What a project is and what project leads do

Paying the staff every month and running this year’s college tech fest both need people, money and a plan, yet only one of them is a project. The difference decides who leads the work, how it is funded and when it is finished, and it matters at scale: India’s central government alone monitors 1,981 ongoing infrastructure projects, each with its own budget, team and end.

Below you will sort twelve pieces of work into projects, operations, programmes and portfolio decisions with a program that shows its reasons, see who is accountable for what in a plan-driven team and in a Scrum team, and judge a finished project the way the profession now does: by the value it delivered, not only by its date and its bill.

A project has an end and a unique context; operations repeat

The tech fest starts when the organising committee forms and ends when the last stall is packed away. It has never happened in exactly this way before: this year’s theme, venue, sponsors and volunteers are its own. Payroll is the opposite. It runs every month in the same way, and the organisation would stop working if it ended.

PMI’s PMBOK Guide, as its PMP Examination Content Outline quotes it, defines a projectproject: Temporary work in a unique context, undertaken to create value: it has a planned start and end and makes a result that has not been made in exactly this way before. as “a temporary initiative in a unique context undertaken to create value”. Temporary means the work has a beginning and an end; unique context means its own goals, conditions, approach and people. Work that repeats to keep the organisation running is operationsoperations: The work that repeats to keep an organisation running, such as payroll, support or backups. It has no planned end; finished projects usually hand their results over to it.: payroll, the helpdesk, the nightly backups.

The program below puts four questions to twelve invented pieces of work, in a fixed order, and the first yes decides. Read its rules in the first lines, then the reason it prints for each item:

"""Sorts twelve invented pieces of work into project, operations, programme or portfolio decision.

Each item answers four questions with yes or no, and the first rule that applies decides:
  1. Does it choose which work to fund, against the organisation's strategy?   -> portfolio decision
  2. Does it run several related projects together for a benefit none gives alone? -> programme
  3. Is it the same work, done again and again?                                  -> operations
  4. Does it have a planned end and a result that has not been made before?     -> project
"""

from collections import Counter

# (work, chooses what to fund, groups related projects, repeats, has an end and a new result)
ITEMS = [
    ("Run this year's college tech fest", False, False, False, True),
    ("Pay the staff every month", False, False, True, False),
    ("Add UPI AutoPay to a food app", False, False, False, True),
    ("Upgrade the hostel Wi-Fi", False, False, False, True),
    ("City metro plan: 3 lines and a depot", False, True, False, True),
    ("Yearly budget round for new work", True, False, True, False),
    ("Answer helpdesk tickets", False, False, True, False),
    ("Move a data centre to a new site", False, False, False, True),
    ("Bank's new app, branches and call centre", False, True, False, True),
    ("Stop two projects, start one", True, False, False, False),
    ("Back up the servers every night", False, False, True, False),
    ("Write the first run of a new course", False, False, False, True),
]


def classify(funds, groups, repeats, new_result):
    """Returns the kind of work and the rule that decided it."""
    if funds:
        return "portfolio decision", "chooses what to fund"
    if groups:
        return "programme", "related projects, one benefit"
    if repeats:
        return "operations", "the same work again"
    if new_result:
        return "project", "an end and a new result"
    return "unclear", "answer the questions again"


counts = Counter()
print(f"{'Work':<42} {'Kind':<19} Why")
for work, funds, groups, repeats, new_result in ITEMS:
    kind, why = classify(funds, groups, repeats, new_result)
    counts[kind] += 1
    print(f"{work:<42} {kind:<19} {why}")

print()
for kind in ("project", "operations", "programme", "portfolio decision"):
    print(f"{kind}: {counts[kind]}")

Work                                       Kind                Why
Run this year's college tech fest          project             an end and a new result
Pay the staff every month                  operations          the same work again
Add UPI AutoPay to a food app              project             an end and a new result
Upgrade the hostel Wi-Fi                   project             an end and a new result
City metro plan: 3 lines and a depot       programme           related projects, one benefit
Yearly budget round for new work           portfolio decision  chooses what to fund
Answer helpdesk tickets                    operations          the same work again
Move a data centre to a new site           project             an end and a new result
Bank's new app, branches and call centre   programme           related projects, one benefit
Stop two projects, start one               portfolio decision  chooses what to fund
Back up the servers every night            operations          the same work again
Write the first run of a new course        project             an end and a new result

project: 5
operations: 3
programme: 2
portfolio decision: 2

Five of the twelve are projects and three are operations. Two items need the earlier questions. The yearly budget round repeats, yet it is not operations, because what it does is choose which work to fund; and the city’s metro plan has an end, yet it is more than one project, because it runs several projects together. The next section names both.

Programmes and portfolios group work for benefits and strategy

A city that builds one metro line has a project. A city that builds three lines, a depot and a fare-card system, and needs all of them working together before people ride in large numbers, has a programmeprogramme: Related projects and other work managed together for a benefit that none of them would deliver alone, such as a city's metro lines, depot and fare system.: related projects managed together for a benefit that none of them would deliver alone. The outline adds that projects can be stand-alone efforts or part of a portfolio or a programme.

A portfolioportfolio: All the programmes, projects and operations an organisation chooses to fund for its strategy; portfolio decisions start, pause and stop work. sits above both. It is all the work an organisation chooses to fund for its strategy: programmes, stand-alone projects and the operations they hand over to. A portfolio decision chooses, starts, pauses or stops work, which is why the yearly budget round and “stop two projects, start one” were sorted as portfolio decisions. Step through one organisation’s work, from the choice at the top to the projects at the bottom:

The last step shows the hand-over that links the levels. A finished project becomes operations, and the benefit the portfolio paid for starts only once that operation runs. The governance module of this track returns to portfolios and programmes with the tools organisations use to choose between them.

Who leads depends on the way the team works

In a plan-driven project, often called predictive, one project managerThe person an organisation makes accountable for leading a project to its objectives: the plan, the team's coordination, risks, reporting and the hand-over. is usually accountable for leading the work to its objectives. A sponsor, the senior person who wants the result and funds it, gives the project its authority and takes the big decisions; the project manager plans, coordinates, reports and raises what the sponsor must decide.

The 2020 Scrum Guide works differently. A Scrum Team is one Scrum Master, one Product Owner and Developers, typically ten or fewer people, with no sub-teams or hierarchies, and the Guide defines three accountabilities. The Product Owner is accountable for maximising the value of the product and for managing the Product Backlog, including its order. The Developers are accountable for their plan for each Sprint, for quality through a Definition of Done and for adapting their plan each day. The Scrum Master is accountable for establishing Scrum and for the team’s effectiveness. The Guide names no project manager at all.

Here is the same kind of feature, a new payment option, delivered by two invented teams: a vendor team in Pune working to a fixed plan and contract, and a Scrum team in Bengaluru.

The work of managing a project does not disappear in Scrum. Budgets, contracts, dependencies on other teams and reports to stakeholders still need someone; in a Scrum team they are shared between the accountabilities and the people around the team, and a job title such as “delivery lead” is whatever an organisation decides it is, not a role any standard defines.

Success means value delivered, not only scope, time and cost

For decades a project counted as a success when it delivered its scope on time and within budget. In 1999 Roger Atkinson, writing in the International Journal of Project Management, called cost, time and quality “two best guesses and a phenomenon” and argued for other criteria. PMI’s 2026 outline now says the same in its own way: it reframes success from schedule, budget and scope to the value stakeholders receive and the outcomes they wanted, describing a project perceived to have delivered value “worth the effort and expense”.

Follow one invented feature from its idea to its benefit. Watch where the project ends and where its value is measured:

The AutoPay project met its date and its budget, and three months after hand-over AutoPay carried 4 % of repeat orders against the 20 % its business case promised. On the old measures it succeeded; on the measure it was funded for, it has not yet. That is why the business case names the benefit and why someone, often the sponsor or the business owner, must be named before hand-over to measure it after the project team has gone.

1,981 central projects show the scale of project work in India

One government’s own monitoring shows how much project work a country runs. The figures below come from India’s central monitoring, and the program after them checks what the published totals mean for a single project.

The program works out the averages with exact decimal arithmetic, so its last digits can be trusted:

"""What one month's MoSPI flash report on central infrastructure projects (Rs 150 crore and above) says on average.

The inputs are the figures of the press release the lesson cites: 1,981 ongoing projects with a total revised cost of
Rs 42.78 lakh crore, Rs 20.36 lakh crore spent so far, and 814 of the projects in the mega class (Rs 1,000 crore and
above). Money is worked in crore with the decimal module (1 lakh crore = 100,000 crore), and rounded half up.
"""

from decimal import ROUND_HALF_UP, Decimal

LAKH_CRORE = Decimal(100_000)  # crore in one lakh crore

projects = 1981
mega = 814
revised_cost = Decimal("42.78") * LAKH_CRORE  # crore
spent = Decimal("20.36") * LAKH_CRORE  # crore


def r(value, places="0.1"):
    return value.quantize(Decimal(places), rounding=ROUND_HALF_UP)


average = revised_cost / projects
print(f"Revised cost of all projects: {r(revised_cost, '1'):,} crore")
print(f"Average revised cost per project: {r(average, '1'):,} crore")
print(f"Share of the revised cost spent so far: {r(spent / revised_cost * 100, '0.01')} %")
print(f"Mega projects: {mega} of {projects:,} ({r(Decimal(mega) / projects * 100)} %)")
print(f"Major projects (Rs 150 to 1,000 crore): {projects - mega:,} ({r(Decimal(projects - mega) / projects * 100)} %)")

Revised cost of all projects: 4,278,000 crore
Average revised cost per project: 2,160 crore
Share of the revised cost spent so far: 47.59 %
Mega projects: 814 of 1,981 (41.1 %)
Major projects (Rs 150 to 1,000 crore): 1,167 (58.9 %)

Two in five of the projects, 41.1 %, are in the mega class, and 47.59 % of the revised cost had been spent, the same share the release itself gives, which is a quick check that the program read the figures correctly.

Practise: sort, assign and judge

Sort, assign and judge a few more cases. Each answer is explained as soon as you check it, and the concepts you find hardest come back first:

Interview questions

Warm-up (fresher to 2–5 years): what makes a piece of work a project rather than operations?

A project is temporary and works in a unique context: it has a planned start and end, and it makes something that has not been made in exactly this way before, such as a new feature, an event or a move to a new office. Operations are the work that repeats to keep the organisation running, such as payroll, support or backups, and they have no planned end. Size does not decide: a weekend upgrade is a project, and running a railway every day is operations. The two are linked, because a finished project usually hands its result over to operations, and the benefit it was funded for starts only when that operation runs. In an interview, give one example of each from your own work and say where one became the other.

Two harder questions follow, about judging a finished project and about leading without a project manager. Say your answer out loud first, then compare it with the first step:

Key takeaways

• A project is temporary and unique, with a planned end and a new result, while operations repeat to keep the organisation running; size decides neither.
• A programme runs related projects together for a benefit none delivers alone, and a portfolio decides which work to fund, start or stop for the strategy.
• Plan-driven projects usually have one accountable project manager, while the 2020 Scrum Guide names no project manager and gives its decisions to the Product Owner, the Developers and the Scrum Master.
• A project ends at hand-over, but its success is judged by the value it delivered, so someone must own measuring the benefit afterwards.
• India’s central monitoring alone counts 1,981 ongoing infrastructure projects, and a short program turns the published totals into an average for one project.