Was this newsletter forwarded to you? Sign up to get it in your inbox.
The sponsor who leaves
Your engagement was signed by an organization and carried by one person. The day that person changes jobs, their successor inherits a commitment they never chose. Here is how to build an engagement that survives the departure.
An engagement is signed with an organization and runs on a person. As long as that person stays, you cannot tell the difference. The day they change jobs, you find out which of the two you really had. To my mind, it is the most predictable risk in this work, and the least prepared for: everyone knows people change jobs, and engagements still get written as if ours will be the exception. What follows is the work you do while that person is still there.
Francis Beaulieu
Why this matters to you now
The Project Management Institute has been surveying thousands of practitioners about their projects since 2006. In the 2018 edition of its Pulse of the Profession, based on 4,455 practitioners, the top cause of failure was neither budget nor skill. It was a change in the organization's priorities, named by 39% of respondents. Inadequate sponsor support came further down, at 26%. The same respondents estimated that only 62% of the projects completed in their organization that year had a sponsor who was actively supportive.
These numbers come from people managing internal projects and judging their own work, each able to tick up to three causes, and they date from 2018. As far as I know, nobody has measured the same thing for consulting engagements. What survives is a ranking: shifting priorities come ahead of weak sponsor support.
The table does not measure departures. I would still ask you to reread those two causes with a departure in mind. Priorities change because the person who carried them is gone, and support disappears for the same reason. One departure can bring both on the same day.
Here is the arithmetic, with a number I am setting as an assumption, not a measurement. Say any given person has a one-in-five chance of changing jobs during your engagement. If your engagement rests on one person, it has a one-in-five chance of becoming an orphan. If it rests on three people whose departures are independent, the chance of losing all three drops to 0.8%. But the chance of losing at least one rises to 48.8%. Three anchors do not spare you departures. They make you live through more of them, but no single departure takes everything at once.
The calculation hides two traps. The first is the word "independent." A reorganization does not move people one at a time. It moves whole departments. Three anchors in the same department will leave together, and you are back to one in five. The second is quieter: the calculation assumes any one of the three can defend the engagement alone. Three names that each hold a different piece protect you from losing everything, not from the jolts. So the useful rule fits on one line: names in different places, rather than more names.
The end of the year makes this more pressing. Edition #32 on the budget written without you showed that next year's budgets are prepared in the fall, and organization charts sometimes move with them. Edition #22 on the phantom engagement already listed "sponsor gone or replaced without handoff" among the signals of an engagement that is drifting, and proposed a rescoping meeting once more than 30% of the client-side decision-makers had changed. Its indicators were written in advance, but they fired once the departure had happened. This one is about what you do before, while nothing is urgent yet.
Your sponsor may well leave. What they take with them is largely up to you.
Pricing: your exposure on the day they leave
The move: for every active engagement, work out how much work you have done and not yet invoiced. Then make sure that number never exceeds one phase.
Why it changes everything: what happens when a client stops an engagement partway depends on where you work. In most provinces, it is the termination clause in your contract, if you have one. In Quebec, the Civil Code answers by default, more clearly than most people think. Article 2125 says the client "may unilaterally resiliate the contract even though the work or provision of service is already in progress" (resiliate is Quebec's legal English for terminate). Article 2129 then requires the client to pay, "in proportion to the agreed price," the actual costs and expenses and "the value of the work performed." Either way, a successor can stop your engagement tomorrow morning, and what follows is an argument about what your work was worth.
The risk sits in that word "value." The value of half-finished work is hard to show when nothing has been delivered: a diagnostic three-quarters done, interviews with no synthesis, an analysis with no recommendation. Take an illustrative $120,000 engagement over six months, and a departure in month four. If you invoice at the end, you have $80,000 of work to get recognized by someone who did not choose you. If you invoice monthly against an accepted deliverable, you have $20,000 at stake, and the first $60,000 is already locked in by decisions the organization made.
The trap to avoid: assuming monthly invoicing solves it. A monthly invoice with no deliverable proves you worked, not what you produced. What turns your work into an organizational decision is a phase that closes on something the client accepted, in writing. Edition #35 on the proposal nobody turned down wanted a first phase that stands on its own, to get the signature. The same rule, applied to every phase, caps what you lose if someone else stops the engagement.
The warning: this is not legal advice. Have your lawyer review your template, starting with what it says about termination, before you change its structure.
This week: two numbers per active engagement: the work done and not invoiced as of today, and the invoices sent but not yet paid, since a suspended engagement often suspends payments with it. If the first one exceeds a month of fees, suggest to your client that you close out the current phase with an interim deliverable, framed as a checkpoint for them. It is a change to the contract, small but real: with a public-sector client, it will take an amendment. For your next engagements, the split goes into the proposal, where it costs nothing.
Sales and business development: three names before you sign
The move: before you sign, write down three names: the person who carries the engagement, the person whose budget pays for it, and the person who will live with the result. If the same name fills two boxes, your engagement hangs by one thread.
If your client is an owner-managed business, where the same person signs, pays and lives with the result, the three boxes will always carry one name, and the departure takes another form: a retirement, a succession, a sale. The field note in edition #22 told exactly that story: the CEO of a family business retiring six months into the engagement. The rest of this section applies mostly to organizations where your sponsor has a boss.
Why now: in 1997, Ronald Mitchell, Bradley Agle and Donna Wood proposed a theory of stakeholder salience that is still a reference point. A stakeholder matters to an organization according to three attributes: power, legitimacy and urgency. Two of their observations about those attributes shed light on a sponsor's departure. The attributes, they write, are "variable, not steady state." And power is "transitory: it can be acquired as well as lost."
Their theory is about how a firm's managers prioritize the people who matter to it. I am transposing it to your engagement, and the transposition is mine. On signing day, your sponsor often holds all three attributes: the power to sign, the legitimacy of the person who asked for the project, the urgency of the person who needs it. On the day they leave, the three come apart. Power stays with the position and passes to the successor. Urgency leaves with the person. The engagement's legitimacy has to be earned again, from someone who never asked for it. Your three boxes follow that split: the person whose budget pays holds the power, the person who carries the engagement gives it its legitimacy, and the person who will live with the result carries the urgency. Three names at signing means three attributes that will not all leave on the same day.
This is a different exercise from edition #20 on the client you abandon, which mapped the neighbouring decision-makers to find the next engagement. Here you are looking for the people who keep this one alive. And the field note in edition #3 on discovery already described a consultant interviewing stakeholders the sponsor does not choose. That is often where the other two names are.
The trap to avoid: looking for the other names behind your sponsor's back. They will find out, and they will read it as distrust. Ask the question openly, during scoping: "If you are not in the room on the day this engagement has to be defended, who defends it?" A solid sponsor has an answer. One who does not has just told you something.
This week: three boxes for every active engagement, and one question for every empty box. Mid-engagement, ask it at a milestone, as a planning question: "Who else should receive the phase two deliverable?"
Collaboration networks: the person leaving, the person arriving
The move: for every active engagement, get two commitments from your sponsor now, for the day they might be called elsewhere: that they will introduce you themselves to whoever takes over, and that the engagement will appear in their handover notes, in three lines in their own words. Asked during scoping or at a milestone, it is an ordinary request. Asked after the announcement, it comes too late.
The mechanism: the person leaving no longer has the power to defend your engagement, but for a few weeks they still have enough credibility to make an introduction. An introduction from the predecessor carries more weight than an email from you, landing with no context in an inbox that is already full. And since the person taking over is often not named right away, the handover notes are sometimes the only introduction there is: the successor, or whoever covers in the meantime, will read them before meeting you.
The person arriving starts by judging what they did not choose. That is normal, and even healthy: an inherited commitment deserves to be understood before it is continued. Your first meeting is there to give them, in twenty minutes, what they need to decide: why the engagement exists, what has been decided, what remains, and what stopping now would cost. It is not there to resell the engagement. Someone new is wary of anyone selling them something in their first week, and they are right to be. Prepared this way, that meeting also shortens the rescoping meeting edition #22 proposed within fifteen days.
There is a third relationship, and it matters later. The person leaving takes your work with them to their next job. Stay in touch, without asking for anything. They know what you are worth, and that is rare in someone walking into a new organization.
The trap to avoid: waiting for the announcement to ask for anything, then asking the person leaving to save the engagement. They no longer have the power to, and every commitment they make in their last weeks is one their successor can undo.
This week: for every active engagement, one line: if the announcement came tomorrow, who would introduce you to the person taking over? An answer of "nobody" names the request to make of your sponsor at the next milestone.
Value creation: an engagement with a memory
The move: keep a decision log for the engagement, attach it to the notes of every meeting, and have it corrected at the start of the next one, along with the minutes. With no steering committee, the same log goes out by email after every meeting, with one request: "Correct me by Friday."
Why it holds: in 1991, James Walsh and Gerardo Ungson proposed a framework for organizational memory that is still the starting point for the field. An organization keeps what it knows in six places: individuals, culture, procedures, structures (meaning roles), the physical environment, and external archives. Only one of those places leaves when a person leaves their job. It happens to be the one where most engagements live: in the head of the person who commissioned it, and in yours.
The log itself is not on their list: their framework did not count an organization's internal files, as other researchers have since pointed out. A log sitting in a folder protects nothing. What makes it hold is two of the six places. Corrected and validated at every meeting, it becomes one of the client's procedures and stops being your document. Written into a role rather than a person, it enters the structures: "the finance department signs off on each phase" survives a departure, "Julie signs off on each phase" does not. Each line fits in one sentence: the date, the decision, who made it, why, and what was ruled out.
The test: could someone new understand, alone, in twenty minutes, why the engagement exists and what has already been decided? If the answer runs through you, you have become the engagement's memory. That is a flattering position and a fragile one: the consultant explaining to the client why the client hired them.
The trap to avoid: a log you keep alone and nobody corrects. It lives in the same place as your memory. Edition #34 on the method you never wrote down put your decision rules into the deliverable, so the client could see why you made the calls you made. This log records the client's own decisions, and the client is the one who validates it.
This week: open a blank document and reconstruct the five most important decisions in your main engagement since it started. If you cannot find who made one of them, you have found the line to get confirmed at the next meeting.
AI: count who decides
The move: give a model one month of meeting notes and ask for two things: the list of decisions, and who is behind each one.
What stays out: names. Your meeting notes name people and internal decisions that are not public. Replace each name with a code, P1, P2, P3, always the same one for the same person, and keep the key on your side. The codes have to stay stable, because the second half of the exercise depends on them.
What it shows you: you now know you should keep the log. What you do not know is how many decisions run through the same person. Ask for the decisions made, one per line, with the date and the code of the person who made it. If your notes say "the committee decided," ask instead who proposed each decision: that is the name that leaves. Then do the count per code in a spreadsheet rather than in the model, which can miscount. If one code sits behind almost every decision in the engagement, you have your answer about the single thread, in numbers, and it is harder to ignore than a hunch.
The warning: one month at a time, not nine. Nelson Liu and his co-authors measured this in a study published in 2024 in the Transactions of the Association for Computational Linguistics: model performance drops markedly when the relevant information sits in the middle of a long context. That holds even for models built for long contexts. The ones they tested have since been replaced, but the move that sidesteps the problem costs nothing: one month at a time, then consolidate in your spreadsheet.
The trap to avoid: letting it write the why. Edition #34 already showed that a model will rationalize an answer without giving the real reason for it. Here it would invent the reason for a decision it never saw being made, and a plausible reason in a decision log is worse than an empty cell. The model sorts what was said. It does not know what was thought.
This week: your last three sets of meeting notes, codes instead of names, the decisions as a table, then the count per code in your spreadsheet. If one name sits behind everything, you know which conversation to have before the end of the year.
Like what you read? Get this in your inbox every Tuesday.
Free. No spam. Unsubscribe in one click.