AI consultancy proposals: how to compare costs

How to compare AI consultancy proposals

An AI consultancy proposal should state the problem, deliverables, acceptance criteria, client responsibilities and assumptions behind the price. This page explains how to compare proposals. For current UK service prices, use The AI Consultancy’s pricing page; for a broader delivery budget, use the UK AI implementation cost guide.

Compare like with like

Question	What to request
What is included?	Named deliverables, systems, users and environments
How is completion assessed?	Acceptance criteria, evaluation data and who signs off
What can change the fee?	Assumptions, exclusions and a written change-control process
What continues after launch?	Licences, model usage, hosting, maintenance, support and ownership
What must the client supply?	Data access, subject expertise, security review and staff time

Choose a commercial model that fits the uncertainty

A fixed fee can work when scope and acceptance are sufficiently clear. Time-based work needs a defined review point and spending limit. An outcome-linked arrangement needs an agreed baseline, attribution method and treatment of factors outside the supplier’s control. These are contracting choices, not evidence that any model guarantees a return.

Build the business case separately

List one-off and recurring costs over the same period as the proposed benefits. Distinguish capacity released from cash saved and label forecasts. Include data preparation, staff review, integration and handover. Do not use a generic first-year ROI multiple as a substitute for your own assumptions.

Use the business-case template to test the inputs. If priorities or data access are unclear, start with AI Readiness; if the workflow is defined, discuss AI Implementation.