Artificial intelligence is changing how technology work gets done, raising a commercial question for suppliers and customers alike: if AI can complete work that once required hours of human effort, should clients continue paying primarily for time, licences, or usage?
Outcomes-based pricing offers an alternative. Under this model, some or all of a supplier’s fees are linked to agreed results, such as successfully processed transactions, resolved customer enquiries or measurable operational improvements. As AI and automation take on more complex tasks, businesses are reconsidering how technology value should be measured and priced.
Traditional technology contracts are relatively straightforward. Software is priced per user or subscription, while professional services are often billed by the hour or project. These models provide a clear basis for budgeting, but they do not always reflect the value a customer ultimately receives.
AI complicates that relationship. An agent that completes a task autonomously may deliver value by reducing manual intervention rather than increasing software usage or the number of people involved.
Deloitte's analysis of outcome-based pricing in agentic AI software examines how pricing arrangements can evolve as AI agents perform more work. The commercial challenge is not simply deciding what to charge for, but establishing which results can be measured reliably and who should carry the risk when expectations are not met.
Consider a hypothetical accounts payable operation. An automation provider might traditionally charge an implementation fee and an annual software subscription. Under an outcomes-based arrangement, part of the fee could instead depend on successful invoice processing, reduced rework, or shorter processing times.
The challenge is agreeing on what counts as success. Faster processing does not necessarily mean fewer errors, and an invoice can be processed correctly while the wider payment process remains delayed by missing approvals.
A contract that rewards transaction volume without accounting for quality can encourage the wrong behaviour. Equally, a provider should not necessarily be penalised for delays caused by incomplete customer data or factors outside its control.
Research from HFS Research, based on a survey of 101 financial services firms, highlights this tension. Its 2026 report, Stop Pricing AI Like Labor, found that 82% of respondents still relied on traditional full-time-equivalent, service-level-agreement or licence models. At the same time, 67% had introduced business-outcome terms into AI services contracts.
The findings suggest that interest in outcomes-based pricing is growing, but established commercial models remain firmly in place. Hybrid arrangements that combine traditional fees with measurable outcomes may offer a more practical transition.
Outcomes-based pricing depends on evidence. Both parties need to understand current performance, identify what should improve, and agree on how results will be measured.
Process mining helps by reconstructing how work actually flows through enterprise systems. It can reveal bottlenecks, repeated processing, unnecessary handovers and deviations from intended procedures, giving organisations a baseline against which to assess whether automation has improved performance.
For example, a business introducing AI into accounts payable could track invoice cycle times, exception rates, rework and compliance with approval controls before and after implementation. These measures provide a more credible basis for evaluating performance than simply counting the tasks an agent completes.
Process intelligence also helps distinguish an agent's contribution from the wider process. If an AI agent processes an invoice in seconds but the invoice then waits days for approval, measuring the agent alone would overstate the improvement delivered to the business.
Pure outcomes-based pricing will not suit every engagement. Some services have clearly defined outputs, while others involve complex implementation work or depend on factors neither party can fully control.
A hybrid arrangement can combine a fixed or recurring fee for the platform and ongoing support with a variable component linked to agreed results. Before entering such a contract, customers and providers should define the outcome, establish a baseline, clarify responsibility for external factors and agree on how performance will be monitored.
As AI changes how technology services are delivered, organisations will need to reconsider how they define and measure value. Not every contract needs to become outcomes-based. The immediate priority is to establish which business results can be measured reliably, and whether the commercial arrangement reflects the value the customer is actually seeking.
That requires more than a new pricing model. It requires a clear understanding of how work gets done, what needs to improve, and whether the investment has delivered the expected result.
─────────────────────────────────────────────────
What is outcomes-based pricing?
Outcomes-based pricing links some or all of a supplier's fees to predefined results, such as completed tasks, resolved customer enquiries, or measurable operational improvements.
How is outcomes-based pricing different from usage-based pricing?
Usage-based pricing charges according to consumption, such as API calls or transactions processed. Outcomes-based pricing charges according to an agreed result. A hybrid model can combine both.
Why is AI changing technology pricing models?
AI can perform work previously measured through human labour or software usage. This encourages providers and customers to price based on completed work, productivity gains, and other measurable results.
What are the risks of outcomes-based pricing?
Disputes can arise over how outcomes are defined, whether quality standards are met, how improvements are attributed, and which party is responsible for factors affecting performance.
How can process intelligence support outcomes-based pricing?
Process intelligence can establish baseline performance, identify operational bottlenecks, and track process changes, helping both parties assess whether an intervention has delivered measurable improvements.
Get Updates and announcements from the Verdant Data Team
You can unsubscribe any time. Learn more about our Privacy Policy
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
%20(1).jpg)
.jpg)
.jpg)
.png)
