A consulting or in-company project asks you to solve a defined problem for a real organisation and deliver recommendations. It is assessed on the rigour of the reasoning behind the recommendation, not on whether the client adopts it — which is the single most common misunderstanding, and the one that determines how you should write.

The difficulty is not the analysis. It is that you are serving two readers with incompatible preferences: a client who wants a short answer, and a marker who wants to see the working. This guide covers how to scope the brief, run the project, and produce something that satisfies both.

1. Understand what is actually being assessed

Your client will judge the project on whether the recommendation is useful and cheap to implement. Your university will judge it on whether the recommendation is defensible — whether you gathered appropriate evidence, considered alternatives, and can justify the method that got you from one to the other.

These pull in different directions. A recommendation the client loves, arrived at by asking three people and trusting your instinct, fails academically. A meticulously evidenced recommendation the client cannot afford still passes.

Practical consequence: when the two conflict, write for the marker and manage the client. A project that is adopted enthusiastically and marked poorly is the common failure mode on these modules, and it is entirely avoidable. If your programme offers a dissertation as an alternative route, our guide to writing a business management dissertation sets out what that option asks for instead — it is a genuinely different trade, not an easier one.

2. Scope the brief before you agree to it

Organisations describe problems at the wrong altitude. “Our staff turnover is too high” is a symptom, a strategy question and a payroll question at once, and it is not deliverable in a single student project.

Before you commit, agree three things in writing with the client, even if only by email:

1. The question. One sentence, answerable, with a defined boundary. “Which factors most strongly predict turnover among first-year warehouse staff at the Leeds site” beats “why do people leave”.
2. What is explicitly out of scope. Name it. This is the clause you will point at in week eight when someone asks you to also look at the pay structure.
3. What access you are being given. Which data, which people, by when. A project that depends on interviews nobody has agreed to is a project with no methodology.

Then check the scoped question against your assessment criteria before you confirm it. If the brief cannot generate the analysis your module requires, renegotiate now — it does not get easier later. Our guide to writing a statement of the problem covers turning a vague concern into a researchable question, and the structure carries over almost exactly.

3. Treat the proposal as a contract

Most programmes require a short proposal before fieldwork. Use it properly: it is the document that protects you when the project changes underneath you, which it will.

It should state the question, the method, the access you have been promised, the timeline, and — critically — what happens if access falls through. A named fallback is not pessimism; it is the difference between a delayed project and a failed one. Our research proposal template and guide covers the academic components, and there is a broader step-by-step research proposal walkthrough on our sister site.

4. Choose a method you can actually complete

The constraint on these projects is access, not ambition. A survey of 400 employees is a better design than eight interviews and a worse plan if the client will not distribute it.

Work backwards from what you have been granted. Common workable designs:

That last row matters. A project with almost no internal access is still viable as a rigorous options appraisal built on published evidence — it is not a failed project, it is a different one. Say so explicitly in the methodology rather than apologising for it throughout.

5. Manage the client relationship

Three failure patterns account for most difficulties, and all three are predictable.

The client goes quiet. Sponsors get busy or leave. Agree a single named contact and a fallback contact at the start, keep a short written record of every decision, and escalate to your supervisor early rather than heroically absorbing delay.

The client wants a different project. Mid-project, priorities shift and you are asked to look at something else. You can often accommodate it — but only as an additional section, never as a replacement for the scoped question your assessment depends on.

The client wants a particular answer. This is the one with integrity implications. If a sponsor makes clear which conclusion would be welcome, your obligation is to the evidence. Where the evidence does not support their preferred option, say so and show why; a well-evidenced negative finding is a strong academic result even when it is an unwelcome commercial one.

6. Handle confidentiality before it becomes a problem

Most client projects involve some commercially sensitive material. Agree at the outset who will read the report — your supervisor, a second marker and possibly an external examiner — and confirm the client is content with that readership.

Where sensitivity is genuine, the usual solutions are to anonymise the organisation, to aggregate figures so individual performance cannot be inferred, or to place sensitive detail in a confidential appendix that is submitted but not circulated. Agree which of these applies before you write, not when you are formatting the final document.

7. Write two documents, not one

This is the structural insight that resolves the dual-audience problem. Produce a short client-facing deliverable and a full academic report, sharing evidence but not shape.

Check your handbook for whether the client deliverable is itself assessed — on many programmes it is, precisely because writing for a non-academic audience is a skill being tested. If so, the register shift is part of the mark rather than an afterthought.

8. Make the recommendations defensible

Recommendations are where these projects are won and lost. A weak set asserts what should happen; a strong set shows the reasoning and the alternatives.

For each recommendation, make four things visible: the evidence it rests on, the alternatives you considered and rejected, what implementation would require in cost and time, and how the organisation would know whether it worked. That fourth element is the one most students omit, and it is often what separates a good project from an excellent one.

Be proportionate. Three well-evidenced recommendations beat eleven plausible ones, and a recommendation your evidence cannot support should be labelled as a suggestion for further investigation rather than dressed up as a finding.

9. Common mistakes

10. If the project collapses

Sometimes the client withdraws, restructures, or simply stops responding. This is common enough that programmes have processes for it, and the worst response is to keep waiting quietly.

Tell your supervisor immediately. The usual salvage routes are to convert to a desk-based study using published sector data, to reframe the project around the access you did get, or to write up the engagement itself — including the access failure — as a documented finding. All three are viable; none of them works if you raise it in the final fortnight.

Frequently asked questions

Is a consulting project easier than a dissertation?

No, and it fails differently. It trades some research depth for real constraints — a client, external deadlines and access you do not control. The marking criteria differ rather than relax.

Am I marked on whether the client implements my recommendation?

Generally not. Assessment focuses on the rigour of the reasoning behind the recommendation. Adoption depends on budgets and politics you have no control over.

Do I still need a literature review?

Almost always. An applied project still needs a theoretical frame to justify the method and interpret findings. Check your handbook for the expected weighting, which is usually lighter than in a dissertation but rarely zero.

What if my client will not let me publish the data?

Agree the readership and confidentiality arrangements before you write. Anonymising the organisation, aggregating figures, or placing sensitive material in a confidential appendix are all normally acceptable to universities.

What happens if my client pulls out?

Tell your supervisor immediately. Projects are commonly salvaged by converting to a desk-based study, reframing around the access you obtained, or documenting the engagement failure as a finding.

Should the client report and the academic report be the same document?

No. They open differently, weight method and literature differently and are judged on different things. Produce both, sharing evidence but not structure, and check whether the client deliverable is separately assessed.

How many recommendations should I make?

Fewer than you are tempted to. Three recommendations with evidence, alternatives, implementation requirements and success measures are worth more than a longer list of assertions.