For the last year or so, there’s been plenty of discussion around OutcomenBased Pricing (OBP) in CCaaS and CX. Rather than paying for user licences or token usage to add AI capability to existing software systems, organisations are charged based on the results achieved by AI agents.
The pricing model has met with significant scepticism. While only paying for results sounds great, practitioners and CX leaders have questioned its practical application and pushed back on the promises made by vendors.
A lot of outcome-based pricing is actually tied to workflow outputs rather than business outcomes. It can be very hard to measure and attribute what, specifically, contributed to a particular result.
Though the concept may have been oversold and organisations need to be cautious when adopting the model, there is definitely a deliverable benefit to it.
Prominent software vendors like Salesforce and Zendesk have actively championed outcome-based pricing. During a recent visit to Melbourne, Jon Aniano, Senior Vice President and General Manager of CX for Zendesk, explained Zendesk’s OBP approach, where clients are billed only when an AI agent successfully resolves a customer issue or enquiry without escalating the interaction to a human agent:“Zendesk charges for automated resolutions—that’s actually fully resolved customer service interactions that did not have a human in the loop. We’re not talking about deflections; we’re not talking about a customer starting, not getting an answer, and walking away. These are all LLM-evaluated outcomes where there has actually been a resolution, and that’s what you get charged for with Zendesk’s outcomes-based pricing today.”
Even though there is a benefit, CX leaders and their organisations need to be careful and prepared. Audrey William, Founder and Principal Analyst at CrayonIQ, advises, “Evaluating outcome-based commercial structures requires a comprehensive analysis of the total cost ecosystem. Successfully delivering a customer outcome depends on the combined performance of underlying AI models, orchestration layers, enterprise integrations, knowledge bases, human advisers, and compliance frameworks.”
William notes that organisations evaluating these modern commercial models must maintain complete visibility over a number of core areas:
Cost-driver transparency: Clear line-of-sight into the underlying costs across both computational inputs and human labour.
Operational ratios: Real-time metrics tracking the balance between automated processes and human-driven tasks.
Governance controls: Explicit oversight mechanisms when delegating customer interactions to autonomous AI agents to mitigate operational, legal, and reputational risk.
Value & satisfaction metrics: Transparent, measurable benchmarks for evaluating true business value and customer satisfaction.
William comments, “Most CCaaS vendors are still stuck charging per agent seat, but that model’s shifting; AI cuts right across the whole enterprise now. Buyers are increasingly exploring the outcome-as-a-service model. There are many definitions out there as to what constitutes this model. The real trick with outcome-based pricing is proving the value to the CFO. Are you charging for the second handoff while the first one’s free? What does pay-per-resolution entail? At the end of the day, boardrooms are asking for a clear five-year forecast on their AI spend. Traditional vendors must pay attention here if they want to stay in the game.”
Defining success and defending the metric
A core challenge of outcome-based pricing lies in defining an “outcome.” Tying pricing to broad indicators like Customer Satisfaction (CSAT) or Net Promoter Score (NPS) introduces too many variables, making precise attribution difficult. Conversely, charging for simple actions like submitting a form or providing a greeting reverts OBP back to basic consumption- or usage-based pricing.
In CCaaS, the standard unit of measurement is often the resolved conversation or resolved ticket. Yet, establishing when an issue is genuinely solved can be difficult. If the metric is overly broad, it becomes impossible to measure accurately. If it is too granular, the vendor is essentially charging for product usage rather than true value.
Is an interaction resolved when an AI agent closes the chat, or when the customer does not reopen the ticket within 72 hours? Aniano explains how Zendesk caters to this by introducing strict heuristics into their model: “Zendesk’s automated resolution model mandates that an interaction must involve zero human intervention and pass an LLM adjudicator that verifies true resolution. If a customer reopens a ticket or files a similar request within a 72-hour window, the resolution is invalidated.
“When you look at what some other companies are doing in the market, they aren’t going as far; they’re just saying, ‘Hey, if the ticket was closed, it counts as a resolution.’ That leans more toward basic usage-based pricing. From the very beginning, we’ve said we want to make sure we’re charging you for actual resolutions.”
CCaaS and CX platforms rely heavily on external ecosystems: CRM databases, inventory systems, APIs, and clean knowledge bases. If a CX AI agent fails to deliver an outcome because the client’s internal API timed out or their knowledge base contains outdated info, does the vendor still get paid despite performing the task correctly? Or, when an AI agent handles 80% of an issue and escalates the remaining 20% to a human agent, does partial credit get assigned to the vendor?
Managing financial volatility
Pure outcome-based revenue fluctuates with customer business volume, macroeconomic conditions, and seasonality. A sudden drop in a customer’s incoming call volume can directly hit the vendor’s top line, making revenue forecasting and investor reporting difficult.
Alternatively, if an automated CX campaign or AI agent performs significantly better than expected, the customer could face a monthly bill that far exceeds what a traditional subscription would have cost, leading to contract pushback.
AI-powered CCaaS solutions rarely deliver perfect outcomes on day one. AI agents and routing models require training data, knowledge base cleanup, and fine-tuning. Defining who pays for system overhead during the initial ramp-up phase needs to be carefully negotiated.
Despite these commercial forecasting challenges, the alignment between customer ROI and vendor revenue remains a compelling growth engine. As Aniano highlights, “Aligning the value we deliver to how a customer pays us gives us the incentive to drive more resolutions for them, make the AI better, and help them adopt. It means they’re only paying us when they’ve got an outcome that they feel is valuable to them.”
Outcome-based pricing in CCaaS and CX represents a fundamental shift in how enterprise software value is calculated. While challenges around attribution, operational baselines, and contract terms persist, market adoption continues to move forward.
The vendors that succeed won’t just offer advanced automation; they will provide clear measurement terms, full audit visibility, and balanced hybrid structures that align platform performance with client growth.