a credit-based system that you apply in any way that you want, but it applies somehow to the value created.

Speaker 1 39:33

Yeah, so hopefully your credit system ties to value creation, right? If you've tuned it well, right? I think uh yeah, I don't want to go down a rabbit hole.

Speaker 39:41

I can't if you want, but hopefully, because I I really like the way you went with it.

Speaker 1 39:47

Maybe not even the perpetual side. I go to Expedia or whatever, and it's two-sided marketplace, so it doesn't quite work. But I use their search service, right? I don't pay them a subscription, right? I it's just a website. I've got an account, I go in, I do my thing, and I leave, right? They have a different business model, but they're not charging me a subscription to use it. Instead, it's just like, hey, come in, yeah, log into Salesforce. And if you want to do anything, you need to spend credits, but you why you're not gonna sign up for a 100 seats for an annual contract. Just use it. If you want Rajiv, who's on your sales team, to have be able to log in with his account, you're gonna spend some of that credit allowance and just go from there.

Speaker 40:24

I see. Pure credits, no seats, go to town, get value from it. I like it. This is great.

Speaker 1 40:30

It's contrarian. I don't know what that would go for on polymarket, but I'm gonna put it out there.

Speaker 40:35

I think it's a great way to I think it's a great way to end this segment because I think it's a lot of amazing, insightful thinking. So now we're gonna go to the game.

Spark Tank AI History Quiz

Speaker 40:44

Welcome to the Spark Tank. Today we're joined by Dan Balcauski, the founder and chief pricing officer at Product Tranquility. A former software engineer, Dan helps companies price cutting-edge innovation and quantify value in the modern software landscape. But today we're stripping away the modern stats pricing tiers and taking you back to the fundamental roots of machine intelligence. We are putting your computer engineering background and your analytical reflexes to the ultimate test with the algorithmic origins challenge. We're going deep into the wild history, foundational research, and iconic lore of artificial intelligence. Dan, are you ready to prove that you are sharp in the history of artificial intelligence as you are in B2B product strategy?

Speaker 1 41:26

I hope so. I did not prepare for this pop quiz, but I'm ready.

Speaker 41:30

That's right. That's the most fun about this show. You don't come in prepared, you don't even know what the subject is, and we're gonna have some fun together. All right. So this is multiple choice question, three possible answers. You guess, we have some fun chatting about it. Okay. All right, question one While AI feels like a modern phenomenon, the term artificial intelligence was originally coined at a legendary summer research project held in 1956, organized by John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon. Where did this founding conference take place? A. I know MIT. Do you know it? I know. It's Dartmouth. Oh. Nailed it! There you go. All right, so for those who want to play this game at home, it's MIT, Stanford, and Dartmouth. Those are A, B, and C. And the the Dan nailed the correct answer. The Dartmouth Summer Research Project on Artificial Intelligence in 1956 is widely considered the birthplace of AI as a formal academic discipline. John McCarthy coined the term artificial intelligence in his proposal, boldly predicting that a two-month, ten-man study could make significant progress towards getting machines to simulate human intelligence. And it only took us another 70 years to get close. All right. You nailed question one, question two. Alan Turing introduced his famous benchmark for machine intelligence in his 1950 paper, Computing, Machinery, and Intelligence. While the world knows it as the Turing test, what was the original name Turing gave to his hypothetical experiment? A the cognitive mirror test. B the imitation game. Or C the machine consciousness protocol. Why'd you pick A?

Speaker 1 43:36

Imitation Game. That was his work on the Enigma machine, I think. And I don't think he talked anything about consciousness in the during this paper.

Speaker 43:48

Alright, so that's why you're saying cognitive neuros.

Speaker 1 43:51

Yeah.

Speaker 43:52

That's amazing. Is that correct? But it's actually the most obvious one. It's the imitation game. Long before Alan read a movie about his life called The Imitation Game, Alan Turing used that exact phrase to ask if a machine could fool a human into thinking it was alive. So Turing originally called it the imitation game based on a party game where an interrogator tries to distinguish between a man and a woman communicating via text, and Turing proposed replacing one contestant with a computer to see if the interrogator could tell the difference. Question number three. Philosopher Nick Bostrom popularized a famous 2003 thought experiment to explain instrumental convergence and alignment risk. In it, an ultra-intelligent AI is given a seemingly harmless, mundane goal, but ends up destroying humanity and consuming all physical matter in the universe to achieve it. What product was the AI tasked with manufacturing? A, paper clips. B, wooden pencils, C, plastic water bottles.

Speaker 1 45:12

Paperclips.

Speaker 45:14

And why'd you give that answer?

Speaker 1 45:16

Paperclip maximizers. That's the uh that's the thought experiment.

Speaker 45:21

You nailed it. You nailed it. It is the paperclip maximizer correct illustrates how an AI doesn't need to be evil or hate humans to destroy civilization. It just needs a goal with no bounds. If a superintelligence is told to maximize clip production, it will logically convert Earth, humans, and solar systems into paperclip factories. It's the ultimate cautionary tale in AI safety. The world doesn't end with a supervillain robot. It ends because a machine took its office supply quota way too seriously. Crazy. All right, do you want to try number four?

Speaker 1 45:56

I would love to. Cue it up.

Speaker 45:59

All right, number four. In 2019, Turing Award winner Rich Sutton wrote a famous essay that sparked intense debate around the machine learning world. In it, he argued that seven years of AI research proved that hard-coding human domain knowledge into algorithms always loses in the long run to general methods that leverage two specific factors. What is the title of this iconic essay? A the bitter lesson. B attention is all you need. C beyond human engineering.

Speaker 1 46:35

A the bitter lesson.

Speaker 46:38

Why'd you say that? All right. Most people have to kind of work their logic into it. So you can get the answers. This is awesome. So in the bitter lesson, Sutton argues that the biggest breakthrough in AI doesn't come from human cleverness or handcrafted rules, but from relying on computation and learning, specifically search and scaling. It proved bitterly disappointing to AI researchers who wanted human intuition to be the secret sauce, showing instead that brute force compute always wins. It forced AI researchers to accept the harsh truth: your clever human hand-in-jured rules aren't going to be giving more compute to a massive neural net.

Speaker 1 47:27

I will say that going back to our Harvey conversation earlier, the bitter lesson, depending how AI gets better or not, would be my biggest bet on why that hypothesis doesn't hold the conversation we had earlier. We've we spent all this time teaching machines how to help organizations do the thing that they do. And if we just let them loose in an organization, they would figure it out and help them better than we could design those systems. So we'll see.

Speaker 47:55

We'll see. I look, I'm banking on the same thing for what my company's doing. I believe that my 20 years of knowledge and all the things that we've learned will translate into better AI for my clients in the mid-market. But we're gonna find out.

Speaker 1 48:07

We're gonna find out. I think we're all gonna find out.

Speaker 48:10

That's part of the fun. I think experts, expertise, even with compute, still matters. Okay.

How Dan Got Into Tech

Speaker 48:17

Did you always know you wanted to work in technology? Was there a specific moment or project that sparked your passion? How did you discover that? What got you sparked?

Speaker 1 48:26

Yeah. Interesting question. I was always, I guess, very talented in math and the sciences. And then my I uh timing tends to be everything. I graduated high school in '99, so peak.com mania just happened to align my junior year of high school. One of the AP courses I could take was AP computer science. Learned C and object-oriented programming when it was all right, computer engineering, gonna go be retired by the time I'm 25. Didn't realize we were at the peak of the mania. I graduated into the the post-crash, things turned out differently. No, no regrets. It didn't have anything to do with me either, but that was the trajectory there. And then just I'm uh in insatiably curious. The reason I studied computer engineering is because I had I dealt with computers, I understood how software worked, but then I always had this question of, yeah, but how does someone program the first computer? I understand we have these this bootloader and it loads an operating system and then a write application on top of it, but like how did we teach the first computer to do anything?

unknown 49:31

Yeah.

Speaker 1 49:31

So I studied computer engineering to figure that out, and then I was like, okay, that's interesting. What's next? And so went went from there.

Speaker 49:37

That's super cool. It's actually one of the reasons I really enjoyed electrical engineering. We got down to the chemistry and built up from there.

Speaker 1 49:43

So our computer engineering undergrad was a third double E, a third comp sci, and a third there was computer engineering specifically. So we had transistor classes and plenty of double E classes with Maxwell's laws and magnetic fields traveling around wire and a bunch of classes I didn't do so well in that domain.

Speaker 50:01

But you understood how things work.

Speaker 1 50:03

Yes. Exactly.

Sabbatical Lessons On Human Value

Speaker 50:06

In 2018, you took a 21-country sabbatical, including volunteering in the dishing temple at an intentional community in Sweden and helping a language school in Colombia. When you completely stepped away from SaaS spreadsheets and CAC or LTV calculations, what did that experience teach you about how human beings fundamentally perceive value and fair exchange?

Speaker 1 50:32

Oh wow. Those are those are you set me up with an entirely different frame of mind than you closed the question with. I don't know that I was particularly concerned in that time period about value exchange. I would say kind of the biggest thing I did during that time period was I actually did two 10-day meditation retreats. And if anything taught me there about perceptions of value, it's a realization that most of our perceptions of value are not things that we chose up front. They were accidents of our own upbringing, and that the Buddhist tradition, there is only craving and aversion. And so the things that we think we want are just illusions of our mind telling us that this is the thing that's going to make us happy. And not to say that there isn't real value, but it is deep in your mind. It does not exist in our bits and our atoms. So I think that gives a lot of power to marketers out there to if you can help shape what that value is in the minds of your customers, this is why you have a job and we'll continue until the bitter lesson takes over.

Speaker 51:45

That's great insight. Let's take you back more to the current.

Why Pricing Turns Political

Speaker 51:49

As a program leader for Kellogg's executive ed program in product strategy, you teach formal business framework concepts. So, what is a single piece of classic business school pricing theory like price elasticity curves or conjoint analysis that looked great on a whiteboard but fails on contact with an early stage SaaS startup?

Speaker 1 52:11

Yeah, no, price elasticity is a great tool for economists, but talk to any B2B CEO of any sort and ask them what their demand curve is, and you're gonna get a blank stare. It doesn't exist in reality, especially in B2B SaaS for a whole bunch of reasons I won't get into. But I think being an engineer by background, this was definitely a mistake that I had, a bias that I've had to correct over time. I used to think the pricing was a math problem. It turns out it's a political problem for the most part. There's often those disagreements in the senior executive room that I hinted at earlier of it's not that nobody has any idea what they're doing, but each everyone has an idea of what they should do, and none of them overlap. It's not the solution to a math problem that you need to solve in that situation.

Speaker 53:00

So it's not just this priceless CD curve, it's much more about how to understand the frame of mind of the people that are in the room and construct your answer on that puzzle. We always asked our guests to name a historical event or person or movement that inspires you, and you answered the framers of the U.S. Constitution. What in particular about that journey lights you up?

Speaker 1 53:21

We talked a lot about AI, and everyone's trying to figure out how to get these agents to do what I want. Anyone who's been in management has had this struggle ever of I've got all these humans and I need them to work towards a common goal, and it's so complicated. There's an endless number of books in the business section of your local bookstore on management theory. And I just think that the the framers of the Constitution just did such an amazing job of understanding human nature and then constructing those frameworks. Obviously, they had benefits of history, but they didn't have this in the 1700s, they didn't have technology or the access to resources we have. Just the level of understanding they had about independent agents and how they'll try to work around your system and all of the checks and balances that they built into that frame that's survived so well. Not to say that they those bounds don't constantly get tested by different agents in a system. But I think anyone who's building today with these systems are like, man, if I could think at that level how to set up my guardrails, if I could create a constitution for my AI agents so they did what they wanted within some bounds and not, I don't know, go out and attack another billion-dollar tech company. Like I think we'd be in such a better place, just at the individual and the societal level. So I think it's just highly inspirational to understand what they did.

Speaker 54:43

Yeah, they seem to they really understood human behavior. I don't think they sat there thinking, oh, this will be a 300 million person country, right? That would be leader of the free world back when, right? They didn't even have light bulbs back then. Electricity, nuclear power, any of the things that we have today. So to be able to forecast how human behavior was with the right guardrails as you talked about, it's just really incredible and how it evolved.

Rapid-Fire Curiosities And Closing

Speaker 55:10

What's a piece of technology from your childhood that you genuinely miss? And what would you bring back if you could?

Speaker 1 55:15

Man, everything has just gone so obsolete. I don't know there's any single piece of technology. Even uh I guess the stuff that hasn't gone obsolete, I just remember my dad had this giant vinyl collection, and I grew up in the 80s, so we had cassettes and then CDs, and I always used to rib him, be like, Why are you keeping all these vinyl records around? Old stuff, right? And even that's made a comeback. I bought my partner a a record player for a birthday this this year, so that's not gone out of style.

Speaker 55:43

There's something about taking your music and having it embedded into a physical form and then listening to it, and it's not perfect, it's got this full range of sound to it, and then it's something you can put up on your wall too, right? The algebra cover. So there's something beautiful about it. My son is like this, he my oldest son, he loves this. He has yeah, he has he's always loved vinyl records forever.

Speaker 1 56:05

Yeah, interesting times.

Speaker 56:07

Yeah. What's something you're currently learning or trying to get better at that has nothing to do with advancing your career?

Speaker 1 56:13

That's surprisingly difficult. Everything just went through my head in some way tangential to running a business. I would say, like everyone else, trying to just figure out this wave of innovation, right? Which is much broader than any business purpose. It seems like every day we're getting hit with oh, have you used Grockbot? Have you used Muse? Have you used dots, right, to run your life? And so that feels like a full-time job just to keep your head above water and not get lost in conversation. And obviously, those things go much beyond career progression. But that's I would say most of my world these days hinges around a lot of the AI topics we've already touched on.

Speaker 56:54

So it's amazing. It's amazing. You're in it. You're in it, you're living it, especially down in Austin. What's something you thought you'd have figured out by now, but you're still completely confused by?

Speaker 1 57:04

You know, in general, running a business. I've been running Biden for seven, eight years at this point. Yeah. I think like anything else, right? You uh I think a healthy way to view those type of things is like a video game, right? You like hit level one, you think you're the boss, you go to level two, you're like, oh, these bosses are harder. Level three, these bosses are even harder.

Speaker 57:20

What's a question you wish people would ask you more often? And what's a question you wish they'd stop asking?

Speaker 1 57:26

Got some real zingers in here. I find it difficult to answer the question because I generally love conversations, so I don't think that there's a question I really get tired of answering. Maybe when I teach the course, if people are like, can I get an extension on an assignment? That's a question I really don't love answering over and over again.

Speaker 57:48

That's especially hard in exec head.

Speaker 1 57:50

Yeah, yeah. There's no penalty for you're just not going to get feedback if you decide it late. Stop asking me.

Speaker 57:55

If you could sit in on any meeting happening anywhere in the world right now just to observe how it's run, what meeting would you choose?

Speaker 1 58:03

Oh, that AI summit that just happened at the White House. I would have loved to be a fly in the wall for that thing.

unknown 58:09

That would be great, actually.

Speaker 1 58:10

It's like you've got all of these high-powered CEOs in a room together. I want to see what it looks like behind closed doors when those that type gets because every single one of them walks into any other room and they're the most important person in that room, bar none. So what happens when they all get together in that room? Who speaks first? And do they cut each other off as the other one's speaking? And do they pull the same shenanigans that they do with their employees?

Speaker 58:36

And then they have to come in with some value to bring.

Speaker 1 58:40

Yeah.

Speaker 58:40

They have to come in with some agenda, some value they they want to bring, some message they want to get across, and then answer any questions or have a conversation, right?

Speaker 1 58:49

Yeah. It always is interesting. I feel like people go through multiple stages in their career, which is like I want to be in the room where the decisions are happening, and the next stage is please get me out of this room.

Speaker 59:02

So I I love that answer. That would have been a very I think there's what 20 people in the room that all were top executives of some of the biggest companies out in the world today, making all kinds of decisions.

Speaker 1 59:20

And responsible for at least 98% of the SP value at this point.

Speaker 59:26

Pretty much that. It was interesting how they were communicating with the leader of the free world.

Speaker 1 59:30

Yeah.

Speaker 59:31

Dan, that was great. I really enjoyed talking with you today. You're a heartfelt guy. You really understand pricing and depth and the thinking behind it. And just a you're just a fun guy to be around. So thank you so much for joining.

Speaker 1 59:45

I appreciate you having me. It was a good conversation. I enjoyed the game too. That was that was fun.

Speaker 59:49

I love how you aced the game. All right. Thanks for listening. If you enjoyed the pod, please take a moment to rate it and comment. You can find us on Apple, Spotify, YouTube, and everywhere podcasts can be found. The show is produced by Anand Shah and edited by Laura Ballant. Production assistants by Taryn Talley. I'm your host, Rajiv Parik from Position Squared, a leading growth marketing company based in Silicon Valley. Come visit us at Position2.com. This has been an F Funding production. We'll catch you next time. And remember, folks, be ever curious.