The Complete Guide to Multi-Criteria Decision Analysis (MCDA)

Multi-Criteria Decision Analysis is not a single technique. It is a family of structured methods designed to evaluate options when multiple, often conflicting, criteria matter. Whether you are selecting a vendor, choosing a technology stack, or deciding which product feature to build next, MCDA provides the analytical rigor that gut feel cannot.

The fundamental challenge MCDA solves is incommensurability: how do you compare cost (measured in dollars) against reliability (measured in uptime percentage) against team satisfaction (measured on a subjective scale)? Simple weighted scoring can do this, but it makes hidden assumptions about tradeoffs that more sophisticated methods make explicit.

The four major MCDA methods each handle this challenge differently. Understanding their strengths and limitations is essential for choosing the right approach.

AHP, the Analytic Hierarchy Process developed by Thomas Saaty in the 1980s, structures decisions as hierarchies: the goal at the top, criteria in the middle, and alternatives at the bottom. Decision-makers provide pairwise comparisons ("Is cost twice as important as speed, or three times?"), and AHP converts these judgments into mathematical weights. A consistency ratio flags contradictory comparisons. AHP is particularly powerful when criteria importance is subjective and stakeholders disagree, because the pairwise comparison process forces explicit discussion about tradeoffs.

TOPSIS, the Technique for Order of Preference by Similarity to Ideal Solution, takes a geometric approach. It identifies the ideal solution (best score on every criterion) and the anti-ideal solution (worst score on every criterion), then ranks alternatives by their distance from each. The option closest to the ideal and farthest from the anti-ideal wins. TOPSIS works best when you have quantifiable data for all criteria and want a straightforward ranking without the overhead of pairwise comparisons.

ELECTRE (Elimination and Choice Expressing Reality) introduces the concept of outranking. Rather than producing a single score, ELECTRE determines which alternatives outperform others on enough criteria to be considered superior, while checking that no single criterion creates a disqualifying weakness. This non-compensatory logic is critical for decisions where certain minimum thresholds must be met. A vendor with excellent pricing cannot compensate for failing your security requirements. ELECTRE captures this reality.

PROMETHEE (Preference Ranking Organization Method for Enrichment Evaluations) builds on outranking with preference functions. For each criterion, you define how differences between alternatives translate into preferences. A price difference of five dollars might be negligible, but five hundred dollars is significant. PROMETHEE's six preference function types (usual, U-shape, V-shape, level, linear, and Gaussian) let you model these thresholds precisely. The resulting positive and negative flows provide both a complete ranking and insight into why each alternative ranks where it does.

Beyond these four, dozens of other MCDA methods exist: MAUT (Multi-Attribute Utility Theory), MACBETH, VIKOR, COPRAS, and more. Each makes different assumptions about preference structures, compensation between criteria, and uncertainty handling. In practice, the best approach is often to run multiple methods and look for convergence. When AHP, TOPSIS, and PROMETHEE all agree on the top choice, you have strong confidence. When they disagree, the disagreement itself reveals important structural features of your decision.

The practical barrier to MCDA adoption has always been effort. Setting up an AHP analysis manually requires building comparison matrices, computing eigenvectors, and checking consistency ratios. TOPSIS needs normalized decision matrices and weighted calculations. Running three methods on the same decision traditionally means days of spreadsheet work or expensive consulting engagements.

This is precisely the problem that decision intelligence platforms solve. SolveRight runs your decision through up to 155 frameworks simultaneously, including all major MCDA methods. You describe your decision in natural language, and the platform extracts criteria, maps them to appropriate frameworks, computes scores, detects cross-framework contradictions, and produces a ranked recommendation with full transparency into how each score was derived.

The combination of automated MCDA with cross-framework contradiction detection is particularly powerful. If AHP ranks Option A first but ELECTRE eliminates it due to a non-compensatory weakness, SolveRight flags this disagreement and explains why the methods diverge. This is insight that no single method can provide, and it is the difference between rigorous analysis and false confidence.

For teams adopting MCDA for the first time, start with a clear problem: a decision with at least three options and at least four criteria. Define your criteria before looking at the options to avoid anchoring bias. Weight criteria before evaluating options to prevent reverse-engineering weights to justify a preferred choice. These process disciplines matter as much as the mathematical methods.

MCDA is not about replacing human judgment. It is about structuring human judgment so that biases are surfaced, tradeoffs are explicit, and decisions are defensible. The best decision-makers combine analytical rigor with domain expertise, using MCDA to ensure their experience is applied systematically rather than selectively.