How to Run a Consulting Engagement, Step by Step: Scoping to Close

What is a consulting engagement, and what does the client really pay for?

A consulting engagement is a defined piece of work in which an outside adviser helps a client understand a problem and decide what to do about it. Some engagements are diagnostics that end in a recommendation. Others continue into implementation. Either way, the client is not buying your hours or your slides. They are paying to make a better decision than they would have made alone, and to start acting on it.

That framing matters because it tells you what to leave out. Data collected just in case, reports that describe your method at length and presentations where findings land as surprises all feel productive and add little. Everything in a well-run engagement serves one key question and one decision.

This is not only work for large firms. Plenty of independent consultants, trainers and coaches now diagnose and advise. The International Coaching Federation's 2025 Global Coaching Study, conducted with PwC, reported that 57% of coaches also offer consulting services. The steps below work at any size.

What are the steps of a consulting engagement?

Most practitioners describe a cycle that runs from entry to exit. Peter Block's Flawless Consulting describes phases from entry and contracting through discovery, feedback and the decision to act, to implementation and then extension or termination. Edgar Schein's work on process consultation adds the idea that the client owns the problem and the helper's first job is to understand it. A practical version for small engagements has seven phases.

How do you scope a consulting engagement?

Ask for an hour with the person who will own the decision before you propose anything. Write their request in their own words, then ask what has happened that makes it matter now, what has been tried, what decision the work will support and what success would look like six months after you finish. Look for at least one number and one date.

Then draft a key question in one sentence: what should [who] do about [what] by [when], given [constraint]. When the request names a solution, such as a new system or a restructure, the key question usually sits one level above it. In the example used in our kit, a manufacturer asked for help choosing planning software; the agreed question became how to bring on-time delivery from 81% back above 95% in six months without adding headcount. Software stayed on the list as one possible answer.

Finish with a readiness check. Does the sponsor own the decision? Will people and data be available within two weeks? Is the client open to a finding they did not expect? Is there any conflict of interest? It is much easier to reshape or decline an engagement before it starts.

What should a consulting engagement letter include?

An engagement letter is the written version of the scoping decisions. Keep it short and plain enough that the client reads it. Draft the substance in your own words, then have a qualified lawyer formalise it, because liability, tax, governing law and data protection rules differ by country.

Two items prevent most disputes: an out-of-scope list with a rule that anything not listed is out of scope unless changed in writing, and a delay rule that moves your dates if access or data arrives late. Ownership of work deserves explicit words too. In the UK, government guidance explains that the creator of a commissioned work is generally the first owner of copyright unless there is a written agreement to the contrary, and the US Copyright Office explains that a contractor's work is a work made for hire only in limited categories with a signed written agreement. Agree who owns what.

• The key question and the decision owner
• Scope in and out, and the change rule
• Deliverables with format, date and sign-off
• What you need from the client, with dates and a delay rule
• Fees, expenses, payment milestones and tax treatment
• Confidentiality, data handling and what happens at the end
• Ownership of deliverables and of your own tools
• Independence, limits of your advice and how either side can end the engagement

How do issue trees and hypotheses speed up diagnosis?

Hypothesis-driven problem solving, closely associated with the large strategy firms, means writing down what you think might be causing the problem in a form evidence could disprove, then going to find that evidence. It replaces months of data collection with a focused plan.

Start with an issue tree. Break the key question into three to five branches that do not overlap and together cover it; consultants often call this MECE, for mutually exclusive and collectively exhaustive. Choose a logic that fits how the client's work flows, such as process steps or types of cause. Under each branch write a specific hypothesis. "Planning could be better" is not testable. "More than a third of orders are re-sequenced inside the three-day window" is.

For each hypothesis, note what you would see if it were true, what you would see if it were false and the cheapest analysis that tells the two apart. Test first the hypothesis most likely to change your plan. A hypothesis that is not supported is still a useful result: it tells the client where not to spend effort.

How should you run interviews and request data?

Prepare one interview guide per group, with the same opening and closing for everyone and each question linked to a hypothesis. Ask for recent specific events rather than opinions: "tell me about the last time an order went out late" produces evidence, while "why are deliveries late?" produces theories. Keep what was said separate from what you think it means, and type notes up within a day using codes rather than names.

Send the data request in the first week, with a reason for every item. Ask for the least data that answers the question, and ask for personal data to be removed or pseudonymised before it is shared. If you handle personal data for a client, data protection law usually applies to you too: the European Data Protection Board's guide for small businesses notes that the GDPR applies to sub-contractors processing personal data on behalf of an organisation. Agree your role and the handling rules in the letter.

Observation and benchmarks round out the picture. Time and count the work rather than judging people, and look first for the client's own best performance as a benchmark, because it is hard to argue that something cannot work here when it already does on one shift.

How do you turn findings into recommendations?

Log every finding with its sources, type of evidence and a confidence rating. A single opinion is a lead, not a finding. Hard data with a second independent source is strong enough for the body of the report. Then write the so-what for each finding: one sentence on what it means for the decision.

Group report-ready findings into three or four key messages that together answer the question. Then list real options, including doing nothing, the client's favourite and ideas the client team already has. Agree criteria and weights with the client before anyone scores, score each option so that 5 is always best, and look hard at the trade-offs of the top option. Every recommendation the client accepts needs an owner, dates and a measure with a baseline and a target.

How should a consulting report and decision meeting be structured?

Put the answer first. The executive summary states the question, the answer in one or two sentences, three key messages with their strongest evidence, recommendations with owners and the decisions needed. This structure owes much to Barbara Minto's pyramid principle: lead with the main point, group supporting ideas so they do not overlap and move detail to appendices. Make every heading a full-sentence message rather than a topic, and include the hypotheses you tested and did not find, so the client does not spend effort there.

Before the decision meeting, walk every influential stakeholder through the findings that affect them, one to one, and ask for their honest reaction. This pre-wiring often improves the recommendations and almost always reduces defensiveness. In the meeting, state the decisions needed on the first slide, give the answer on the second, discuss more than you present and finish by deciding each recommendation: yes, no or not yet, with an owner. When challenged, go back to the source, period and sample; if someone brings a new fact, check it rather than defending the finding.

How do you close a consulting engagement and use AI safely?

Close with a review conversation with the sponsor, a handover pack, records returned or deleted as agreed, access removed, a final invoice, internal lessons learned and check-in dates at 30, 60 and 90 days. Measure value with the client against the baselines you agreed, and ask in writing before using any case study, letting the client choose whether they are named.

AI assistants can draft interview guides, summarise pseudonymised notes, suggest analyses and tighten report wording. They should not decide the key question, rate evidence or choose recommendations. Check your engagement letter and the client's policies first, use approved tools for confidential material and check every fact yourself. Frameworks such as the US NIST AI Risk Management Framework, released in January 2023 with a generative AI profile in July 2024, are useful reading on managing these risks, and in the EU the AI Act's AI literacy duty for providers and deployers has applied since 2 February 2025.

If you want to see where your own skills stand, an AssessAll assessment such as Strategic Thinking and Business Judgment gives you a starting point, and a Jobulary individual development plan turns what you find into practice goals. Bodhih Training's Consulting Engagement Kit puts every step above into working files: a scoping sheet, an engagement letter discussion sheet, a 17-sheet engagement workbook, interview and hypothesis forms, report templates and an AI prompt pack.