The Pyramid Under Pressure: Why AI Should Push UK Consultancies Beyond the Billable Hour

The billable hour has always been a convenient fiction. It pretends that value is created in direct proportion to time spent, that a junior analyst's hour and a partner's hour are comparable units and that clients are buying duration rather than judgement. For decades, UK management consultancies have made that fiction work through the pyramid model: a small number of senior people win and oversee the work, while a larger number of junior and mid-level people produce the analysis, drafts and deliverables that make the advice tangible.

In 2026, that pyramid is under strain. AI tools can now draft a market summary, clean a dataset, build a first-pass model, summarise interviews and produce a competent slide deck in a fraction of the time. The work has not disappeared, but the labour required to produce it has fallen sharply. The squeeze is not mainly at the top, where board access and trusted judgement still command a premium, or at the bottom, where graduate intake can be flexed. It is in the middle: the experienced consultant who used to add value by managing a small team of producers while also doing a great deal of production themselves.

The middle is where the model breaks

The traditional consultancy economics depended on leverage. A partner sold a project, a manager designed the approach and a team of analysts and consultants delivered it. Each layer was billable, and each layer learned by doing. The model was not just a profit engine; it was a training system. Junior people learned through repetition, supervised mistakes and exposure to client rooms.

AI disrupts both sides of that equation. If a mid-level consultant can use AI to do the work of two juniors, utilisation falls unless the consultancy wins more work or changes what it sells. If clients know that a first draft can be produced in an afternoon, they begin to question the hours on the invoice. The result is a quiet repricing of routine analytical work, even where the billable hour remains on the contract.

UK clients are not waiting for consultancies to resolve this philosophically. Procurement teams in central government, NHS bodies, local authorities and large private sector organisations are already asking suppliers to demonstrate productivity gains from AI. Some are demanding that efficiency savings be passed on. Others are moving towards fixed-fee packages, outcome-based fees or retained advisory subscriptions. The direction of travel is clear: clients want to buy a result, a capability or a period of access, not an open-ended allocation of junior time.

Why the hour is the wrong unit

The billable hour is not wrong because time is irrelevant. It is wrong because it measures input rather than output, and it becomes actively misleading when AI changes the ratio between the two. A consultancy that bills by the hour has an incentive to preserve labour-intensive processes, even when a faster method would serve the client better. That is not a sustainable position in a market where clients can see the tools for themselves.

There is also a talent problem. If firms respond to AI by cutting junior and mid-level roles, they may protect short-term margins while hollowing out the pipeline of future managers. The people who will lead client relationships in 2030 need to learn how to frame problems, interrogate AI output, handle ambiguity and manage stakeholders. Those skills are not acquired by watching a partner present. They are acquired through supervised practice.

Protecting the pipeline, not just the margin

UK consultancies therefore face a choice. They can use AI to sell the same work more cheaply, entering a race to the bottom against freelancers and software. Or they can use AI to change the shape of the work, moving from selling production capacity to selling judgement, assurance and implementation. The second route is harder, but it is more defensible.

Managers can take practical steps. First, audit where time is actually spent. Most firms have a rough sense that junior staff produce decks and models, but they rarely separate low-judgement production from high-judgement analysis. A simple activity review will show which tasks AI can absorb and which still require human oversight.

Second, reprice by deliverable, not by headcount. A fixed-fee diagnostic, a retained advisory service or a milestone-based implementation fee can align price with value while giving the client cost certainty. The billable hour can remain as an internal costing tool, but it should not be the client-facing unit for every engagement.

Third, be transparent about AI use. Clients are increasingly sophisticated about what these tools can and cannot do. A consultancy that explains how it uses AI, where human review sits and what quality controls are in place is more credible than one that quietly reduces hours and hopes nobody notices.

Fourth, redesign the talent pathway. If juniors are doing less production, they need more structured exposure to client problems, quality review, stakeholder management and commercial judgement. Apprenticeship routes, internal rotations and supervised client-facing work can replace some of the learning that used to happen through repetitive delivery. The apprenticeship levy can support this, but only if the training is genuinely integrated into project work rather than treated as a compliance exercise.

A practical reset for managers

For managers inside UK consultancies, the immediate priorities are unglamorous. Review utilisation targets. They were built for a pre-AI delivery model and may now encourage the wrong behaviour. Review project staffing. A team of five may no longer be the right shape for a project that once required six. Review quality assurance. Faster output increases the risk of plausible but wrong analysis, so review needs to be sharper, not lighter.

Most importantly, review what the firm is selling. Clients still pay for clarity, accountability and the confidence that someone experienced has looked at the problem properly. Those things are not commoditised by AI. They are made more valuable when the surrounding production work becomes cheaper.

MCon UK readers will recognise the tension. The billable hour is familiar, auditable and easy to explain to finance. But familiarity is not the same as fitness. In 2026, the consultancies that thrive will be those that use AI to improve the quality of judgement rather than simply to reduce the number of hours. The ones that cling to the old pyramid may find that the middle disappears first, and that the top has nobody left to stand on.