The 5 Phases of a Consulting Engagement, Explained

Most descriptions of "the consulting process" stop at the deliverable: gather data, build the deck, present the findings. That covers two of the five phases. An engagement doesn't end when the client accepts a recommendation. It ends when the recommendation is running, the client can maintain it without the consultant in the room, and both sides have formally closed the relationship. A proposal or statement of work that ends at "deliver the recommendations" is missing a phase of structure.

This guide walks the five phases in sequence, with the sub-activities, checklists, and completion criteria each one requires, written for the person running the engagement, not the client receiving it.

The five phases at a glance

Phase	Purpose	Key deliverable	Where it commonly breaks
1. Entry	Learn enough about each other to agree on scope and approach.	Signed contract or letter of agreement, with terms of reference.	The preliminary diagnosis quietly becomes a full diagnostic project before a contract exists.
2. Diagnosis	Examine the problem and its causes in depth.	A confirmed problem statement with evidence-tested causes.	Symptoms get treated as the problem; findings are saved for one big reveal instead of shared as they emerge.
3. Action Planning	Develop, evaluate, and get client sign-off on a solution.	A client-selected recommendation with an implementation plan attached.	The recommendation describes the end state and leaves out the path to it.
4. Implementation	Turn the recommendation into operating practice.	The new practice running under normal conditions, monitored and owned by the client.	Scoped out by default instead of by an explicit, budgeted decision.
5. Termination	Close the job and the relationship.	Final report, settled commitments, an agreed withdrawal.	A fade-out instead of a decision; withdrawal never discussed until it's already overdue.

Phase 1: Entry

Entry ends when a contract, or its practical equivalent, is signed. Kubr's model breaks it into five sub-activities: first contacts with the client, a preliminary problem diagnosis, assignment planning, a proposal to the client, and the consulting contract itself.

Keep the preliminary diagnosis short

This is not the in-depth diagnosis that follows in Phase 2. It's deliberately brief: one to four days for most assignments, five to ten for complex, multi-area ones. Anything needing an extensive diagnostic survey (a turnaround, a merger, a major reorganization) isn't a preliminary diagnosis; it's a separately scoped project. The purpose is orientation, not solutioning: move from the general to the particular, from overall objectives down to specific reasons for substandard performance, not the reverse. Working function-by-function first, hoping the picture assembles at the end, tends to misdirect the effort.

What belongs in the terms of reference

Whether the client drafts it, the consultant does, or it gets built jointly, the terms of reference is the document both sides point to when the engagement's scope gets tested later. A complete one covers:

• The problem(s) to be solved, stated plainly.
• Objectives and expected results: what "done" looks like.
• Background and supporting information the consultant needs to start.
• A budget estimate or resource limit.
• A timetable: start and completion dates, key stages, control dates.
• Interim and final reporting: dates, form, and who receives them.
• Inputs the client must provide: data, staff time, administrative support.
• Exclusions: what will not be the object of the assignment. This is the working anchor against scope creep, not a one-time disclaimer.
• Constraints likely to affect the project.
• The profile and competence required of the consultant or team.
• Contact persons and addresses.

A deeper walkthrough of drafting each of these, plus the proposal structure that responds to them, is in the terms of reference guide.

Choosing a contracting form

Kubr sets out three forms, in ascending formality: a verbal agreement for repeat business or small, trusted assignments; a letter of agreement, the prevailing form in most countries; and a written contract, required by law for most public-sector work and advisable for anything large, complex, or cross-jurisdiction. Whatever the form, the same ground needs covering: parties, scope, work products, inputs, fees and billing, confidentiality and conflict of interest, copyright, liability limits, subcontractors, and when either party may invoke termination (the item most often left implicit). The full checklist, and where a statement of work differs from these lighter forms, is in the compliance consulting SOW guide.

Completion criteria: a contract or its equivalent is signed, the terms of reference are agreed (explicit or implicit), and the assignment strategy (roles, mode, pace, resourcing by phase) is set.

Phase 2: Diagnosis

Kubr calls Diagnosis "the first fully operational phase." The work is to examine the problem and its causes in depth and build the information base the next phase's solution work depends on. Diagnosis does not include working out solutions (that's Phase 3), though in practice diagnostic interviews routinely surface embryonic ideas, and those should be captured, not suppressed, while the diagnostic work itself stays disciplined.

Five dimensions to fix before collecting a single fact

Dimension	The question it answers
Substance / identity	What is the problem, precisely, and against what standard is "poor" or "low" being measured?
Location	Which units, which locations, how widespread, and does it touch customers or suppliers outside the organization?
Ownership	Who is affected and wants it solved; who is likely to resist a fix?
Magnitude	How large in absolute terms (money, time, capacity) and relative terms (against other problems, against total turnover)?
Time perspective	Since when, how frequent, trending which direction, and is this a future problem being anticipated rather than a current one?

Kubr, drawing on Nadler and Hibino, also recommends opening with purpose rather than fault-finding: not "what's wrong here," but "what are we trying to accomplish." Narrowing an array of possible purposes to a single focus purpose (one that survives the filter of management's desires, financial benefit, cost, time limits, and stakeholder support) keeps the diagnosis pointed at something worth solving, not just something broken.

Five ways diagnosis goes wrong

• Mistaking symptoms for problems. Falling sales or rising absenteeism are usually symptoms, not root problems.
• Starting with a preconceived cause. "We already know why" shortcuts the fact-finding that would test it.
• A single technical lens. An engineer sees an engineering problem; a systems person sees a systems problem. Neither alone is diagnosis.
• Only top management's framing. Ignoring how the problem is perceived elsewhere in the organization misses half the picture.
• An unclear focus purpose. Working productively on the wrong thing is still working on the wrong thing.

A gap analysis or a risk assessment engagement is largely a Diagnosis-phase exercise conducted against a defined framework; see the AML program gap analysis guide and the BSA/AML risk assessment guide for what that diagnostic work covers in a compliance-specific context.

Completion criteria: the problem is restated and reconfirmed with the client rather than assumed from the contract: genuine disagreement over the original framing surfaces more often than proposals admit. Causes are identified and evidence-tested, not merely hypothesized. The client feedback loop stays live throughout, rather than saved for one end-of-phase reveal; silence tends to breed rumor and resistance, not patience.

Phase 3: Action Planning

Action Planning finds, evaluates, and gets client sign-off on a solution, plus the plan for implementing it. The sub-activities: developing solutions, evaluating alternatives, presenting proposals to the client, and planning for implementation.

Getting from many ideas to one

Kubr describes creative problem-solving as five stages, two analytical and three requiring suspended judgment: preparation (restate the problem several ways), effort (generate volume without filtering), incubation (step away and let it sit), insight (the moment it clicks), and evaluation: only now, filter. The barriers worth naming out loud in a session are as much cultural as technical: respect for a senior person's framing is hard to challenge even when invited to, and a team's own past success is one of the more common reasons it stops questioning what made it successful.

Presenting the recommendation honestly

Kubr is specific about what "honest" means when a recommendation goes to the client. Four things have to be on the table:

• The risks involved: an untested solution, likely resistance, realistic cost-overrun potential.
• What the client must sustain: the discipline, transfers, or changes required of senior staff to make the solution work.
• What could not be completed: scope gaps and data limits, named rather than glossed over.
• Future perspectives: whether the solution anticipates growth or will need revisiting soon.

A recommendation that describes only the end state, with no path to it, is incomplete by this standard: "an effective action proposal shows not only what to implement but also how to do it." An independent-testing engagement is a useful contrast: it has its own required scope and cadence layered on top of this general shape, covered in the independent testing guide. Fee discussions belong here too, in the proposal's financial-terms section; the mechanics of setting them are in the consulting pricing models guide.

The decision belongs to the client

It is the client's decision, never an outcome the consultant imposes. A client who feels a solution was pushed on them tends to disengage during implementation and assign blame when results disappoint. This is the pivot point of the engagement: everything downstream depends on the client owning the choice, not merely not objecting to it.

Completion criteria: at least one evaluated, client-selected solution; an implementation plan, even a high-level one, included in the proposal itself; an explicit client decision on record, not inferred from silence.

Phase 4: Implementation

Kubr calls Implementation the "acid test" for everything developed in Diagnosis and Action Planning: turning the proposal into operating reality, monitoring it, correcting it as false assumptions surface, because they will, and building the client's own capability to run it without the consultant.

Whether the consultant stays involved at all is a real decision, not a default. It's reasonable to skip when the problem is straightforward and the client showed real capability during Diagnosis. Where staying involved makes sense, the arrangements scale down from a full team: shrink the on-site presence progressively, keep one consultant through implementation while pulling in specialists as needed, handle only the hardest tasks and leave the rest to the client, visit at agreed checkpoints, or stay on-call. Budget pressure is a design problem, not an automatic reason to drop implementation support; a more economical assignment design can free the resources to stay involved instead of abandoning the client at the recommendation.

This is also the phase most engagements never reach: by Kubr's own estimate, probably not more than 30 to 50 percent of assignments include implementation at all. Usually that reflects client preference or budget, not consultant negligence, but it should be an explicit line in the proposal, decided during Entry, not something that quietly falls off the scope once the deck is delivered.

A new way of working holds up better when the approved method is taught from the outset, rather than letting people develop ad hoc habits that later need unlearning; practice happens in short, spaced sessions rather than one long push; goals are demanding but realistic and trackable; and evidence of improvement gets recorded and fed back regularly. After the new practice goes live, backsliding is the main risk: retiring the old forms and tools, not just deprecating them, and auditing the new practice periodically, the way an annual books audit runs, are what keep it from quietly reverting.

Completion criteria: the new practice operates under normal, not consultant-supervised, conditions; monitoring is owned by the client; training is complete for the roles that need it; and safeguards against backsliding are structurally in place, not just verbally agreed. For many compliance engagements, what's operating at the end of this phase is a documented compliance management system the client now owns.

Phase 5: Termination

Two things terminate here, not one: the job (completed, discontinued, or continuing without the consultant; pick one explicitly rather than leaving it ambiguous) and the relationship, which determines whether repeat business is even on the table. A consultant convinced the job succeeded while the client is simply waiting for them to leave is the exact failure this phase exists to prevent.

Timing runs in both directions

Terminating too early leaves work incomplete, often because the client overestimated their own readiness or the budget ran out first. Terminating too late usually means a technically difficult project without enough client training to take over, a job scope that kept quietly expanding, or, stated plainly, the consultant staying because the revenue is convenient. The discipline that prevents both: discuss the withdrawal point at the start, and revisit it at every phase transition. Watch for withdrawal signals, like a client becoming less available, and take them seriously even when there's a professional case for wanting more time.

What evaluation actually checks

Two things get evaluated. The benefits to the client: has the stated purpose been achieved, what results came out of it, and what could not be. Kubr groups client benefit into six categories: new capabilities, systems, relationships, opportunities, behavior, and performance, with performance treated as the overriding goal; capability-building that never shows up in performance risks being a costly academic exercise. The second object is the consulting process itself: was the contract's design realistic, were both sides' inputs adequate, did the engagement maximize client learning transfer rather than just deliver an answer. Interim evaluations at the close of Diagnosis and Action Planning catch problems while they're still fixable; the end-of-assignment evaluation matters most, but shouldn't be the only one.

Follow-up isn't a retainer

Follow-up is bounded and tied to the completed assignment: quarterly check-ins over a defined period, for example. A retainer is an open-ended ongoing relationship, and most retainers grow out of an assignment that was already delivered successfully; clients rarely enter an open-ended arrangement with a consultant they haven't already seen produce results.

Completion criteria: evaluation completed, or explicitly scheduled as a follow-up if it's too early to measure results; the final report delivered; commitments settled, financial and otherwise; withdrawal timing agreed by both parties rather than unilaterally announced; and follow-up or retainer status decided explicitly, not left to drift.

Where engagements break down, phase by phase

• Entry: the preliminary diagnosis quietly runs long and becomes a full diagnostic survey nobody scoped or budgeted for.
• Diagnosis: the team diagnoses a symptom, or saves every finding for one big reveal instead of confirming the problem with the client along the way.
• Action Planning: the recommendation describes the destination and skips the route: no implementation plan, no honest account of what wasn't covered.
• Implementation: support ends the day the deck is delivered, by default, with no explicit decision or line in the proposal that said so.
• Termination: the relationship fades out instead of closing: no final report, no settled commitments, no agreed withdrawal date.

Primary sources

• Milan Kubr (ed.), Management Consulting: A Guide to the Profession, 3rd ed. (Geneva: International Labour Office, 1996). The source of the five-phase model (Entry, Diagnosis, Action Planning, Implementation, Termination) and its sub-activities, checklists, and completion criteria used throughout this guide.
• Paul N. Friga, The McKinsey Engagement: A Powerful Toolkit for More Efficient and Effective Team Problem Solving (McGraw-Hill, 2008). A supplementary analytical-process framework for the diagnostic and problem-solving work inside Diagnosis and Action Planning.
• Institute of Management Consultants USA (IMC USA): the U.S. professional body for certified management consultants, publisher of the profession's code of ethics and the Certified Management Consultant (CMC) standard.