An organisation chart is finished the moment somebody signs it. The reorganisation it describes is finished much later, when the people named on it are actually sitting in the jobs it gave them, and it is entirely possible for the first thing to happen and the second thing never to. The distance between the two is where restructures quietly fail, and it is almost never inside the scope of the work that produced the chart.
A European travel and tourism group was days from completing the structural blueprint for a new digital function. The proposal for the next phase opened by saying the thing most follow-on proposals leave implied: a structure is not an operating model, and a structure on its own cannot be run. It broke the difference into three streams of activity, and one of them was the organisation itself, carried down past the boxes.
That is the finding this piece exists to carry. The organisational output was specified, in writing, as a complete mapping of individuals to the integrated design, a review of the role profiles and headcounts that already existed, the key gaps that remained once the mapping was done, and a transition plan carrying the team and individual moves that had been agreed. Underneath all of it, a competency model. Not a chart with an approval date. A list of people, each against a job, with the leftovers written down.
The challenge
The reason this scope is rare is that it does not parallelise and it does not survive weak data, and both of those show up in the plan rather than in the rhetoric.
It does not parallelise. Report mapping and role descriptions were to be split into two waves: priority areas by the end of the first month, everything else by the end of the second. Nobody stages a task in waves when the task is easy. Mapping a person to a role is a judgement made one at a time by someone who knows both, and the only way to compress it is to decide which parts of the organisation get the careful version first. Publishing that choice as two dates beats pretending the whole population resolves at once.
It does not survive weak data either, and the proposal named its own dependency instead of hiding it. The integrated structure was to be designed from a fact-based cost and resource review. If that review was not complete by the middle of the first month, the design would be built instead from headcount and job-title data held in HR. Putting the degradation path in the scope document, with a date on it, converts a future argument into a decision already taken.
A job title is a string somebody typed once and nobody has since had reason to correct. Grade is a payroll fact, accurate about what a person is paid and silent on what they can do. Skills fit is usually recorded nowhere at all. Designing from the first two and asserting the third is how you get a structure that is defensible on paper and wrong about the people inside it. The proposal does not claim the fallback is equivalent. It says which version you get, and when.
Then there is the sequencing, which is the part worth putting in front of any executive about to announce a restructure. The competency model, the role mapping, the grade and skills fit and the gap analysis all sit inside the quarter of work. The communication to the organisation and the first transfers sit outside it, in a later quarter, and the proposal states plainly that the earlier work is done in order to be ready for them. Recruitment runs on the same logic and appears as milestones of its own: a recruitment plan agreed inside the quarter, key roles in place at the far end of it. That ordering contains an admission: once you have mapped everybody you can map, some seats are still empty, and the gap list is the recruitment brief. An organisation that announces first finds the same gaps in public, from the people who read the chart and could not find themselves on it.
The competency model is what makes the rest reusable. Grade harmonisation, skills fit and career path definition all appear as named activities, and none of them is a one-time exercise. This is where the argument reaches something we sell: skills held as data against roles, so fit and gap can be asked as a question rather than reassembled by hand every time the structure moves. Synapsa exists to hold that record after the consultants leave.
The approach
The proposal set out what would exist at the end, dimension by dimension, and the discipline is in how specific each artefact is.
For process, level-two flows with roles for around thirty main processes, five taken down to level three, with the first ten sequenced ahead of the remainder. The plan names its ordering rule too: start with the complex and critical areas, and put the actual process owner in the room rather than designing around them. Awareness and training material is listed as an output of the operating model itself rather than a follow-on project. A process nobody has been walked through is a document.
For governance, a described gating cycle plus a defined set of customer, product and executive forums. Governance as a meeting cadence with named attendees is far easier to run than a policy document. For the boundaries, formal rules of engagement between the new digital function and three commercial parts of the business, treated as an artefact in its own right. The standard failure of a new central function is unowned edges, and the only defence is writing them down before anyone is standing on them.
Put the process stream and the organisation stream side by side and the join becomes obvious: the flows were specified as level-two flows with roles. A role description means very little alone and a great deal when it is anchored to the steps a person is accountable for in a named process.
The outcome
This was a proposal. The individual mapping, the transition plan, the competency model and the process flows are scoped deliverables, and this piece reports no result for any of them. What can be said honestly is that the scope was drawn in the right unit, and that the unit is the rare part.
If the same engagement were scoped today, three things would change and none of them is the argument above.
The thirty process descriptions would not start in a workshop. Most of those processes already run through systems that write event logs, and process mining over those logs gives you the actual flow, with its variants and rework and real handoff points, before anyone draws the intended one. The workshop is then spent on the difference between the two, which is the only part worth an expensive room.
The people data would be treated as a data-readiness problem rather than an HR extract. The fallback to job titles is only available because nobody has ever been accountable for the lineage of the people record: where a grade came from, when a skill was last assessed, which system wins when two disagree. That is the same work we do on any estate before a model is allowed near it, and it is cheaper once than at every reorganisation.
And the cautious language-model pilots now appearing in HR functions change less than they look like they will. A model can draft a role description from a process flow and save a genuine week. It cannot tell you whether a particular person fits a particular grade, because the evidence for that judgement is not in any document it can read. Our Consult work starts by separating those two categories, since a pilot aimed at the second fails slowly and expensively.
A reorganisation announced and a reorganisation executed are different deliverables with different completion dates. Only one of them is a chart.