An Independent Advisor or a Consultancy

A board that has decided to put AI into its operations has two kinds of help available, and they are usually compared on price when they should be compared on what they are for.

A consultancy sells capacity. People, method, a team that stays until the thing is built, and an organisation that carries delivery risk. An independent advisor sells a position: a view on what should happen, argued and defended, by somebody who will not be staffing the result.

This page states where each is the right purchase. It is written by an independent advisor, which is a conflict of interest, so the section on what an independent cannot do is longer than the section on what one can.

What a consultancy is for#

Work that needs hands. If the decision has been taken, the scope is known, and what remains is execution across several workstreams at once, capacity is the thing to buy and an individual cannot supply it.

Three cases where the firm is the right answer and an independent is not:

• Delivery at scale. A programme running in six business units needs people in six business units. An independent can shape it and cannot run it.
• Carried risk. A firm can take contractual responsibility for an outcome, carry professional indemnity proportionate to the engagement, and continue if one person is unavailable. An individual cannot do any of those three.
• Systems work. Integration, data engineering, model deployment and the surrounding technical build belong to people who do that for a living.

What an independent is for#

A decision that has not been taken, where the obstacle is that nobody in the room can say the unpopular thing.

The structural argument is about incentives rather than ability. A firm that implements has a commercial interest in the answer being implementation. That is not dishonesty; it is the shape of the business. When the right answer to "what should we do about AI in claims handling" is "nothing yet, your decision rights are undefined and you would be automating a process you cannot describe", a firm whose revenue comes from the build is being asked to argue itself out of work.

An independent has a different interest, and it should be named rather than implied. I am paid for advice, so I have an interest in you concluding that you need advice. What I do not have is an interest in which tool you buy, which vendor you select, or how large the programme becomes. Those are the three places where advice most often bends.

The question that tells you which one you need#

Can you write down, in one sentence, the decision this work is meant to implement?

If you can, buy capacity. If you cannot, buying capacity will take the decision for you by default, through the accumulated choices of whoever is doing the work. That outcome has a name in this research: drift. It is the most expensive way to decide anything, because the decision still gets made and nobody is accountable for it.

What to ask either one#

The same five questions work on a sole practitioner and on a global firm, and the answers separate them faster than a credentials list.

• What would you tell us not to do? An answer that contains no refusals is a sales document.
• What does this engagement produce that outlives you? Something written that a board can be held to, or nothing.
• Where does your revenue come from if we follow your advice? Both of us should be able to answer this without discomfort.
• Who carries the consequence if this is wrong, and how would we find out? The second half is the harder half.
• What have you been wrong about? A practitioner with no corrections has either not been doing it long or is not recording them.

What this page cannot tell you#

There is no comparative evidence on the outcomes of independent advice against firm advice. No controlled comparison exists, in this field or near it, and the engagements are too heterogeneous for one to be straightforward to design. Anybody offering you a figure on this is giving you a sales claim.

What is established, in an adjacent literature, is narrower and still useful. Dell'Acqua and colleagues found with Boston Consulting Group consultants that AI help raised performance inside its competence and lowered it outside, which is a finding about advice-takers rather than advisors, and it transfers as a caution: the quality of help depends on whether the problem is inside the helper's range, and neither party reliably knows where that edge is.

What would settle it#

A study tracking decisions rather than engagements: same class of organisation, same category of decision, advice from an independent in one arm and a firm in the other, outcomes assessed at two years by somebody who did not give the advice. Nobody has run it. Until somebody does, the argument on this page is an argument about incentive structures, and it should be read as one.

Where this sits in my own argument#

Most of this research is about judgement inside organisations, and who is accountable for it when a machine takes part. Buying advice is the same question pointed outward. An organisation that cannot say which decision a programme implements has the same defect whether it hires one person or forty, and the defect is not solved by either purchase. It is solved by somebody writing the decision down.

Related SuperSkills research#

On the decision this work is supposed to implement, decision quality and the delegation boundary map. On what happens when nobody writes it down, design versus drift. On the limits of oversight at scale, how much can one firm check. On what my own advisory work produces, the case studies, which state what they do not show.

Key sources

• Dell'Acqua, F. et al. (2023). Field experimental evidence on the jagged technological frontier.

About this research#

Written by Rahim Hirji, author of SuperSkills (Kogan Page, 2026), independent advisor to chief executives and boards, and founder of The SuperSkills Intelligence Company.

This page describes a category of service I sell. It is not an independent comparison, and the conflict is stated above rather than disclosed at the foot.

How this research works · Reviewed quarterly · Found an error? Tell me and it is corrected on the page.