We changed AI pricing three times before we had it right
What happens when your AI products evolve faster than your pricing model? Here's what we learned from changing ours three times.
Lucy Field
Vice President, Product Management at Zendesk
Laatst gewijzigd 18 november 2026
Over the last two years, Zendesk has been pioneering outcome-based pricing for AI, where customers pay for the outcomes AI delivers. And when you're building something without a playbook, you have to be prepared to change course as you learn. We've done exactly that, changing our pricing model three times in two years.
First, we extended seats to a price per automated resolution. Then we introduced tiered resolutions. Most recently, we introduced Zendesk Unlimited. Each version was right for what we knew at the time, and each went out of date once our products and our customers showed us what came next.
Here’s how we worked through those changes, what we learned along the way, and what I’d pass on to anyone starting the same work.
Moving from seats to resolutions
By 2024, it was clear that seats no longer described what customers were buying from us. With AI, customers could receive value when no human was involved. The number of people logging in was becoming a less meaningful measure of the value customers were getting. So we priced our AI agents per Automated Resolution. A customer paid when an AI agent resolved one of their customers' issues, giving the customer and Zendesk the same goal from the start.
That first version held an assumption that valuable work would always look like a conversation ending in a resolution. For a while that held. Most of what our AI did was answer customers, and a resolved conversation was a fair way to measure its work.
Not all resolutions are equal
Our second model added tiers to Automated Resolutions. As Zendesk’s AI covered more use cases and became more advanced, it seemed a reasonable way to introduce value-based pricing that reflected our offering. Over 2025, we invested in building out this new iteration of our pricing model, while also transforming the underlying platform to enable the level of flexibility and scale we expected to need over the coming years. Meanwhile, the Zendesk product teams were innovating at lightning speed. As we neared our launch of Automated Resolution Tiers, we came to a sobering realization that the market and the product had already superseded the pricing model we’d spent so much time developing. And the reason was becoming increasingly clear. AI was starting to deliver value in ways that had little to do with resolving a customer conversation.
Why AI pricing needed to move beyond conversations
This product shift reflects a bigger change in service. AI can increasingly understand intent, coordinate systems and take action. In fact, our products started doing work that had no conversation in it at all.
We call this broader shift Outcome AI™, a focus on the value AI delivers, rather than the interactions it handles. And our pricing had to catch up. Zendesk was evolving from a customer service platform primarily used by humans into a much broader AI-powered automation platform. In this world, huge amounts of value are delivered without a human and without a conversation. Resolved conversations did not account for any of it.
With this major shift in mind, we went right back to the drawing board. We evaluated all of the value that Zendesk was delivering, built a framework to categorise and articulate that value, and asked ourselves how we could ensure our customers were getting the most out of Zendesk as possible while still maintaining visibility and control over costs. The answer we landed on was: Unlimited Seats and an Outcome Budget with easy to understand outcome types.
A budget in place of a price list: how the Outcome Budget works
The next challenge was translating that framework into a pricing model customers could easily understand and manage.
The easy route was to list every AI feature, work out what each one cost us to run, and pass those costs on separately. But that could leave customers with a bill they couldn’t predict and a long list of prices to compare. Other enterprise companies have gone that way, with hundreds of distinct actions but we didn’t want our customers to need a spreadsheet to estimate what they would spend each month.
With an Outcome Budget, each plan includes a budget in dollars. Conversational and Operational AI outcomes draw down on it as they are delivered, so a single budget covers every eligible type of outcome. Customers can see what is included, how many outcomes have been counted and how their Outcome Budget is being used. Notifications and controls keep them informed well before a limit.
Why we introduced unlimited seats
That pace of innovation also led us to a bigger question: if AI is now driving value at scale, why should adding another team member to Zendesk be a budget conversation at all? The new plans come with unlimited seats, making it easier for people across the company to access the service work and insights relevant to their roles.
Service interactions can reveal more than what a support team needs to resolve a request. Marketing teams can understand how campaigns affect customer questions. Product teams can spot recurring issues and emerging trends. Logistics teams can see where fulfillment creates friction. With broader access, these teams can bring their expertise to service and learn from what customers are experiencing. That shifts the human role from handling every request to providing the expertise and empathy that the work needs, and it lets customers bring people and AI together without counting licences. It also answered something customers had raised in our research, that paying for seats and for automated resolutions at the same time felt like paying twice for the same work.
Pricing impacts the whole company, and a model built entirely on outcomes means aligning the entire company strategy and model around this singular goal.
Building a pricing model that can evolve
The reason we could keep changing is that we invested in platform foundations ahead of the pricing model. Every version of the model touches billing, quoting, reporting and checkout experiences, and if each change had meant rebuilding those systems, we would have stopped after the first one. We rebuilt the platform around reusable building blocks, such as offerings, meters, allowances and transactions, so that putting a new product on the rate card means combining those blocks in configuration.
What cannot be automated however is change management. Pricing impacts the whole company, and a model built entirely on outcomes means aligning the entire company strategy and model around this singular goal. Consistent and clear communication became one of our core jobs as a project team. At one point we had four names in circulation for the new pricing model, which confused the people building it, never mind customers. These types of challenges may sound small but every confusion is a paper cut that erodes customer trust. We needed to establish robust systems for tracking, change management and communication to ensure the entire company was moving to the same drum beat at all times.
What comes next
I'd love to say that the title of this article is true and that we now have the luxury of being right but we all know that's not true.
This new model is launching into a world being upended by new technology, a market that is shifting constantly and new AI capabilities are being delivered daily. We expect our customers' needs to change and our product to continue to evolve - our Outcome AI™ Platform is designed to evolve with them.
One thing that hasn’t changed: our customers need to be able to trust that we’re building what they need, with pricing that remains transparent, predictable and within their control. We’re grateful to the many customers who have helped shape this model through their feedback, and who continue to work with us as we evolve it. That ongoing partnership is what will help us get it right.
We have changed direction three times as an organisation and are learning how to do it well as we know it won’t be the last.
Lucy Field
Vice President, Product Management at Zendesk
Lucy Field is VP of Product at Zendesk, where her remit spans the Foundation platform — security, privacy, infrastructure, AI development and monetization — including the move to Outcome AI™ Pricing, the first complete shift away from seat-based pricing at enterprise scale. She loves leading teams, reinventing how work gets done and exploring new technologies.
Explore Zendesk Unlimited
Outcome AI™ Platform Plans bring unlimited seats together with an Outcome Budget for eligible conversational and operational outcomes.