Quantitative Methods

Multi-criteria Decision Analysis and Weighting Things That Do not Compare

Turn a vendor scorecard from a number-laundering exercise into an honest statement of what you are trading against what — and find the one question the whole decision actually turns on.

  • Advanced
  • 14 min total
  • 13 chapters

What decision this helps you make: Which option to choose when each one wins on a different dimension, and — more usefully — which single trade-off the choice really depends on, so the argument can be had about that instead of about the spreadsheet.

What this topic is

Multi-criteria decision analysis is a family of methods for choosing between options that are measured in different units — dollars, weeks, risk ratings, fit scores. The common form scores each option on each criterion, normalises the scores to a shared scale, applies weights, and adds them up. The value is not the total. The value is that the trade-offs become explicit numbers someone can disagree with, instead of implicit judgements nobody can locate.

Why it matters

Every significant selection decision — vendor, site, acquisition target, project portfolio — is multi-criteria, and it will be resolved by weights whether or not anyone writes them down. Unwritten weights cannot be audited, cannot be argued with, and drift toward whoever speaks last. Written weights, done properly, do something better than pick a winner: they identify the small number of trade-offs the answer is sensitive to, and reduce a fourteen-criterion argument to one question the business can actually answer.

Who should learn it

Anyone running a selection process, sitting on a committee that scores options, or trying to work out why a scorecard produced a winner that feels wrong.

What you will understand

  • Why a weight is meaningless without the range of the criterion, and how swing weighting fixes it
  • How to run a sensitivity analysis that reduces a many-criteria decision to the one trade-off it turns on
  • When a weighted sum is the wrong shape entirely — hurdles, correlated criteria, and preference dependence
  • How to use the method to locate disagreement rather than to manufacture a winner

Prerequisites

Common misconception

"Cost is 40% of the decision" is a complete statement of a preference. It is not a statement at all until you say what range of cost is on the table. Forty percent weight on a criterion that varies by $550,000 across the shortlist means something entirely different from forty percent on one that varies by $40,000, yet committees assign weights before seeing the options and then apply them to whatever spread shows up. That single omission — importance stated without range — is the most common reason a scorecard produces a winner the room does not believe.