Most group leadership curricula assume authority a capability-centre manager does not hold. The decision-rights audit, cohorting rule and measurement fix.
I spent years on the buying side of this — curating faculty and designing leadership programmes for enterprise clients inside executive education. A lot of that work was for groups with several operating companies under one roof: capability centres, shared-services arms, portfolio businesses sitting under one investor.
The pattern I kept seeing took me far too long to understand. The same programme, the same faculty, the same workbook, would land in one entity and die in another. Not land badly — die. High scores, warm feedback, nothing different in March.
For a long time I assumed it was a content problem, so we localised. Different case studies, local facilitators, regional examples. It helped at the margins and it did not fix anything, because it was never the content.
Take any leadership curriculum off the shelf — yours, a provider's, the one the group bought last year — and read it only for the verbs.
Decide. Delegate. Allocate. Prioritise. Hire. Promote. Hold accountable. Escalate. Sponsor. Reallocate.
They are all authority verbs. Every one of them assumes the person in the room holds the thing being exercised.
A manager in a capability centre often holds almost none of them. They own delivery, quality, the standard of the work, and whether their best people stay. The decisions those verbs describe sit in another entity, frequently in another time zone, with someone who does not report to them and to whom they do not report.
You cannot practise delegating something you do not hold. You can role-play it, and people will, politely, because they are in a room that cost money. Then they will go back to a job where the thing they practised is not available to them, and nothing will change, and the programme will have evaluated well.
It is worth being clear that none of this is new and none of it is a deficiency in the people in the room.
Stanley Davis and Paul Lawrence published Problems of Matrix Organizations in Harvard Business Review in May 1978. From a study of companies running some form of matrix, they identified nine pathologies to which the matrix design is specifically vulnerable, along with prevention and treatment.
I am going to be careful here, because the habit I am arguing against is the one where someone names a famous paper to borrow its authority. The article is behind a paywall and I have not read it, so I am not going to tell you what the nine are, or paraphrase them, or quietly imply I know. What I can tell you is the thing that matters for your design: nearly fifty years ago, two people who studied matrix organisations concluded that its characteristic failures belong to the design, not to the managers inside it.
That is the whole argument. The difficulties your participants are describing are properties of the structure they work in. A development programme is not a treatment for a structure. It can equip people to work well inside one — which is a real and valuable thing to sell — but only if you design it for the authority they actually have.
So the first act of design is not writing objectives. It is finding out what they hold.
Take your draft list of target behaviours. Against each one, mark which of three is true for the people who will actually be in the room — not for the level the group framework says they are:
Three columns, one pass, usually an hour with two or three people who actually do the job. Here is what it looks like on a composite brief — reconstructed from the shape of several, not from any one organisation:
| Target behaviour as written by the group | Reality for a capability-centre manager | Column |
|---|---|---|
| "Allocates budget to the highest-value work" | Budget is set in the parent entity against a global plan; they consume it | Is told |
| "Builds a diverse pipeline through hiring" | They sit on panels and write the role brief; headcount and band are approved elsewhere | Influences |
| "Holds the team accountable to quality standards" | Theirs, entirely. The standard of the delivered work is the one thing nobody else owns | Decides |
| "Prioritises competing demands from stakeholders" | Priorities arrive from two global function leads who do not speak to each other | Influences — and this is the real job |
| "Develops successors and manages succession" | They know who is ready; the slate is run in the parent entity | Influences |
| "Decides when to escalate versus absorb" | Theirs. Daily. Nobody has ever taught it to them | Decides |
Now the three columns tell you three different things.
The Decides column is your practice. These are the behaviours where a participant can leave the room, do the thing on Thursday, and come back with evidence. Every piece of application work, every between-session commitment, every manager conversation should sit here. In the composite above, "decides when to escalate versus absorb" was not in the original brief at all and turned out to be the single most useful module in the programme — because it was real, daily, and unaddressed.
The Influences column is the programme. This is the counter-intuitive part and it is where most of the value is. Influencing without authority is the actual content of a multi-entity leadership job, it is genuinely teachable, and almost nobody designs for it explicitly because the group's framework does not have a competency called that. Half your design should live here: how to carry a case into a forum you do not chair, how to disagree upward across an entity boundary, what to do when two global leads give you contradictory priorities and neither will blink.
The Is told column comes out. All of it. Teaching a behaviour that the learner has no route to perform is how a programme earns a 4.6 and changes nothing — and worse, it is how participants learn that the programme is not about their job. One module of authority verbs they cannot exercise costs you the credibility of the other five.
There is one exception worth making deliberately. A behaviour in Is told can stay if you reframe it from exercising the decision to working with it: not "allocates budget," but "makes the case for a reallocation and runs the team through one they disagree with." That is a real skill, it is in Influences where it belongs, and it is honest about who holds the money.
Titles do not survive an entity boundary.
"Manager" in a capability centre can mean sixty people, three shifts and a 24-hour handover. "Director" in a small operating company under the same group can mean four people and a shared assistant. Put them in the same cohort because the HR system says they are both band 4, and you have put someone carrying a P&L in a breakout with someone who books the meeting rooms. Neither of them will say anything true for the rest of the week, and the quiet in that room is usually read as low engagement.
So group by two things you can actually measure:
A cohort where everyone is mostly in Influences can do serious work on influence without authority, because every person in the room recognises the problem. A mixed cohort cannot, because the people with authority quietly think the others are complaining.
When a sponsor insists on title-based cohorts — and they often will, because the nomination list comes out of the HR system that way — the thing to put in front of them is the audit, not an opinion. It is much harder to argue with a table showing that two nominees have four of twelve behaviours in common.
The group has a leadership framework. The entity usually has its own. And the entity's is the one people are actually rated against.
Design to the group's framework alone and you have built a programme nobody is measured on. Design to the entity's alone and you have built five unrelated programmes and lost the sponsor who is paying for one.
Use both, for different jobs:
| Carries | Why | |
|---|---|---|
| The group framework | The narrative. The sponsor conversation, the executive deck, the reason this is one programme and not five, the language of the steering group | It is the only language in which the investment makes sense as a single thing |
| The entity's framework | The objectives, the practice, the feedback language, the words on the handout | It is the language of the conversation each participant will have with their own manager in March, and that conversation is where behaviour change either survives or doesn't |
Concretely, that means a mapping table in your design document — group competency on the left, each entity's nearest equivalent across the columns, and the gaps marked honestly as gaps. It is an afternoon of work and it is the single artefact that most often decides whether the programme is renewed, because it is the thing that lets a group sponsor and a local HR director read the same page and both recognise their own words.
This is the multi-entity version of a problem every consultant meets: designing into a competency framework the client already has rather than importing your own. The multi-entity case is just the hard version — two frameworks, one of them with teeth.
This is the part that gets skipped, and it is the part that decides whether anyone can tell you the programme worked.
A Level 3 measurement plan rests on an assumption so ordinary that nobody writes it down: somebody observes the behaviour at work. In a single-entity organisation that somebody is obvious — the participant's manager, who sees the behaviour and also writes the appraisal.
In a multi-entity structure those are two different people, in two different legal entities, and frequently they have never spoken.
So the evidence and the authority to record it are in different buildings. An L3 plan that says "manager observes" is not a plan. It is a sentence that will quietly fail eight weeks after launch, and the failure will look like low response rates rather than a design error.
The fix is unglamorous. Name the observer per behaviour, and name their entity. One row per behaviour:
| Behaviour | Where the evidence is visible | Observer | Their entity | Instrument | When |
|---|---|---|---|---|---|
| Runs a structured escalation call | The call itself | Functional lead, global ops | Parent | 6-item observation note, 5 min | Weeks 4, 8, 12 |
| Argues a reallocation case in the planning forum | Forum minutes and the decision | Forum chair | Parent | Did the case get made; was it adopted | Next planning cycle |
| Holds a quality conversation with an underperformer | After the conversation | Administrative manager | Local | 3-question check-in | Week 6 and week 12 |
| Runs a development conversation against the entity framework | The conversation record | Administrative manager | Local | Existing appraisal field | Mid-year cycle |
Four rows and the design has told you something the brief never would: two of the four observers do not work for the person paying for the programme. That is not a measurement inconvenience. It is a design dependency, and it has to be closed before launch, not after.
What closing it looks like, in practice:
Everything downstream of this — the chain from a Level 2 assessment item to a Level 4 business number — works exactly as it does anywhere else, and I have written out how to build that chain separately. The multi-entity case adds one requirement and only one: every observer has an entity written next to their name, and anyone outside the sponsor's entity is a dependency you close in writing before launch.
Sometimes you run the audit and most of the behaviour list lands in Is told. That is not a reason to pad the design out and hope. It is the most valuable thing you will say in the whole engagement, and it needs saying in the first conversation rather than the last.
The words I have found work, more or less:
"I have mapped your twelve target behaviours against what the people in your nominated cohort can actually decide. Three of them they own outright. Six of them they influence but do not hold. Three of them arrive as decisions they implement. I can build you a strong programme on the nine, and I would like to — the six in the middle are the real job and nobody has ever taught them. The three at the end I would take out, because teaching someone a behaviour they have no route to perform costs us the credibility of everything else. If those three matter to you, they are a decision-rights conversation, not a development one, and I am happy to write that up separately."
That is a harder sentence to say than "yes, we can cover all twelve." It also does three things at once: it shrinks the scope you are being asked to be responsible for, it demonstrates judgement no competitor's proposal will contain, and it tells the sponsor something true about their organisation that nobody has told them. In my experience it does not lose the work. It is frequently the reason you get it.
This structure is no longer a niche.
Zinnov and nasscom's India GCC Landscape 2026 report puts 2,117 GCCs operating in India across 3,728 units, employing 2.36 million people, at $98.4bn in revenue — 32% growth since FY2021. Of those, 506 are run by Forbes Global 2000 companies, 583 are mid-market, and 504 are PE-backed (figures as of March 2026).
Every one of those is the structure described above: managers whose work is global and whose authority is local, rated against one framework and developed against another.
And the leadership development aimed at them is being sold at serious prices. One published example, because almost nobody publishes: IIM Bangalore's Leadership Development Programme to Build Resilient GCCs runs ten weeks for direct reports of country heads, at ₹4,00,000 non-residential and ₹4,55,000 residential per participant, excluding GST, for a cohort of 20 August to 17 October 2026.
That is the number a client is comparing your engagement to, whether or not anyone in the room has looked it up. I have found that the providers in this segment publish their prices and not their methods, and the advisory firms publish their diagnosis and not their design. Which leaves the part in the middle — the objectives, the cohorts, the mapping table, the observer rows — to whoever is actually holding the pen.
If that is you, the four moves above are the whole method. Run the audit before you write anything. Cohort by span and rights. Use the group's framework for the story and the entity's for the objectives. And put an entity next to every observer's name.
One thing to try, free. The hardest part of all of this is the conversation where you find out what your client's managers can actually decide — and most discovery calls never get there, because the stated problem sounds like a training problem and everyone is relieved to agree. Our Diagnostic Conversation Coach is free, needs no signup and no email, and it works on exactly that: the questions that separate a capability gap from an authority gap, and what to ask next when the answer surprises you. If it does not sharpen the way you run a discovery call, nothing else here will interest you either.
The worked examples in this article are composite reconstructions drawn from the shape of several engagements. They are not records of any identifiable organisation, and no client dialogue is reproduced.