When Should a Consultant Say No?

Imagine the assignment without knowing the client’s name. Sales performance varies across regions, management wants to understand the difference, the company has extensive customer information and the commercial organisation contains enough variation to compare territories, representatives and accounts. A consulting team could reasonably examine segmentation, targeting, incentives, salesforce effectiveness, customer behaviour and the opportunities available in different parts of the market.

None of those analytical techniques is unusual, and the work could be carried out with great technical competence. The difficulty appears when one fact is returned to the centre of the analysis: the product was OxyContin, a powerful opioid whose commercial performance could not sensibly be considered independently of prescribing behaviour, addiction, misuse and patient harm.

The methods therefore remained useful while the meaning of the recommendation changed. Increasing sales was still a measurable commercial objective, but a complete professional assessment had to consider the consequences attached to achieving it, including consequences that sat outside the narrow language of sales productivity.

Technical accuracy does not remove the need to judge the objective being optimised.

Consulting organisations invest heavily in teaching people how to structure a problem, test evidence, analyse data and develop recommendations that can be implemented. Those capabilities are fundamental to the profession, but they become insufficient when the engagement itself contains a wider risk that management has chosen to exclude from the framing.

The adviser then has to consider whether the original question is complete. A proposal may ask how to increase sales, reduce headcount, automate a decision, identify a more profitable customer group or accelerate a regulatory strategy, while the consequences of doing so extend beyond the metric that has been selected for improvement.

This is where professional judgement earns its place. The consultant is being paid partly because somebody outside the operating hierarchy should be able to see a problem with enough distance to challenge the assumptions around it, including the assumptions that made the engagement attractive in the first place.

One question cuts through much of the complexity: if the recommendation succeeds exactly as intended, is the consultancy comfortable with the outcome it has helped create? The value of the question is that it shifts attention from the mechanics of delivery to the purpose of the engagement, which is often where the more difficult judgement sits.

Suppose the client gains market share, changes customer behaviour and implements the recommendation almost perfectly. Those results may look excellent in a project close-out presentation, yet the board of the consulting firm should still be able to explain why it wanted that outcome to occur and why the engagement was consistent with the firm’s professional responsibilities.

Client acceptance processes usually consider capability, commercial attractiveness, available people, likely margin and the potential value of a longer relationship. I would add the success test before the proposal is signed because it forces the consultancy to examine the result before commercial momentum makes that examination more difficult.

Imagine an engagement that produces €2 million in fees with an attractive margin, uses the team efficiently and creates a credible path towards further projects. Most consulting dashboards would record that relationship positively, particularly if invoices are paid on time and senior executives continue asking for more work.

Now include a low-probability but very large regulatory or reputational consequence, together with the possibility that future clients begin questioning the judgement behind the firm’s work. The expected economics change because consulting firms sell something that is difficult to isolate in the accounts: confidence that their advice combines technical competence with independent professional judgement.

Reputation therefore behaves differently in professional services from a conventional marketing asset. Once clients start wondering whether commercial incentives can override judgement, the concern reaches into the product they are buying rather than remaining at the level of communications or brand management.

Good consultants spend enough time inside a client organisation to understand why people behave as they do. They learn the company’s history, internal constraints, personalities, competitors and commercial pressures, and that knowledge is one of the reasons experienced advisers can often see possibilities that remain hidden in a more superficial analysis.

Long relationships also change the adviser. Language that sounded unusual during the first month becomes normal after the sixth, internal frustrations acquire context and the objectives discussed repeatedly in steering meetings begin to feel like the obvious problems the consulting team has been hired to solve. Familiarity can therefore improve the work while making independent challenge more difficult.

The governance response is not distance for its own sake. It is to make sure that somebody with sufficient authority remains capable of examining the engagement from outside the commercial relationship and asking why the firm is helping the client pursue this particular objective.

Consulting partnerships contain a structural tension because the professionals with the deepest knowledge of a major client are often the same people responsible for developing the account. Their experience is valuable when the firm considers new work, yet the commercial success of the relationship may also affect their revenue responsibility, professional standing and future opportunities.

That combination does not make their judgement unreliable, although it does make independent review sensible. A partner advocating for another €5 million engagement should contribute strongly to the acceptance decision, while the firm should also ensure that somebody who is not commercially dependent on the account can challenge the purpose, scope and risk of the work.

The need becomes more important when an individual client represents a substantial proportion of revenue because financial dependency can change the tone of professional challenge long before anybody consciously decides to compromise a standard.

The Purdue controversy was not the only period in which McKinsey’s client and engagement governance came under serious scrutiny. In South Africa, the firm faced investigations connected with work for state-owned enterprises and, in December 2024, McKinsey Africa agreed to pay more than US$120 million to resolve a US Department of Justice investigation into a bribery scheme involving government officials and consulting contracts.

Years earlier, McKinsey had also committed to return fees from projects involving Transnet and South African Airways where Regiments Capital had been involved, with the firm describing weaknesses in previous safeguards and outlining reforms to its global risk and compliance procedures. By May 2021, the settlement with Transnet concerning those fees, including interest, totalled R870 million.

I would be careful about treating different cases as though they were evidence of one identical failure because the facts, conduct and legal questions were different. For consulting boards, however, the recurring governance issue is difficult to ignore: decisions about whom the firm serves, what work it accepts, which intermediaries or partners it relies on and who can independently challenge a valuable relationship belong inside the commercial model rather than at its edge.

The lesson becomes more relevant as consulting firms help companies introduce generative AI, workforce automation, customer profiling, automated pricing, surveillance systems and algorithmic decision-making. Many of these assignments begin with a commercially reasonable question, such as how much labour cost can be removed, which decisions can be automated or which customer groups can be targeted more precisely.

The calculation may be technically straightforward while the advisory problem is broader. Management also needs to understand which capabilities disappear with the jobs being removed, how customers experience automated decisions, what data is being used, whether the model introduces discriminatory outcomes and who remains responsible when the system causes damage that nobody intended.

As analytical work becomes easier to automate, I suspect this part of consulting becomes more important rather than less. The value of an adviser will increasingly depend on the ability to interpret consequences, challenge framing and recognise when an apparently efficient solution creates a different problem elsewhere in the organisation.

It would be convenient to see this as a governance problem belonging mainly to enormous consulting firms, yet smaller advisory companies may face a more immediate version because one client can represent 20, 30 or even 40 per cent of annual revenue. Losing that relationship can affect hiring, investment, cash flow and the year’s profit target almost immediately.

That dependency changes the context in which professional judgement is exercised. An uncomfortable request may be accepted because the relationship is important, followed by an extension that looks close enough to the original scope to approve, and then by further work that would probably have received much more scrutiny if a new client had proposed it at the beginning of the relationship.

Governance is most useful when the difficult decision has been considered before the money is at risk. Smaller firms do not need elaborate compliance departments, but they do need clarity about which engagements require independent review, what types of work they will refuse and who has enough authority to walk away from revenue when the commercial pressure is strongest.

The easiest way to tell this story is to describe the settlements and conclude that stronger compliance was required. The more useful question is how an organisation filled with highly capable people can become so focused on solving a client’s commercial problem that the legitimacy of the objective receives less scrutiny than the quality of the analysis.

Consulting requires proximity because advisers need to understand the organisations they serve, but the profession also depends on retaining enough independence to challenge the client when its assumptions, incentives or objectives deserve examination. That balance becomes harder as relationships become larger and more profitable, which is precisely why it cannot be left entirely to individual judgement at the point when a new engagement is being sold.

McKinsey’s subsequent investment in stronger risk management, client selection and compliance demonstrates how seriously those systems can matter once a firm has experienced the cost of getting them wrong. For other consultancies, the opportunity is to consider the same problem earlier and at a scale appropriate to their own business.

A consultancy is shaped not only by the advice it gives, but by the work it decides is worth accepting.

For CEOs and boards of advisory firms, that makes client selection part of strategy. Revenue still matters, as it should, although the more difficult task is understanding which revenue strengthens the business and which can quietly weaken the judgement on which the business depends.