You close a six-week engagement for a UK e-commerce business. The brief was customer service automation. You billed £38,000. Twelve months later, the client sends a good news email: their support headcount has reduced by three roles, saving around £90,000 a year. They are delighted with you.
You got paid once. They will receive the benefit every year until they change the system.
This is the structural problem with day-rate billing for AI services work. The value you create keeps compounding after you have left the building. The pricing model that bills the clock captures only the first few turns of it.
What is value-based pricing and how does it differ from a day rate?
Value-based pricing ties your fees to measurable outcomes the work produces rather than the hours it takes to deliver. For an AI services firm, the practical shape is typically a hybrid: a fixed base covering discovery, scoping, and delivery, plus a variable component linked to agreed business improvements such as cost savings per quarter, workflow throughput, or revenue attributable to the engagement.
A day rate caps your revenue at hours multiplied by your rate. Once the invoice is paid, your financial relationship with the client’s outcome ends. A value-based structure captures a share of the economic improvement you generate, which for AI work is often significantly larger than the delivery cost.
Bessemer Venture Partners’ analysis of how AI software companies price their products shows a common hybrid structure: a platform or project fee covering the fixed cost of delivery, with additional fees tied to verified usage or outcomes. Services firms can mirror this directly. The discovery and delivery phases stay on a fixed base. The outcome-linked component is where value pricing lives. Stripe’s analysis of AI pricing confirms the same pattern: pairing a platform fee with variable pricing for outcome-dependent elements delivers budget predictability for the client while preserving margin for the provider.
Why does AI services work expose the problem more than traditional consulting?
AI automation and implementation work tends to generate compounding, recurring value for clients. Once a system is running, the savings arrive year after year with little additional input from you. Bessemer Venture Partners documents this with concrete examples: Intercom prices its AI support agent at $0.99 per automated resolution, and legal AI vendor EvenUp prices per AI-generated demand package, not per hour of work done.
McKinsey’s research on AI adoption reports early deployments producing 20 to 50 per cent reductions in call-centre volumes through automation in some sectors. AI-enabled personalisation has driven 5 to 15 per cent revenue uplift in marketing use cases. These are the kinds of gains that, once baselined, create room for a performance component tied to verified improvements.
Day-rate billing treats the first week of the engagement and the last week the same way. A client’s finance director can usually see the asymmetry clearly once you name it: you delivered the capability, they run the savings, and the fee structure should reflect that split more honestly.
What do value-based proposals actually look like in practice?
Three structures work directly for UK AI services firms, and each fits into a standard contract. The critical preparation is a structured value discovery session during scoping: establish baseline KPIs, agree on the measurement method, and document the baseline before any delivery work begins. Monetizely’s research on outcome-based AI pricing identifies this baselining step as the single point where value-based contracts succeed or fail.
Customer service automation: hybrid cost-savings model
A fixed base for discovery and delivery (for example, £15,000 and £25,000 respectively), followed by an outcome fee calculated as a percentage of documented reductions in fully-loaded support costs, measured quarterly. A guardrail clause removes the outcome fee if customer satisfaction scores drop below the agreed baseline. This reflects Intercom’s per-resolution logic translated into a consultancy context. Monetizely’s framework describes this as the Efficiency Tier: pricing anchored to cost savings rather than hours billed.
Back-office workflow automation: per-workflow fees
Define the unit of work clearly, a fully processed invoice with no manual intervention, for instance, and price per unit with tiered discounts above volume thresholds. An accuracy-linked SLA reduces the fee if quality targets are missed. Compliance AI firm Graph AI uses a per-case model for AI compliance processing, as documented in Bessemer’s analysis. The same structure applies to finance-team automation in professional services.
Revenue-side AI: performance fee
Lead qualification or personalised outreach can carry a performance fee tied to attributed new revenue. The attribution method must be agreed up front, typically CRM tagging by the sales team, and a performance cap limits client risk to a manageable multiple of the fixed fees. Stripe notes that clients accept hybrid structures because they reconcile budget predictability with visible value links.
When does value-based pricing not apply?
Value-based pricing needs two things to hold up: reliable baseline data on the metric you plan to improve, and a defensible method for attributing results to your work rather than to other variables in the business. If a client cannot tell you their current ticket volume, invoice throughput, or pipeline conversion rate before you scope the project, the baseline condition is already broken.
Three further situations make the model impractical. First, early-stage or experimental deployments where outcome variance is high. If neither you nor the client can estimate what the impact will be, tying fees to a moving target creates delivery risk that outweighs the margin upside. Second, revenue-share agreements suffer from attribution disputes. If pipeline conversion improves, the client’s CRM may credit the sales team rather than the AI workflow you built. Agreeing the attribution method before signing is essential; if the client will not commit to a method, default to a fixed fee. Third, some larger enterprises and public-sector procurement frameworks still mandate time-and-materials billing, and outcome fees will not fit the standard paperwork.
What makes value-based pricing legally and commercially defensible in the UK?
UK regulatory requirements for AI work carry direct cost implications that must appear as line items in your proposals. The ICO’s AI and data protection guidance requires Data Protection Impact Assessments, transparency notice updates, and human-in-the-loop provisions for automated decision-making. NCSC guidance adds threat modelling and access controls. FCA expectations cover governance, operational resilience, and model risk for any client in regulated financial services.
These are not optional line items you can absorb into a flat daily rate. The ICO fined Clearview AI £7.5 million in 2022 for unlawful biometric data collection. The Experian enforcement action in 2020 highlighted failures around automated profiling transparency. A client who deploys AI without adequate protections carries that exposure. When your proposal prices the governance work explicitly, you are reducing both parties’ risk, and that is a commercial argument as well as a compliance one.
If clients trade in the EU or handle EU residents’ data, the EU AI Act adds risk management, technical documentation, and human oversight requirements for high-risk applications in recruitment, credit, and essential services. UK firms serving EU clients should treat Act compliance as a distinct line item, not a bracketed note in the project scope.
The CMA’s principle of transparent, fair AI dealing provides a final commercial reason: clients who understand what they are paying for, and what protections are built into the engagement, are more likely to renew than those who see only a day rate on a statement of work.
Value-based pricing in AI services requires one preparatory shift before anything else: a structured conversation in discovery about what success looks like in pounds, percentages, or volume. If you can measure it and attribute it to the work, you can price it. If you can do both, you have the commercial foundation to stop letting the clock set your ceiling.
