Leadership and People Skills
Display

Theme

Text size

Font

Night is a warm, low-blue theme for reading after dark. Your choices are remembered on this device.

Download PDF(PDF)
Session 3

Organisation Structure

Session by Prof. Vidyanand Jha · Robbins, Judge & Vohra, Chapter 14 · Cases: The Restaurant; Acme and Omega

Why structure matters

Greg James’s team in Session 2 didn’t fail for lack of effort or goodwill. It failed for lack of an architecture: who owns what, who reports to whom, where decisions sit. That architecture is not an org chart drawn after the fact. It is designed, and the design decides what the people inside can do.

An organisational structure defines how job tasks are formally divided, grouped and coordinated.

Structure serves effectiveness, but whose?

Prof. Jha opened with a bigger question. Structure exists to make an organisation effective, but effectiveness isn’t one number. Employees, customers, shareholders and regulators each judge it differently, and their measures pull in different directions: higher salaries please employees and cut short-term returns. A manager could track hundreds of measures; the skill is choosing which ones go on the dashboard.

The strategic constituencies model says an organisation organises itself around whoever is most critical to its survival. When the US airline industry was heavily regulated, winning route rights was the survival issue, so lawyers rose to run airlines. A consumer-goods firm elevates sales; a cash-starved firm elevates finance; a public-sector unit orients itself to its ministry.

The seven dials

Every structure ever designed is a setting of the same seven elements.

ElementWhat it meansThe question it answers
Work specialisationHow far work is divided into separate jobsHow narrow is each job?
DepartmentalisationThe basis for grouping jobsHow are we organised into units?
Chain of commandThe unbroken line of authority from top to bottomWho do I take orders from?
Span of controlHow many people one manager directsHow many people per manager?
Centralisation and decentralisationWhere decisions are madeHow much authority sits lower down?
FormalisationHow far jobs are standardised by written rulesHow rule-bound is the work?
Boundary spanningContact across group and organisational boundariesWho reads the outside world?

A mechanistic firm turns specialisation, centralisation and formalisation up; an early-stage startup has them all near zero; Indian Railways sits near the maximum. Neither extreme is right everywhere.

Prof. Jha’s three dimensions

The seven elements can be grouped into three dimensions:

  1. Complexity, made up of horizontal differentiation (division of labour across jobs), vertical differentiation (the number of layers) and spatial differentiation (spread across locations). More of any one makes the organisation more complex.
  2. Standardisation, used interchangeably with formalisation: how far rules and procedures fix the work.
  3. Centralisation versus decentralisation: where decisions are made.

Work specialisation: gains, then a ceiling

Specialisation raises productivity, up to a point. Ford’s moving assembly line cut the time to assemble a Model T chassis from about 12.5 hours to about 1.5 hours. But the gains flatten and then reverse as boredom, errors, absenteeism and turnover set in.

Productivity and specialisation

Economies tasks mastered faster Optimum Human diseconomies boredom, errors, absence, turnover Work specialisation → Productivity →
The optimum moves right (more specialisation) for simple physical work and left for complex, knowledge or creative work.

McDonald’s runs extreme specialisation: consistent and fast, with high staff turnover. Toyota builds in rotation and quality participation and gets lower turnover and higher productivity per worker.

The work-redesign movement grew from this ceiling. Rotation had limits in practice (you can’t make the cook the security guard). Job enlargement added more tasks at the same level, which doesn’t help if they are all dull. Job enrichment gave people a slice of the boss’s job, which does. An IIM Calcutta colleague’s experiment at a post office replaced single-task counters with single-window service: queues shortened and staff satisfaction rose.

Departmentalisation: five ways to group jobs

BasisHow it worksExample
FunctionalGroup similar specialists togetherAcme; cooks with cooks, waiters with waiters
Product or serviceEach line has its own resourcesTata Group: Steel, Motors, TCS
GeographicGroup by regionRegional structures in consumer goods
ProcessGroup around stages of the workflowKitchen → runner → pantry → floor
CustomerGroup by type of customerA bank’s retail, corporate and private-wealth arms

Every department builds its own identity and its own “us versus them” (“we bring in the money”, “we sell it”). That is differentiation, and an organisation needs as much of it as its environment demands. But the specialists must then be pulled back together, which is integration: cross-functional teams, regular all-hands, a shared goal everyone serves. Good structure does both.

Chain of command and span of control

Unity of command means each person reports to one boss; two bosses with conflicting instructions produce paralysis. Span of control sets the shape of the whole organisation.

4,096 front-line workers, two spans of control

Span of 46 levels of management · 1,365 managers

Level 1: 1
Level 2: 4
Level 3: 16
Level 4: 64
Level 5: 256
Level 6: 1,024
Front line: 4,096

Span of 84 levels of management · 585 managers

Level 1: 1
Level 2: 8
Level 3: 64
Level 4: 512
Front line: 4,096
Doubling the span from 4 to 8 removes two levels of management and 780 managers. Wide spans are cheaper and faster but only work when the work is routine and staff are skilled.

Span depends on how critical the task is, how skilled the staff are and how complex the work is. There is a human trade-off too: customers prefer flat organisations (a complaint is fixed faster with fewer layers), while many employees prefer tall ones, because layers mean more promotions and titles.

What is a supervisor actually for?

Every layer of hierarchy costs money. A chicken chain with three layers of managers carries overheads that a single-owner shop selling similar food doesn’t, so the lean shop can undercut it. Prof. Jha split the supervisor’s job in two:

  • Control (making sure people don’t slack) is increasingly hard to justify in knowledge work.
  • Coordination (making sure hand-offs between people and processes work) remains essential.

Quality shows the same shift: end-of-line inspection is giving way to in-process quality, where each step checks the one before it. Every layer must earn its keep.

Centralisation, decentralisation and formalisation

Centralised organisations decide at the top: consistent and controlled, but slow, and the boss becomes a single point of failure. Decentralised ones push authority down: faster, more engaging, and they develop future leaders, at the risk of inconsistency. Amul is decentralised by design: farmers and village cooperatives decide locally while the federation handles processing and distribution. Most firms mix the two, centralising brand, strategy and capital, and decentralising front-line service.

The fly in the soup

Prof. Jha’s running example. Under centralisation the waiter says “I’m not authorised, I’ll escalate”, and days later a letter offers a free soup: slow and useless, and the story distorts as it travels up. Under decentralisation the waiter replaces the soup at once, because that’s where the information is. But pure decentralisation invites abuse. The elegant answer is to control the premises of the decision, not the decision itself: every waiter may give away five soups a month, no questions asked, and anything beyond that is reviewed. Decentralised and controlled at once.

Formalisation ranges from near-total (a space mission control room) to none at all (developing a film concept).

Boundary spanning is mostly about the edge between the organisation and its environment: HR reads the hiring market, sales reads demand, top management reads everything. Citing Karl Weick’s idea of enactment, Prof. Jha stressed that there is no single objective environment. People see what their training and interests let them see: a finance person reads finance signals, a marketer reads market signals, and the same forecast produces opposite strategies in two firms.

Eight common structures

StructureStrengthsWeaknessesExample
SimpleFast, flexible, clear accountabilityCan’t scale; everything depends on one personA kirana store
BureaucracyEfficient for high-volume routine work; consistentRigid; rule-following crowds out problem-solvingIndian Railways
FunctionalDeep expertise; clear career pathsSilos; poor coordination across functionsTraditional manufacturing
DivisionalEach division focuses on its market; clear accountabilityDuplicated resources; divisions competeTata, Reliance
MatrixShares scarce specialists; cross-functional expertiseTwo bosses: breaks unity of command and creates role conflictConsulting, aerospace
VirtualSmall core, minimal overhead; buys best-in-class partnersLoss of control; culture is hard to buildA brand that outsources manufacturing
TeamBreaks silos; high ownershipHigh coordination cost without shared directionMission and product teams
CircularSignals equality and opennessUnclear authority; confusing to hierarchical thinkersSome NGOs

The matrix’s built-in conflict

The matrix deliberately gives people two bosses: a functional manager and a project manager. When they want different things from the same person, that is the role conflict from Session 2, now built into the organisation chart. A matrix works only with strong psychological safety and clear rules for resolving conflicts.

How you employ people is a structural choice

Prof. Jha took virtual structures past the textbook. Casualisation, where firms want “switch on, switch off” labour rather than permanent staff, is the structural root of the gig economy. Its human cost is precarity: insecure work leads people to postpone families and life decisions. His reading suggestion was Michael Sandel’s What Money Can’t Buy.

Mechanistic versus organic

The eight structures sit along a spectrum between two ideal types.

MechanisticOrganic
SpecialisationHigh; narrow rolesLow; overlapping, cross-functional
AuthorityRigid hierarchyFollows expertise, not position
Span of controlNarrow; close supervisionWide; self-direction expected
FormalisationHigh; explicit rulesLow; method left to judgement
DecisionsCentralisedDecentralised, participative
InformationTop-down, filteredFlows in every direction
Adapting to changeSlowFast
Best forStable environments, cost control, routine workDynamic environments, innovation, novel work

A traditional joint family, where elders decide and roles are fixed, is mechanistic. A nuclear family with two working partners who debate every decision is organic. Neither is better; one is built for stability, the other for change.

Four factors decide the fit

Strategy Cost minimisation → mechanistic. Innovation → organic. Imitation (copy what works, adapt it) → a mix. The most important factor
Size Large firms need rules to coordinate thousands of people. Small ones can still coordinate informally.
Technology Routine (same inputs, same outputs) → mechanistic. Non-routine, custom work → organic.
Environment Stable, predictable → mechanistic. Dynamic and uncertain → organic.

Structure shapes behaviour

FeatureEffect on people
High specialisationShort-term efficiency; long-term boredom and absence once variety drops too low, whatever the pay
Narrow spanCan feel suffocating to skilled staff, but helps learning on complex, high-stakes work
High formalisationFewer errors on routine work; less initiative when something new happens
DecentralisationHigher satisfaction and commitment, when people feel able to handle the responsibility
Individual fitPeople who need clarity thrive in mechanistic structures; people who need autonomy in organic ones

Downsizing, the permanent, deliberate reduction of the workforce, belongs here too. Done as a redesign of the work, it brings lasting speed. Done as a payroll cut without changing the work, it produces survivor syndrome: the remaining staff carry the load, morale falls and knowledge walks out of the door. The test is simple: did we redesign the work, or just remove the people?

The cases

The Restaurant: structure evolving in real time

A diner grows through four stages, and its owner works through the seven dials without knowing their names.

Stage 1Simple structureThree people do everything. Task variety keeps everyone engaged.
Stage 2SpecialisationCooks, dishwashers and waiters appear, and with them the first coordination problem. Everything still depends on the owner being there.
Stage 3First management layerThe owner’s span is exceeded. Supervisors are added, and he becomes a manager of managers.
Stage 4Second layerA service pantry adds complexity, and the dilemma: how to gain the benefits of standardisation without losing the warmth that made the place special?

The sensible design is a functional structure with clear roles, spans of no more than about 8–10, unity of command, an operations manager as the coordination bridge, written procedures where precision matters (portions, cost) and discretion where warmth matters (guests). It brings its own risks: bureaucracy creep, silos and regulars missing the owner.

Three routes to consistency

The owner’s real choice is how to make people behave consistently. Prof. Jha offered three routes:

  1. Formalisation: written rules. Powerful but brittle: even a thousand-page manual can’t foresee every exception.
  2. Socialisation: teach the philosophy, then trust judgement. Founders meeting every recruit, a new joiner receiving the founder’s book, or the Udupi restaurant tradition, consistent through shared culture rather than manuals.
  3. Professional socialisation: the profession shapes behaviour. A chartered accountant checks everything because that is what accountants are trained to do.

For a restaurant built on warmth, socialisation preserves more than rules.

Rules also accumulate. Structural inertia means a structure resists change: the world shifts, rules are written to catch up, the world shifts again, and the old rules stay, because adding a rule is easier than removing one. Hold that thought for Session 4.

Why trusted work needs fewer supervisors

Three ideas explain why some organisations manage with far less supervision:

  • McGregor’s Theory X and Theory Y. Theory X assumes people dislike work and must be controlled. Theory Y assumes people will work hard if the work is fulfilling, so poor effort signals a badly designed job, not a lazy worker.
  • Toyota’s jidoka. Any operator can pull the andon cord and stop the whole line on spotting a defect: authority pushed to the front line.
  • Ouchi’s Theory Z. Long-term employment, collective decisions and strong shared values, as in post-war Japanese firms, let people work without being watched.

The common thread: where people are trusted and the work is meaningful, you need less hierarchy.

Acme and Omega: structure is a choice

Two electronics firms with the same parent, product, market and city, and opposite structures.

AcmeOmega
TypeMechanisticOrganic
SpecialisationHigh; narrow rolesLow; fluid roles
FormalisationHigh; memos and documented proceduresVery low; no org charts, no memos
DecisionsThe general manager decidesA management team decides together
Profit todayHigher; wins more contractsLower, but more adaptable

Then a customer split an urgent order for 100 prototype units between the two firms, and both hit the same problems.

AcmeOmega
How problems were handledEvery issue went to the general manager by memo, bounced between departments, and set off a turf war resolved only by escalationDepartment heads met in one room and decided together; engineering suggested a substitute part; the team caught a design error because everyone owned the whole product
DeliveryLate: 18–21 days against a 14-day targetOn time: 11 days
Quality10 defective unitsNo defects

Same problems, opposite handling. The structure decided the outcome.

Don’t crown Acme for being more profitable

Acme’s mechanistic design is a perfect fit for a stable, high-volume market. Change the context (a new technology, a supply shock, a custom order) and Omega’s flexibility becomes the decisive advantage. Always ask: profitable in what environment?

Centralisation also makes the boss the only person who sees the whole product, so everyone expects him to know everything. Prof. Jha’s image: the qualities once attributed to God (all-knowing, all-powerful, everywhere) quietly transfer to the hierarchical superior. No real person is like that, which is why over-centralised structures are fragile. He also drew a useful line between delegation (prepare people, give resources, praise and correct) and relegation (just offloading work onto someone else).

Exploration, exploitation and ambidexterity

This is the idea the whole session built towards.

Two capabilities, two structures

Mechanistic structureOrganic structure Exploitation
AcmeGood fitDo known work cheaper, better and faster. Suits cost leadership.
Poor fitFlexibility costs money you don’t need to spend on routine work.
Exploration
Poor fitA machine that executes but can’t think: new ideas die in the hierarchy.
OmegaGood fitInvent and do new things. Suits differentiation and innovation.

Historically, firms chose one. Then Toyota’s cuts to changeover time and its just-in-time system made mass customisation possible: variety and low cost. Today firms are expected to be ambidextrous: to exploit and explore at the same time.

Knowing you need both doesn’t make it happen. Career paths, rewards and current revenue all sit in exploitation, so the best people and the money flow there and exploration starves. Nor can you simply merge an Acme with an Omega: Acme people feel rudderless under Omega’s style, and Omega people feel suffocated under Acme’s. Many mergers fail on exactly this kind of cultural mismatch.

Beyond the notes: how firms make ambidexterity work

Research describes two common approaches. Structural ambidexterity (O’Reilly and Tushman) separates exploratory units from the core business, with their own processes and culture, linked only at the top by a leadership team that protects their budgets. Contextual ambidexterity (Gibson and Birkinshaw) builds a culture in which individuals divide their own time between improving today’s work and exploring new ideas. Either way, exploration needs ring-fenced money and senior protection, or the core business will absorb it.

The investor-versus-employee paradox

Asked where they’d rather work, almost everyone in class picked Omega. Asked where they’d invest a prize of a lakh, most picked Acme. When our preferences as employees and as investors point in opposite directions, the investor usually wins, so the world drifts towards Acme, and work-life balance erodes. A structural choice with a social consequence.

Where this leads

Acme’s tidy structure was perfect for a stable market. But the things that make a structure efficient (fixed roles, settled rules, a comfortable way of working) are exactly what make it resist change when the world shifts. The structure built in this session becomes the thing that fights you in the next.