Designing the Roles Before Anyone Knows Who Fills Them

The case says the combined company will run on fewer, differently shaped roles, and the arithmetic is usually sound. But a number cannot be executed. Somebody has to say what the combined function consists of — which accountabilities exist, at what level, carrying which decision rights — before anyone can be appointed or released. That definition is the synergy; everything after it is reporting.

The act gets delegated, on a deadline, to the leaders of both sides: the only people who understand the work, and the only people with something to lose from every simplification. Both are true at once, and the second wins quietly. What comes back is the structure in which most of the room is safe.

What actually happened

Polymath’s founder held the human capital side of acquisitions for a TSX-listed consumer packaged goods group with roughly $2–3 billion of revenue and payroll in approximately 21 countries, which grew by buying companies — bolt-on transactions first, and later diligence and integration planning through a mid-size acquisition.

Each of those cases carried a resourcing line, and therefore an instruction that appeared on no plan: before anybody can leave, join or move, somebody has to write down what the combined organization is made of. It arrived as a date.

The first few times, it was done the way it is done everywhere: joint sessions, the receiving function head opposite the acquired function head, both with their charts. Matching titles settled nothing, because acquired populations arrive with position definitions, approval thresholds, grading and spans matching nothing in the acquirer’s estate. Every mapping was argued, and each argument was between two people, one of whose jobs was its subject.

The structures that came out had a family resemblance. Where each side held one role, the design often produced two, split by geography, by segment, or by a transitional phase whose end date nobody enforced. Where a layer could have gone, it was renamed. Where an acquired role had no counterpart, a position appeared to receive its holder. Every choice had a defence, and together they preserved both organizations that entered. Nobody misled anyone. Those leaders were in the room because they knew how the work ran, and no outsider could have designed credibly without them. They were also specifying, in front of a counterpart, a structure that would decide whether they still had a job. Expecting impartiality there is a staffing error, not a character flaw.

The second arrangement was deliberate. Diligence and integration planning sat with one team instead of crossing a boundary where the incentives changed, so the role work could begin before the joint sessions rather than inside them. Target roles were specified against the job architecture first — accountabilities, decision rights, span, grade, the position each would occupy — by people whose own roles were out of scope. Leaders on both sides were consulted as sources; neither authored the structure that would contain them.

The change that did most was smaller and less comfortable. Where a leader did have to contribute, their own outcome was settled in writing first, and that altered the subject of the argument: it stopped being whether a position was needed, which is unanswerable when the answer decides your future, and became whether the position could be described.

Polymath now specifies a client’s target roles before the people who might fill them are in the room, and settles the position of anyone who must be.

Why it happens

Two things about an integration collide, and neither is a mistake on its own.

The first is that role design needs operating knowledge held only inside the two businesses. How a category is actually managed, which approvals genuinely protect something, which span works in a market with fragmented distribution — none of it is in a data room, and no adviser can supply it. The people who hold it are the functional leaders, which is why convening them is correct. The second is that the same integration is how those leaders find out whether they are employed. Deal timetables compress design and selection into the same weeks and often the same forum, because both are urgent and both turn on the same names.

What follows does not look like self-interest. It looks like prudence. Every simplification in a merger destroys a position, and a leader asked to endorse one while their own status is unresolved has a single riskless answer: keep the capability, phase the change, revisit after stabilization. Nobody can prove in advance that the removed thing would go unmissed.

Aggregate those prudent answers and the shape is predictable, though nobody intended it. Both structures survive inside the combined chart. Duplicated roles are separated by geography or segment rather than merged. Layers are renamed. Coordination roles appear to manage the seam the design created. Selection then runs into that structure and ratifies it: by the time appointments are made, the design has already conceded the synergy.

One neighbouring failure needs separating from this. Our cost-programme argument concerns a cut taken without a redesign, where the work never stops and relocates to somebody with less room. This is the opposite. The redesign was done — sessions, documentation, accountabilities, a chart — and authored by people who could not be impartial about it, so what it produced was a reasoned justification for the existing shape. A programme can fail this way while doing everything the methodology asks.

Which is why the usual instruments fall short. A stronger integration office sequences the design and escalates a delay; it cannot supply an impartial author, because it has none. Independent facilitation improves the conversation and leaves the conclusions with the same participants. A retention framework treats the problem as motivational, paying people to stay while still asking them to design their own jobs. Not one of them alters who authors the structure.

The strongest objection

The strongest case against this: the leaders are the only people who know how the work runs, and a structure drawn by anybody else will be elegant, wrong and rejected on contact with the business.

Concede it fully, because it is true. Designs produced without operating knowledge fail exactly that way, and integrations are littered with target operating models accurate about nothing. But the objection conflates two uses of those leaders. As sources they are irreplaceable: what this team decides, what breaks if an approval goes, why this market needs its own structure. As authors they are compromised, structurally rather than personally. Separating the two keeps the knowledge, loses none of the rigour, and removes the requirement that somebody argue against their own employment.

The harder objection usually ends the conversation: you cannot settle everybody’s outcome first. Somebody is told last, the executive team’s own shape is often unresolved until late, and promising a leader their role before the design exists is either undeliverable or proof the design was predetermined. The constraint is real. This is a gradient, not a switch. For most of the structure a designer can be found whose own role is out of scope: a leader from a part of the business the deal does not touch, the job architecture owner, an HR leader with no line in the new organization. Where nobody neutral exists, make the conflict explicit — name it, have the design challenged by somebody not exposed to it, and let the executive decide knowing who drew it. A declared conflict is manageable. An undeclared one is the default, and it is what produces the chart.

The method

Six steps. One and two settle what the design must achieve and who may draw it; three and four produce the structure; five and six populate and hold it.

1. We write the design brief from the case before anyone draws a box. We convert the resourcing line into what the combined company must be able to do: which decisions have to be takeable and at what level, which capabilities must exist in your estate, what the structure may cost. The trade-off we accept: we delay the first joint session, which everyone wants held early for goodwill. Skip it and the brief becomes whatever the first chart shows.
2. We settle the outcome of everyone who will author the structure, in writing, before they author it. Each contributor is either outside the design or has been told what their own role in it is. Where we cannot get that for a leader, we take them out of the authorship, use them as a source, and record why. The trade-off we accept: we force senior decisions earlier than an integration would, while the transaction is still being celebrated. Skip it and everything after this step is produced by people with an interest in its result.
3. We define positions rather than boxes and names. We give each target role accountabilities, decision rights, a span, a grade and a place in your job architecture, written while it is unknown who might hold it. The trade-off we accept: we are slower than a chart, and we produce a document nobody absorbs at a glance in a steering meeting. Skip it and the design can be compared with nothing — not the case, not your estate, not the acquired one.
4. We reconcile the structure to the brief while it is still unpopulated. We test every position against what the brief says the company must be able to do, and we challenge anything answering to no requirement in it. Positions that exist because a person exists are visible here and invisible later. The trade-off we accept: some of those deletions get described afterwards as removing good people, and we hear it directly. Skip it and the reconciliation waits until after selection, where the only instrument left is a second redundancy.
5. We publish the structure before we open selection, and we fix the criteria before we know the field. We issue the roles with accountabilities and grades attached, and only then do we run selection. The trade-off we accept: an unpopulated structure is uncomfortable to look at, and we give up the option of adjusting a role to suit a preferred candidate. Skip it and the appointments confirm a design already built around the people who get them.
6. We put the structure under change control from the first day of combined operation. Any position created, regraded or inserted after go-live is declared against the design, with a reason and a named approver. The trade-off we accept: we add an obligation on leaders who have just absorbed an acquisition and feel owed some latitude. Skip it and the old shape rebuilds itself one reasonable exception at a time.

Step two carries the outcome. The rest is technique, available to anybody; what decides whether it is applied honestly is whether the person applying it knows where they stand. Run everything but two and you get a well-documented version of the same structure at greater expense.

Evidence and measures

The decisions an integration takes about roles, and who is in the room for each:

Five measures worth tracking over ninety days. None of them requires a new system:

— Target positions defined by accountabilities, decision rights and grade before any candidate name was attached.

— How many authors of the structure had their own role settled in writing first. Begin with the ones who did not.

— Positions in the target structure answering to no requirement in the brief. That count is the part of the organization built to hold people.

— Layers between the chief executive and the front line, in both estates before and combined after. A merged company deeper than either was not designed.

— Positions created, regraded or inserted after go-live, split by whether they were declared or found later.

Read the first measure beside the last. A structure specified before the names and still recognizable when the first planning cycle closes is a resourcing synergy. Anything else was a reorganization timed to a deal.

How Polymath solves this

We start with the resourcing line in your case and the integration calendar already set, and we produce the design brief before any joint session is convened: what the combined company must be able to decide, which capabilities must exist on your side of the transaction, what the structure may cost. From there we specify the target positions against your job architecture — accountabilities, decision rights, span, grade — while it is unknown who will hold them. The discipline is Change Management, because the analysis is the straightforward half and the authorship is the hard one.

We work inside the timetable the deal has already imposed, with the chief human resources officer, the integration lead, the job architecture owner and the receiving function heads. Leaders on both sides are interviewed as sources, and where one has to contribute to the design we settle their own position first and say so in the record. Your HR business partners run selection against criteria we write with them before the field is known, and they own the change-control log from the first combined cycle.

What accumulates is the separation itself, held as a rule rather than a preference. Once an organization has designed a structure before knowing who fills it, that becomes the precedent for the next transaction and for every ordinary org design after it: the target positions are reusable, the receiving architecture is described, and the argument about who holds the pen no longer has to be won from scratch. Your second integration starts from a specified estate instead of two charts and a deadline.

What it costs to do nothing

The first cost is the resourcing line itself, and it is the least interesting. A structure preserving both shapes costs roughly what the case said it would save, and the gap gets reported as phasing rather than a miss. Recoverable, in principle, in a later round. The second cost is that later round. A restructure following an integration by a year is read by everybody as the integration having failed, which is accurate. It runs without the protections the deal provided: no retention framework, no transaction narrative, no expectation that change was coming. And it lands on the population that stayed, who accepted the first structure in good faith.

The third cost is what the organization learns, and it outlives everyone involved. Leaders who watched a design session decide their futures take away an accurate lesson: protect your team by being in the room when the chart is drawn, and protect a position by making it hard to describe. That behaviour is rational, durable, and arrives fully formed at the next acquisition, where the structure is larger.

The move available now is smaller than a programme. Take the design your integration is producing this month and, for each contributor, write down whether they already know what their own job in it will be. The blanks are the parts of the structure you should expect to survive unchanged, and that is the conversation we start from.