Programmes, portfolios and OKRs

An invented edtech company funded ten initiatives for its six teams and started all ten at once. All ten were live by month 7, yet waiting for them cost 4,247 thousand rupees of benefit, 522 thousand more than it had to.

Projects deliver outputs. Programmes and portfolios decide which outputs are worth making, together and in which order, and OKRs tie those choices to the strategy so everyone can see what the work is for.

A project delivers, a programme coordinates, a portfolio chooses

The company sells online courses to working adults and college students. Its training portal for corporate customers is a project: one output with its own scope, plan and team. 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. More in “What a project is and what project leads do” joins related work for a benefit none of it brings alone, here corporate renewals, which need the portal, team dashboards and a sales push together. A portfolioportfolio: All the programmes, projects and operations an organisation chooses to fund for its strategy; portfolio decisions start, pause and stop work. More in “What a project is and what project leads do” is everything the company funds for its strategy. Each level asks its own question:

Programmes are judged by benefits, and portfolios by what the whole set returns for the teams it uses. PMI keeps separate credentials for both levels: the PgMP for programmes, for four to seven years of experience, and the PfMP for portfolios, for eight or more. Step from the strategy down to one project, then follow the benefits back up:

The arrow back up matters most. Portfolios that never hear about benefits can only count outputs and dates.

A portfolio is a capacity choice: start less, finish more

The company’s ten initiatives need 31 team-months between them, and each has a cost of delaycost of delay: What one period of lateness costs for one item: money not earned, a penalty incurred or a risk left open. Dividing it by the item's duration sequences a queue. More in “Lean flow: waste, Little's law and constraints”, the benefit lost for each month it is not live. The program compares starting all ten at once with six in progress, ordered by cost of delay per month of work:

"""Ten initiatives, six teams: what starting everything at once costs an invented edtech company's portfolio.

The company sells online courses to working adults and college students. Its portfolio board has funded ten
initiatives for the year. Each needs one team for a number of months, and each has a cost of delay: the benefit, in
thousand rupees a month, that the company gives up for every month the initiative is not yet live. The company has
six teams.

Policy A starts all ten at once and shares the six teams among them equally (a team cannot split further than one
initiative at a time, so each unfinished initiative gets 6/10 of a team, then more as others finish). Policy B keeps
at most six initiatives in progress, one team each, and starts them in order of cost of delay divided by months of
work, highest first. Neither policy pays any switching cost, which flatters policy A.

The program prints the month each initiative goes live under both policies, then the month of the first benefit, how
many are live by month 3, and the total benefit given up from the start of the year until each one is live.
"""

from fractions import Fraction

TEAMS = 6

# (id, initiative, months of work for one team, cost of delay in thousand rupees a month)
INITIATIVES = [
    ("I01", "Corporate training portal", 6, 210),
    ("I02", "Offline downloads in the app", 3, 150),
    ("I03", "Subtitles in Hindi and Tamil", 2, 110),
    ("I04", "Instalment payments at checkout", 2, 90),
    ("I05", "Mentor chat for learners", 4, 120),
    ("I06", "Course recommendations", 4, 70),
    ("I07", "Accessibility fixes (WCAG 2.2 AA)", 1, 40),
    ("I08", "New analytics warehouse", 5, 45),
    ("I09", "Referral rewards", 1, 30),
    ("I10", "Team dashboards for corporate clients", 3, 60),
]


def start_everything(items, teams):
    """All items in progress at once; the teams are shared equally, at most one team per item."""
    left = {i[0]: Fraction(i[2]) for i in items}
    now, done = Fraction(0), {}
    while left:
        share = min(Fraction(1), Fraction(teams, len(left)))
        step = min(work / share for work in left.values())
        now += step
        for key in list(left):
            left[key] -= step * share
            if left[key] == 0:
                done[key] = now
                del left[key]
    return done


def limit_in_progress(items, teams):
    """At most `teams` items in progress, one team each, started in the given order."""
    free_at = [Fraction(0)] * teams
    done = {}
    for key, _name, months, _cod in items:
        t = min(range(teams), key=lambda k: free_at[k])
        free_at[t] += months
        done[key] = free_at[t]
    return done


def main():
    order = sorted(INITIATIVES, key=lambda i: Fraction(i[3], i[2]), reverse=True)
    a = start_everything(INITIATIVES, TEAMS)
    b = limit_in_progress(order, TEAMS)
    print(f"{'id':<4}{'initiative':<39}{'delay/mo':>9}{'months':>7}{'live A':>8}{'live B':>8}")
    for key, name, months, cod in INITIATIVES:
        print(f"{key:<4}{name:<39}{cod:>9}{months:>7}{float(a[key]):>8.1f}{float(b[key]):>8.1f}")
    print()
    for label, done in (("A, all ten started at once", a), ("B, six in progress, by delay per month of work", b)):
        first = min(done.values())
        by3 = sum(1 for t in done.values() if t <= 3)
        given_up = sum(cod * done[key] for key, _n, _m, cod in INITIATIVES)
        print(f"Policy {label}:")
        print(f"  first benefit in month {float(first):.1f}, {by3} of 10 live by month 3, "
              f"all live by month {float(max(done.values())):.1f}")
        print(f"  benefit given up until each is live: {float(given_up):,.0f} thousand rupees")
    saved = sum(cod * (a[k] - b[k]) for k, _n, _m, cod in INITIATIVES)
    print(f"Policy B gives up {float(saved):,.0f} thousand rupees less, with the same six teams and the same work")


if __name__ == "__main__":
    main()

id  initiative                              delay/mo months  live A  live B
I01 Corporate training portal                    210      6     7.0     6.0
I02 Offline downloads in the app                 150      3     4.0     3.0
I03 Subtitles in Hindi and Tamil                 110      2     3.0     2.0
I04 Instalment payments at checkout               90      2     3.0     2.0
I05 Mentor chat for learners                     120      4     5.0     4.0
I06 Course recommendations                        70      4     5.0     6.0
I07 Accessibility fixes (WCAG 2.2 AA)             40      1     1.7     1.0
I08 New analytics warehouse                       45      5     6.0     7.0
I09 Referral rewards                              30      1     1.7     2.0
I10 Team dashboards for corporate clients         60      3     4.0     5.0

Policy A, all ten started at once:
  first benefit in month 1.7, 4 of 10 live by month 3, all live by month 7.0
  benefit given up until each is live: 4,247 thousand rupees
Policy B, six in progress, by delay per month of work:
  first benefit in month 1.0, 5 of 10 live by month 3, all live by month 7.0
  benefit given up until each is live: 3,725 thousand rupees
Policy B gives up 522 thousand rupees less, with the same six teams and the same work

Six at a time gives up 3,725 thousand rupees instead of 4,247, and its first initiative is live in month 1.0, not 1.7. Every initiative needs the same team-months either way; starting all ten only spreads each team thinner, and the program does not even charge for switching between them. The capacityportfolio capacity: The work a portfolio's teams can do at once; starting more initiatives than that spreads the same teams thinner and adds waiting, not work done. of a portfolio is the work its teams can do at once, and starting beyond it adds waiting, not work done.

Order is the second choice. Most valuable first starts the largest costs of delay; shortest first starts the quickest items. Both are compared below on the same ten.

OKRs connect the portfolio to the strategy

OKRsOKRs: Objectives and key results: a few ambitious objectives, each with about three measurable key results graded from 0.0 to 1.0, public across the organisation and kept apart from employee evaluations., objectives and key results, turn a strategy into a few goals anyone can check. Google’s re:Work guide describes ambitious objectives, each with about three key resultskey result: A measurable result that shows progress towards an objective, such as raising a course completion rate from 18 % to 30 %, graded at the end of the period. graded from 0.0 to 1.0, and expects an average of 0.6 to 0.7. One of the company’s OKRs has the objective “make our courses the first choice for working adults learning data skills” and three key results:

Each row names a measure, its value today and a target, so the quarter’s grade can be worked out. OKRs are public, so every team sees what the others are working towards. Google’s guide adds that OKRs are not employee evaluations and not a shared to-do list. Portfolios use OKRs to say no: work that moves no key result waits.

Two related tools often come up beside OKRs. Scrum.org’s Evidence-Based ManagementEvidence-Based Management: A Scrum.org guide for deciding with evidence: it measures the value an organisation delivers in four key value areas and moves towards its goals in small, measured steps. guide measures value in four key value areaskey value area: One of the four areas in which Evidence-Based Management measures value: current value, unrealised value, time to market and ability to innovate.: current value, unrealised value, time to market and ability to innovate. SAFe’s Lean Portfolio Management applies Lean thinking to funding and governing a portfolio, as the scaling lesson shows.

Filling six teams five ways, and pricing a slip Pro: opens with a Pro, Premium or Ultimate pass

Writing and grading OKRs, and measuring value with EBM Pro: opens with a Pro, Premium or Ultimate pass

Practise

These questions use the company’s levels, its ten initiatives and its first OKR:

Interview questions

Warm-up: how does a programme differ from a portfolio?

A programme coordinates related projects and other work so that together they deliver a benefit none of them would deliver alone, such as corporate renewals that need a portal, dashboards and a sales push. A portfolio is everything an organisation funds for its strategy, related or not, and its job is choosing: which work gets teams, which waits, which stops. A programme is judged by the benefits it realises; a portfolio by the value the whole set returns for the capacity it uses.

Three harder questions ask for decisions with numbers:

Key takeaways

• A project delivers an output, a programme coordinates related work for a benefit, and a portfolio chooses which work deserves the organisation’s teams.
• A portfolio is a capacity choice: six teams on six initiatives finished sooner than on ten, and gave up 522 thousand rupees less benefit.
• OKRs name a few ambitious objectives with measurable key results, graded from 0.0 to 1.0 with about 0.6 to 0.7 expected, public and kept apart from appraisals.
• Evidence-Based Management measures value in four key value areas: current value, unrealised value, time to market and ability to innovate.