Siddharth Ramakrishnan

Writing

The Pricing Ladder

September 19, 2024

Everyone is talking about outcome-based pricing. AI agents are going to come and take over software, and AI agents can actually do things. They’re going to take tasks and complete them end to end, so it doesn’t really make sense to charge for your product on a per seat basis anymore right? $20 a month (per person but you really only need one account) for your AI agent to close 10k support tickets? That doesn’t really seem fair.

The idea is simple: instead of charging based on the product or usage, you charge based on the actual outcomes you help the customer achieve. Instead of $20 a month, why not just charge $0.05 per ticket closed? That seems like a better deal for everyone.

Unfortunately, I don’t think most people realize just how difficult this shift will be.

The Ladder

Chaos Pricing is a ladder. The lowest rung is what we all know and love: basic pay-for-product model, where customers buy your product outright. From there, you move up to seat-based pricing, where companies pay for each user or "seat" that needs access. After that, many companies shift to usage-based pricing—paying for how much of a service or product is actually used, like cost per API call, data usage, or emails sent.

Then, at the top of this pricing ladder is outcome-based pricing, where you get paid based on actual results—clicks, conversions, sales, or whatever outcome you're targeting for your customer.

Moving Up the Ladder Isn’t Easy

At each step, you’re not just changing the way you charge customers; you're fundamentally changing the way your business operates.

For example, at Attentive, we're in the process of moving from a usage-based model (charging per send) to an outcome-based model (charging per click). In theory, this makes a lot of sense—we work in the advertising space, where outcomes like clicks and conversions are easy to define and have been around for decades. But even for us, with a clear outcome to track, the transition has been full of hurdles.

One of the biggest challenges? The technology. Moving from usage-based to outcome-based pricing requires a whole new level of tracking. If you're used to charging based on "seats," you need to build infrastructure that tracks "usage." If you’re charging for usage, you now need to track outcomes. And that means more / different data to manage, more integration with customer systems, and more complexity.

It's much more straightforward to move from one rung of the ladder to the next. Jumping from seat based pricing to outcome based pricing is going to require some tech / data that'll take longer to actually build.

Convincing Your Customers

It’s also not just about the technical hurdles. You also need to convince your customers to come along for the ride. First, they need to believe that whatever you’re tracking (clicks, conversions, etc.) is accurate. And second, they need to believe that they’re still getting more value from this new deal (otherwise why would they change a contract that’s working for them?).

This is the biggest reason why moving up the ladder is so difficult. Changing your pricing model causes uncertainty in customers—if they're used to paying based on usage, suddenly paying for outcomes will come with a lot of open question. If the AI agent closes a ticket but then the customer re-opens it, do we need to pay for that? Is that 1 ticket or 2 tickets? If the AI needs to escalate to the company, how do we deal with that? Who do we blame if a customer is still unhappy after some interaction?

One way to ease the transition is to take it one rung at a time. Moving gradually allows both your team and your customers to adjust without overwhelming anyone. If you try to leap from seat-based pricing all the way to outcome-based pricing, you're probably in for a rough time.

External Factors

Outcome-based pricing also conceptually is a great idea (it’s used in a lot of ways already, like sales compensation), but many outcomes online are not as measurable as people want them to be. In the advertising world, there’s an ongoing back and forth with buyers on “attribution” and figuring out who should get credit when a customer clicks and buys. The Facebook ad that exposed them to the brand in the first place? The Google ad a week later where they actually clicked? Their friend who reminded them they were going to buy something? Internet ads, with 20+ years of precedent, don’t have clear definitions of these things, so I anticipate it’ll be hard for other areas to come to agreements with their customers as well.

Additionally, outcomes might not even be within your system’s control! AI agents can help customers achieve certain goals, but those goals—those outcomes—might not be something you can easily measure or track. And if you can’t track it, how are you supposed to charge for it?

Ultimately, while AI will change the way we approach pricing, it’s not a magic wand that can solve the inherent challenges of changing pricing models.

Disruption

Disruption, everyone's favorite word. In the AI shift, we see the larger incumbents having most of the advantages. They have more compute, more talent, better distribution, the works. But existing players will have a lot harder of a time shifting customers over to different pricing models, and startups will actually have an advantage here. All the tracking and data that you need can be built from scratch so it's exactly what you want. The contract you initially sign will have outcome based pricing (or whatever other pricing model) from the get go! No need to herd cats and convince people that it'll be net better than an existing model. Sales will still be tough, don't get me wrong. You're likely pitching someone a rip and replace of something that works for them, but this seems like a cleaner path forward nonetheless.