# ELECTRE Method Explained: A Practical Decision Guide > Learn how the ELECTRE method compares options across conflicting criteria using outranking, thresholds, and veto logic, with practical limits explained. Source: https://www.electe.net/it/post/electre-method-explained Site guide: https://www.electe.net/it/llms.txt A procurement manager has two credible suppliers on the shortlist. One offers the lower price, but its delivery record is erratic. The other costs more, yet consistently meets agreed dates and quality requirements. A single scorecard can make the cheaper supplier look like the obvious winner, even when a missed delivery would stop production. That's the problem the **ELECTRE method** is designed to address. Rather than forcing every consideration into one compensatory score, it compares alternatives and asks whether one is sufficiently supported, without suffering an unacceptable weakness. This makes it relevant to European SMEs choosing suppliers, software, investments, locations, or other options with conflicting criteria. This guide explains the method in plain language, from its French origins to its thresholds, veto logic, calculation steps, variants, strengths, limitations, and practical fit. The central lesson is simple: **a strong result on one criterion shouldn't automatically cancel out a failure that matters more than the average suggests**. ## Table of Contents - [What ELECTRE Is and Where It Comes From](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#what-electre-is-and-where-it-comes-from) - [The Core Idea: Why ELECTRE Is Non-Compensatory](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#the-core-idea-why-electre-is-non-compensatory) - [Key Concepts: Outranking, Concordance, Discordance, and Thresholds](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#key-concepts-outranking-concordance-discordance-and-thresholds) - [The Classical ELECTRE Procedure: Step by Step](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#the-classical-electre-procedure-step-by-step) - - [Build and prepare the matrix](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#build-and-prepare-the-matrix) - [Apply weights and compare pairs](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#apply-weights-and-compare-pairs) - [A Worked Example: Choosing a Supplier](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#a-worked-example-choosing-a-supplier) - [The ELECTRE Family in Brief](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#the-electre-family-in-brief) - [Strengths and Limitations in Equal Measure](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#strengths-and-limitations-in-equal-measure) - [When to Use ELECTRE, and When to Skip It](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#when-to-use-electre-and-when-to-skip-it) - [ELECTRE Is Not ELECTE](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#electre-is-not-electe) - [Conclusion](https://www.outrank.so/dashboard/articles/267c1d99-ed44-4323-b4e9-a74393eeadb5#conclusion) ## What ELECTRE Is and Where It Comes From **ELECTRE (ÉLimination Et Choix Traduisant la REalité)** is a family of **multi-criteria decision-making methods** developed by **Bernard Roy and colleagues** in France. Its core ideas were conceived in **1964**, used on a live client problem in **1965**, and reported publicly in a **1966 research report**. Roy's first detailed journal article followed in **1968**, providing the basis for ELECTRE I. The [ELECTRE method timeline and origins](https://en.wikipedia.org/wiki/ELECTRE) and the French-language [ELECTRE overview](https://fr.wikipedia.org/wiki/ELECTRE) reflect this development. The work began at SEMA, a French consultancy, in response to a practical business decision. The team faced criteria with different meanings, where a weighted sum could let an excellent result in one area offset a result that managers considered unacceptable. The approach grew from that consulting problem rather than from a purely abstract mathematical exercise. [ELECTE guida ai sistemi decisionali](https://www.electe.net/post/sistemi-di-supporto-alle-decisioni-ai-lascesa-degli-advisor-nella-leadership-aziendale) RemoveUploadDownloadRegenerateAsk AI The ELECTRE family belongs to the **outranking** tradition. It compares alternatives pair by pair, weighing evidence that supports one comparison against evidence that challenges it. The output need not be a complete league table. It can show which alternatives reasonably outrank others, which are blocked by a serious weakness, and which remain incomparable. For managers, this preserves an important judgement. A criterion does not have to become a common utility value before it can influence the decision. A strong overall profile may still fail if one weakness has unacceptable operational consequences. That distinction defines the method. ## The Core Idea: Why ELECTRE Is Non-Compensatory A weighted-sum model is **compensatory**. Give each criterion a weight, score every alternative, multiply, and add. The mathematics permits a high score on one criterion to offset a low score on another, provided the totals work out. Consider a simplified supplier assessment. Supplier A scores **9 on price** and **1 on delivery reliability**. Supplier B scores **4 across both criteria**. If price receives enough weight, Supplier A can win the weighted average. The model may be internally consistent, but it can still produce a decision that operations managers reject immediately. The **ELECTRE method is non-compensatory**. It asks whether A has enough supporting evidence to outrank B, while also checking whether A has a severe weakness on any critical criterion. A strong price advantage can contribute to concordance, but it cannot automatically erase a delivery failure that reaches a veto level. RemoveUploadDownloadRegenerateAsk AI The difference can be stated in one sentence: > **Weighted scoring asks which alternative has the highest total; ELECTRE asks whether one alternative is sufficiently supported and free from unacceptable failure.** That doesn't mean ELECTRE ignores weights. Weights still express the relative importance of criteria. The difference is that weights operate alongside **concordance**, **discordance**, and **veto** conditions rather than serving as the only decision rule. This approach suits situations where criteria aren't naturally commensurable. Price may be measured in euros, delivery in days, quality through inspection scores, and geographic risk through a qualitative assessment. Combining them into one number can create false precision. Pairwise outranking keeps the underlying judgement visible. A practical manager can therefore use the method to distinguish between an alternative that's slightly worse overall and one that's unacceptable because of a critical defect. That distinction is the foundation for the concepts below. ## Key Concepts: Outranking, Concordance, Discordance, and Thresholds The terminology can sound technical, but each concept answers a practical question about a pair of alternatives. **Concept****One-Sentence Definition****Supplier Example****Outranking**Alternative A outranks B when A is at least as good as B on a sufficiently important coalition of criteria, without an unacceptable weakness.A supplier may be cheaper and closer, but it won't outrank another supplier if its defect performance is dangerously poor.**Concordance**Concordance measures how much weighted evidence supports the statement that A is at least as good as B.A concordance value of **0.75** means **75% of the weighted criteria** support A against B.**Discordance and veto**Discordance captures serious opposition to an outranking, while a veto criterion can block it automatically when the performance gap crosses a defined limit.A supplier may lead on price, but still fail the comparison if its quality or delivery weakness breaches a veto threshold.**Indifference threshold, q**The indifference threshold is the gap below which two performances are treated as practically equivalent.A small difference in delivery time may be ignored if it has no operational consequence.**Preference threshold, p**The preference threshold is the gap above which the decision-maker clearly prefers one alternative over another.A sufficiently large quality difference can establish a firm preference rather than a marginal one. **Concordance** is therefore not a simple count of criteria. It reflects their weights. If price and capacity carry more importance than geographic risk, their support contributes more to the index. The result remains a structured judgement, not a raw majority vote. **Discordance** adds the opposing side. It asks whether A performs so much worse than B on a particular criterion that the proposed outranking becomes implausible. The **veto principle** is the strongest form of this check. One criterion can block the relationship, even when the overall supporting coalition is strong. Thresholds make the comparison less brittle. Without them, tiny measurement differences could create artificial preferences. With them, the analyst can distinguish between negligible variation, weak preference, and meaningful superiority. These parameters also create a responsibility for the decision team. Thresholds shouldn't be selected merely because they produce a convenient result. They should reflect operational tolerances, contractual requirements, regulatory concerns, or an explicitly documented managerial judgement. That is why ELECTRE is powerful, but also sensitive. ## The Classical ELECTRE Procedure: Step by Step A practical ELECTRE I analysis starts with a **decision matrix**. Put alternatives in rows and criteria in columns. Record the raw information, such as supplier price, lead time, defect rate, capacity, and risk. ### Build and prepare the matrix First, identify whether each criterion is a benefit or a cost. Higher capacity may be better, while a lower defect rate is better. Qualitative criteria need an agreed scale and clear interpretation before scoring begins. Next, **normalise** the matrix. Normalisation places criteria measured in different units onto comparable scales. The precise formula depends on the ELECTRE variant and the type of data, so the analyst should document the chosen approach rather than treating normalisation as a neutral housekeeping step. RemoveUploadDownloadRegenerateAsk AI ### Apply weights and compare pairs Assign weights to reflect the decision team's priorities. The weights are then used in the **concordance index**, which aggregates the criteria that support the proposition that A is at least as good as B. For every pair, calculate how much A supports the comparison and how badly A falls behind on each criterion. The **discordance index** focuses on the largest or most serious opposition, depending on the variant. Insert the indifference, preference, and veto thresholds before finalising the pairwise relations. Thresholds translate operational judgement into rules, such as treating a small delivery difference as immaterial while blocking a supplier with an unacceptable defect gap. The resulting concordance and discordance conditions are then aggregated. If A has sufficient support and no criterion triggers a veto, the analysis records an outranking relation from A to B. Use scenario analysis to test how the relation changes when assumptions change. A structured [business scenario analysis process](https://www.electe.net/post/what-if-analysis-tool) can help the team examine alternative weights, thresholds, or risk tolerances without hiding those choices. Finally, the method synthesises the outranking graph. ELECTRE I may identify a **kernel**, or a set of acceptable alternatives, rather than a complete ranking. Other family members refine the procedure differently. Every parameter should be justified and reviewed with the people who understand the operational consequences. ## A Worked Example: Choosing a Supplier Take three suppliers and four criteria. Lower values are better for cost, lead time, and defect rate. Capacity is scored from low to high. **Supplier****Cost****Lead Time (days)****Defect Rate (%)****Capacity Score****A**701289**B**821028**C**781436 A simple weighted-sum model could place Supplier A first if low cost and high capacity receive enough weight. In that view, A's weaker quality performance can be compensated for by its advantage elsewhere. An ELECTRE-style reading asks a different question. Does A have enough support to outrank B, and is any weakness severe enough to block that claim? Against B, Supplier A is attractive on **cost** and slightly ahead on **capacity**. But B is better on **lead time** and much better on **defect rate**. If the decision team treats that quality gap as operationally unacceptable, the proposed outranking from A to B can be blocked by a veto on defect performance. That leads to a practical conclusion: - **Weighted sum:** A can still win if its price advantage is allowed to compensate for poor quality. - **ELECTRE:** B can defeat A if the quality gap is treated as a non-compensable weakness. - **Managerial interpretation:** B costs more, but A carries a failure mode the team has decided not to buy at any price. This is intentionally a qualitative illustration, not a full numeric ELECTRE computation. In a real analysis, the team would document the exact weights, thresholds, concordance rule, and veto condition before claiming a formal outranking result. Teams assessing technology providers can apply the same logic through a documented [SaaS provider risk assessment](https://www.electe.net/post/provider-due-diligence) process. ## The ELECTRE Family in Brief The family contains related methods for different decision tasks. - **ELECTRE I** supports choice and selection by building an outranking relation and identifying acceptable alternatives. - **ELECTRE II** extends the approach with stronger and weaker outranking relations to refine ordering. - **ELECTRE III** uses **fuzzy outranking**, representing credibility as degrees rather than relying only on crisp yes-or-no relations. - **ELECTRE IV** works without explicit criterion weights, which can help when stakeholders can't agree on a weighting scheme. - **ELECTRE IS** extends ELECTRE I by introducing **indifference and preference thresholds**. - **ELECTRE TRI** sorts alternatives into predefined categories rather than ranking them from best to worst. It fits tasks such as assigning cases to risk or priority classes. The choice depends on the question. If you need a nuanced comparison under uncertain judgements, ELECTRE III may be more appropriate than a crisp variant. If you need to decide whether an alternative belongs in an approved, monitored, or rejected category, ELECTRE TRI addresses a different problem from ranking. Don't choose a variant by name alone. Define the decision output first, then select the method whose outranking structure matches it. ## Strengths and Limitations in Equal Measure ELECTRE's strongest feature is its ability to represent **non-compensatory logic**. It can preserve a critical weakness instead of burying it inside an average. It also supports thresholds, qualitative assessments, mixed units, and explicit veto rules, giving managers a more realistic way to express operational constraints. The pairwise structure can create an audit trail. A team can inspect why A outranked B, which criteria supported the relation, and which criterion prevented the reverse relation. That is often clearer than presenting a single composite score with no visible account of unacceptable failure. RemoveUploadDownloadRegenerateAsk AI The limitations are just as important. - **Parameter sensitivity:** Weights, indifference thresholds, preference thresholds, and veto settings can be subjective. Small changes may materially alter the outranking result, a recurring point in methodological reviews of ELECTRE's sensitivity and complexity, including this [review article](https://pmc.ncbi.nlm.nih.gov/articles/PMC10656203/) and this broader [overview of ELECTRE methods](https://onlinelibrary.wiley.com/doi/10.1002/mcda.1482). - **Incomplete outcomes:** ELECTRE can produce incomplete rankings, incomparabilities, or intransitive relations. A decision team may not receive one stable ordering of every alternative, a trade-off discussed in this [review of multi-criteria decision methods](https://pmc.ncbi.nlm.nih.gov/articles/PMC10656203/) and in this [survey of outranking methods](https://hal.science/hal-00876980v1/document). - **Higher analytical effort:** Pairwise comparisons, thresholds, and relation synthesis require more work than a basic scorecard. - **No automatic objectivity:** The method structures judgement, but it doesn't eliminate judgement. The team remains responsible for defining what counts as unacceptable. > **Practical rule:** Treat sensitivity testing as part of the decision, not as an optional presentation exercise. ELECTRE is rigorous, but it isn't magic. It makes trade-offs and weaknesses more visible. It can't decide whether a defect threshold is commercially or operationally justified. ## When to Use ELECTRE, and When to Skip It Use ELECTRE when you have a **small set of serious alternatives**, conflicting criteria, and at least one failure that shouldn't be offset by success elsewhere. Supplier selection, industrial location, energy technology, software procurement, and investment screening can all fit this pattern. It's particularly useful when the team needs to explain why a cheaper or faster option lost. The outranking relation can show that the decision wasn't based on an unexplained preference, but on a defined combination of support, opposition, and veto conditions. Skip it when the decision is simple and fully compensatory. If every criterion can reasonably be expressed in money, the trade-offs are accepted, and one alternative clearly dominates, a weighted score or cost-benefit analysis may be faster and easier to defend. [distribution tips for lead magnets](https://supportgpt.app/blog/lead-magnet-examples) It can also be excessive when the shortlist is large or the deadline leaves no time to validate parameters. A basic scorecard is often sufficient for a routine purchase, such as choosing a low-cost office printer where reliability and price have straightforward trade-offs. ## ELECTRE Is Not ELECTE **ELECTRE** is a generic acronym for a decision method used in academic and industry literature since the 1960s. **ELECTE** is a distinctive coined word and the name of an AI-powered data analytics platform for European SMEs. The two are unrelated. Readers looking for the company can find it at [electe.net](https://www.electe.net/). ## Conclusion The ELECTRE method is most useful when averages conceal the decision that managers need to make. It compares alternatives pairwise, measures supporting and opposing evidence, and allows a critical weakness to block an outranking through a veto condition. That non-compensatory insight is its main value. A cheaper supplier isn't automatically better if its defect rate or delivery reliability violates a requirement the business can't accept. At the same time, ELECTRE demands disciplined parameter choices and may produce an incomplete or inconclusive ranking. Use it when those trade-offs are real, document the thresholds carefully, and choose a simpler method when the decision is purely compensatory. **Glossary** - **Alternative:** An option being evaluated, such as a supplier or software product. - **Criterion:** A dimension used to assess alternatives, such as cost, quality, or risk. - **Concordance:** Weighted evidence supporting the claim that one alternative is at least as good as another. - **Discordance:** Evidence that an alternative performs seriously worse on a particular criterion. - **Outranking:** A relation showing that one alternative can reasonably be considered at least as good as another. - **Threshold:** A boundary defining indifference, preference, or veto. - **Veto:** A severe weakness that blocks an outranking despite support from other criteria.