02 · Portfolio Management Basics¶
A portfolio is the full set of projects and programs an organisation is funding at once — often unrelated to each other, competing for the same pool of money and people. Portfolio management answers one question the project or program level cannot: given limited capacity, which of these should we be doing at all? Get this wrong and your best project managers deliver flawless projects that nobody should have funded.
Portfolio vs. program vs. project (recap)¶
| Project | Program | Portfolio | |
|---|---|---|---|
| Question answered | How do we deliver this? | How do these related things add up to a benefit? | Which things should we fund? |
| Scope | Fixed | Related set | Organisation-wide, may be unrelated |
| Owner | Project manager | Program manager | Portfolio manager / PMO / executive committee |
A scoring model for prioritisation¶
Comparing a compliance project to a growth project to a cost-cutting project on "which is more important" is a values argument dressed up as analysis unless you have a shared scoring model. A simple weighted model:
| Criterion | Weight | What it measures |
|---|---|---|
| Strategic alignment | 30% | Does this map to a named strategic objective this year? |
| Financial value (NPV/ROI band) | 25% | Expected return relative to cost |
| Risk (inverted — lower risk scores higher) | 15% | Likelihood of failure or major overrun |
| Urgency / cost of delay | 15% | What happens if we wait a year? |
| Resource feasibility | 15% | Can we actually staff it without gutting something else? |
Each criterion is scored 1–5 by a review board, weighted, and summed to a 100-point scale.
Worked example: scoring four candidate projects¶
| Project | Alignment (30%) | Value (25%) | Risk (15%, inverted) | Urgency (15%) | Feasibility (15%) |
|---|---|---|---|---|---|
| A: Core platform migration | 5 | 3 | 2 | 3 | 3 |
| B: New mobile app | 3 | 5 | 3 | 4 | 4 |
| C: Regulatory compliance update | 5 | 2 | 5 | 5 | 5 |
| D: Internal tooling refresh | 2 | 3 | 4 | 2 | 5 |
Weighted score = Σ(criterion score × weight) × 20 (to put on a 100 scale since raw scores are out of 5):
A = (5*0.30 + 3*0.25 + 2*0.15 + 3*0.15 + 3*0.15) * 20
B = (3*0.30 + 5*0.25 + 3*0.15 + 4*0.15 + 4*0.15) * 20
C = (5*0.30 + 2*0.25 + 5*0.15 + 5*0.15 + 5*0.15) * 20
D = (2*0.30 + 3*0.25 + 4*0.15 + 2*0.15 + 5*0.15) * 20
Verified with python3:
| Project | Weighted score |
|---|---|
| A | 69.0 |
| B | 76.0 |
| C | 85.0 |
| D | 60.0 |
C (compliance) clearly outranks the others despite lower financial value, because its risk, urgency, and feasibility scores are high — this is the model correctly capturing that a compliance deadline with heavy non-compliance penalties outranks a nice-to-have growth bet, even though a value-only ranking would have put B first.
The balance: not just the top-N by score¶
Ranking alone produces a portfolio that's all one type of bet. A mature portfolio review balances the ranked list against a target mix:
| Category | Target % of budget | Current % | Gap |
|---|---|---|---|
| Run-the-business (keep-the-lights-on) | 40% | 55% | Over by 15pts |
| Grow-the-business (new revenue) | 35% | 20% | Under by 15pts |
| Transform-the-business (strategic bets) | 25% | 25% | On target |
A portfolio 55% run-the-business is starving growth even if every individual score is defensible — the review board's job is to notice the mix problem the scoring model alone won't surface, and deliberately fund a lower-scoring growth project over a higher-scoring maintenance one to correct the balance.
Capacity-constrained selection¶
With a ranked list and a fixed budget, selection is a constrained optimisation, not just "fund the top of the list until money runs out" — that greedy approach can leave large amounts unspent or lock out a high-value project that needs a big-but-not-huge slice.
| Project | Score | Cost ($k) | Score per $k |
|---|---|---|---|
| C | 85.0 | 300 | 0.283 |
| B | 76.0 | 900 | 0.084 |
| A | 69.0 | 600 | 0.115 |
| D | 60.0 | 150 | 0.400 |
With a $1,000k budget: funding by raw score (C, B) uses $1,200k — over budget, so B is dropped, leaving $700k unspent capacity idle. Funding by score-per-dollar instead (D, C, A) uses $1,050k — still slightly over. The actual decision a review board makes: fund D + C + A minus a $50k trim on A's scope, capturing three initiatives instead of two, and getting more total strategic value per dollar than chasing the single highest-scoring project.
Portfolio dashboard¶
| Project | Score | Status | Budget used | Schedule | Risk trend |
|---|---|---|---|---|---|
| A | 69.0 | Active | 40% | On track | Stable |
| C | 85.0 | Active | 15% | On track | Improving |
| D | 60.0 | Active | 60% | 2 weeks late | Stable |
| B | 76.0 | Deferred to next cycle | — | — | — |
How It Actually Works¶
Capacity-constrained selection is a version of the knapsack problem: you
have a fixed capacity (budget, or full-time-equivalent headcount) and a set
of candidate projects each with a "weight" (cost) and a "value" (strategic
score), and the objective is to select the subset maximizing total value
without exceeding capacity. Picking strictly top-N by score, the naive
approach, is provably suboptimal against a hard capacity constraint — the
correct heuristic is to rank by value density, score / cost, not raw
score, because a lower-scoring but cheaper project can free capacity for two
more projects than one expensive high-scorer would allow, producing more
total value from the same budget. This is exactly analogous to Level 2
Module 5's FPIF share-ratio math and Module 2's cost-slope crashing
decisions — all three are constrained-optimization problems where the naive
"pick the biggest number" heuristic loses to "pick the best ratio" heuristic
once a hard constraint is binding. True optimal knapsack selection is
computationally expensive at scale (NP-hard), which is the actual technical
reason portfolio tools use the value-density heuristic rather than solving
for a global optimum — it's a fast, usually-near-optimal approximation, not
a shortcut taken out of laziness.
Exercise¶
Score two new candidate projects using the weighted model above (alignment 30%, value 25%, risk 15% inverted, urgency 15%, feasibility 15%): Project E (alignment 4, value 4, risk 3, urgency 2, feasibility 4) and Project F (alignment 2, value 5, risk 1, urgency 5, feasibility 2).
- Compute each weighted score on the 100-point scale and verify your
arithmetic with
python3 -c. - Given a portfolio already spending 55% on run-the-business against a 40% target, argue for or against funding whichever of E/F scores higher, using the category-mix table as part of your reasoning, not just the score.
- List one thing the scoring model in this module cannot tell you that a human review board still has to judge.