Portfolio selection is the adult conversation: given strategy, capacity, and risk appetite, which work do we start, continue, defer, or kill? Everything-is-a-priority is not a portfolio. It is a queue with better fonts.
Criteria before charisma
Write the selection criteria before the first passionate sponsor enters the room. Strategic fit, expected benefit, cost, risk, capacity demand, compliance must-do, and time sensitivity are a typical set. Weight them in daylight. If you invent the scoring after you know who is in the room, you have built a justification engine.
Must-do compliance work still belongs in the model. It consumes the same scarce people. Pretending a regulatory project is 'outside the portfolio' is how you over-commit the architects and then act shocked when the strategic bet slips.
A score is a decision aid, not a priest. Two initiatives can tie. Then you talk about sequencing, option value, and political reality — out loud. A model that always 'proves' the CEO's favourite is not analysis. It is theatre with a spreadsheet.
Capacity truth and a balanced mix
The constraint is almost never money in the abstract. It is the same twelve people everyone wants in Q3. A portfolio that approves work without a capacity view is signing fictional start dates. Show the load. If the integration guild is at 140%, the next approval is a decision to delay something else — or to buy capacity, not to 'go faster.'
Balance the mix: short and long horizon, run and change, high risk / high option value versus dull-but-necessary, customer-facing versus plumbing. A portfolio of only shiny bets starves the platform. A portfolio of only plumbing starves the strategy. Balance is a design choice, not an accident of who shouted.
Dependencies across the portfolio matter. Three programs that all assume the same data platform in the same quarter are not three strategies. They are one collision. Selection includes sequence. 'Yes, later' is a legitimate portfolio decision. It just does not feel like a ribbon-cutting.
Start, stop, continue — the verbs that prove you have a portfolio
Selection is not an annual ceremony. It is a rhythm. New information arrives: a benefit assumption dies, a market moves, a team is stuck. The portfolio should be able to stop or reshape work without treating it as a personal failure. Sunk cost is not a strategy.
Continuing everything to 'keep people busy' is how you fund ghosts. Idle capacity is cheaper than a project that cannot produce a benefit. Redeploy the people. The organisation will invent a reason to keep the ghost if you do not offer a dignified stop.
Document the decision: why this, why not that, what you will review, and what would make you stop. When a sponsor asks six months later why their idea is on the shelf, 'we discussed it in March' is not an answer. A one-page decision record is.
Summary & next steps
Select with published criteria, capacity truth, and a balanced mix. Must-do work still consumes the same people. The portfolio's rare skill is stopping, not starting.
Next: benefits realisation — the part that continues after go-live, when the project team has gone to the next kick-off with the good biscuits.
Practice this
Stakeholder Register
Turn theory into a real deliverable. Generate a stakeholder register with the AI tool and adapt it to your own project.
Open the generator