The owner of a twenty-five person services firm has four AI tools partially deployed across the business. None of them is producing the output the demos promised. There is no internal seniority coordinating the work, and the team has stopped raising new ideas because the last three didn’t land. She is now comparing two consulting proposals: a £4,000 a month fractional CAIO retainer and a £15,000 fixed-fee project. The numbers are close enough that the choice feels like a cash flow question. It isn’t. The two proposals are doing fundamentally different jobs, and one of them will leave her exactly where she is now.
This is the most common mis-purchase pattern in the SME AI advisory market today. Three engagement formats coexist at roughly comparable price points. Each one is designed for a different starting position. Match them wrong and the engagement does not fail loudly, it fails quietly, with the consultant delivering exactly what was specified and the business in the same place it started.
Why the format matters more than the price tag
The three formats answer three different questions. A fractional CAIO answers “what should we do, and how do we govern it”. A project consultant answers “how do we build the thing we have already specified”. A roadmap consultant answers “where should we apply AI in the first place”. The price ranges overlap enough to make them feel comparable, and the surface deliverables sound similar enough that the distinction blurs. The distinction matters anyway.
A founder in the question-where stage who buys a project will get a built thing she did not need. A founder in the question-how stage who buys a fractional CAIO will get good strategy and no implementation. A founder in the question-govern stage who buys a roadmap will get a document that overlaps with what she has already worked out herself. The same money, three different misfires.
What a fractional CAIO actually does, and what it doesn’t
A fractional Chief AI Officer is a senior advisor working part-time, typically one to two days a week, across an initial 90-day engagement that often renews for six to twelve months. The cost lands at £2,000 to £8,000 a month, or £40,000 to £96,000 a year at full senior rates. The role is strategy and governance: defining the AI roadmap, prioritising use cases, selecting and overseeing technology partners, leading change management, and serving as the named accountability point for AI outcomes inside the business.
What a fractional CAIO does not do is implementation. They will not write code, train models, integrate systems, or run a pilot themselves. The research evidence on this is unambiguous. As the SME AI consulting market puts it, a fractional CAIO can design a sensible AI roadmap and govern the programme, but they cannot build it. A founder who contracts a fractional CAIO without engaging delivery partners separately will find herself with excellent strategy and no implemented solutions. That is a real, repeating failure mode.
Fractional CAIO is the right format when the business has multiple AI initiatives in motion or planned, lacks senior internal coordination, and needs governance discipline more than delivery muscle. It is the wrong format when the business has one well-defined use case and just needs it built.
When a fixed-fee project is the right format
Fixed-fee project engagements work best when the use case is clearly defined, the business is confident in the approach, and the primary need is execution. The market has settled into three rough tiers. Basic automation projects, like email response automation or appointment scheduling, run £5,000 to £8,000. Mid-range workflow optimisation, like sales pipeline automation or customer enquiry routing, runs £8,000 to £15,000. Complex custom work, like proprietary content generation or domain-specific prediction models, runs £15,000 to £25,000.
Project engagements perform poorly when the use case is poorly specified or the business is still exploring whether AI is appropriate. A consultant incentivised to deliver scope quickly will build what was asked for, even when the underlying business question has not been validated. The classic failure pattern is the founder who specifies “build a chatbot for our website”, gets a working chatbot in six weeks, and then discovers customers prefer human support channels. The consultant did the work. The business stayed still.
The format suits a business that has already done the strategic thinking, validated the use case with at least one stakeholder interview, and is now buying execution.
When to buy a roadmap before anything else
Roadmap-only engagements are designed to address the validation problem before anyone commits to delivery. A roadmap engagement at £5,000 to £15,000 runs over six to eight weeks and produces a strategic document, not running systems. The output should include business process interviews with named stakeholders, a data readiness assessment, a prioritised use case list with quantified business value, and a recommended sequence with budget tiers attached.
A roadmap-only engagement is the right format when the business is at an early stage of AI adoption, when the founder is sceptical and wants validation before committing, or when previous AI experiments have stalled and the question now is “what should we have done instead”. The output is a document the business uses to make subsequent buying decisions, including which delivery partner to engage. A roadmap that does not lead to subsequent implementation has succeeded as a roadmap and failed as a business initiative. That distinction matters when scoping the engagement.
The sequence that works at SME scale
For a business that has tried AI, stalled, and is now reaching for outside help, the sequence that works is roadmap, then fractional CAIO, then project. Roadmap first, at £8,000 to £15,000, validates use cases and assesses readiness. Then a 3 to 6 month fractional CAIO retainer at £3,000 to £6,000 a month establishes governance and oversees pilot selection. Then a project engagement at £12,000 to £25,000 delivers the highest-priority use case with a measurable outcome attached.
Total first-phase investment lands around £25,000, similar to a single mid-tier project, but the risk profile is substantially better. Each stage informs the next. Each stage has a deliverable. None of them is doing the wrong job.
The format follows the question. The question follows the readiness. Picking on monthly cash flow alone is the failure mode. The fix is the conversation that names which question is actually live in the business right now.
If that conversation is the one you need to have, book a conversation.
