Barnaby Robson

Mental Model · Decision & Risk · Markets & Finance

Cost-Benefit Analysis

Plot initiatives by benefit vs cost to spot quick wins, staged bets and time-sinks.

After Managerial decision-making (simple cost–benefit prioritisation)

This is the lightweight, visual form of cost–benefit analysis. Instead of detailed NPV spreadsheets, you estimate benefit and effort/cost on consistent scales and place items on a 2×2 grid. It’s ideal for backlog grooming, roadmap picks and operational improvements where speed and clarity matter.

Cost-Benefit Analysis matrix showing Prioritise, Consider, Maybe, Don't do.

How it works

Axes

  • Y: Benefit / Reward – expected impact on the goal (revenue, risk reduction, time saved, customer value).
  • X: Cost / Effort – money, people, time, risk, complexity.

Quadrants

  • High benefit, low cost — Prioritise: quick wins and high-leverage moves.
  • High benefit, high cost — Consider: stage, prototype, or seek funding; de-risk before full commit.
  • Low benefit, low cost — Maybe: do only if dependencies are simple or as filler work.
  • Low benefit, high cost — Don’t do: decline or rethink.

Use-cases

Backlog grooming and sprint planning.

Portfolio and roadmap selection.

Ops improvements and process changes.

Compliance and risk mitigations triage.

Step-by-step

  1. Define benefit and cost scales – e.g. 1–5 or T-shirt sizes; write what each level means.

  2. Score together – cross-functional group estimates both axes; capture assumptions.

  3. Plot the items – place each initiative on the grid; cluster similar ones.

  4. Decide –

    • Prioritise the top-left first.

    • For top-right, break into milestones or run a small proof-of-value.

    • Bottom-left only if they unblock others.

    • Bottom-right is a no unless the idea can be redesigned.

  5. Add guardrails – owner, success metric, review date; re-plot when facts change.

Pitfalls & Cautions

Hand-wavy scales – define benefit and cost levels up front or estimates drift.

Optimism bias – inflate benefit, deflate effort; counter with ranges and a quick premortem.

One metric only – if benefits are mixed (revenue + risk), agree a composite or use separate grids.

Crowding the middle – force choices by limiting how many can sit in “Consider” or “Maybe”.

Never upgrading evidence – for big bets in “Consider”, follow up with a proper NPV/EV analysis.