# Decision matrix: decisions you can defend

A decision matrix compares options against weighted criteria. It shows why one option ranks first.

-   3+ viable options
-   Weighted criteria
-   Clear rationale

Also helps with:[Annual planning](https://www.reflacto.com/moments/annual-planning#situation-cut-projects)

## Compare options with weighted criteria

The matrix scores every option against the same criteria. Weights show priorities; reasons and risks show how sound the result is.

1.  ### Define the options

    List only serious alternatives at a comparable level.

2.  ### Weight the criteria

    Set the criteria and their weights before scoring.

3.  ### Score the options

    Score every option on the same scale using available evidence.

4.  ### Explain the front-runner

    Compare the results. Explain why the front-runner ranks first.

5.  ### Check the risks

    Record downsides, uncertain assumptions, and mitigations.

## When this method helps

Use it for three or more viable options when others need to follow the reasoning.

### Key benefits

-   Shows which option best matches the chosen priorities
-   Makes scores and reasoning easy to follow
-   Surfaces overlooked factors and risks

## Limits and when to pause

The result is only as sound as its criteria, weights, and evidence. A high score is not proof; show ties and weak evidence.

## Decision matrix examples

The five steps stay the same whatever you decide: which tool to buy, what to do first, or how the team works. As in Reflacto, weights run from 1 (low) to 5 (very high) and scores from 1 (poor) to 5 (excellent). Each example shows the steps separately and then how you could share the result.

### Choosing a tool

Pick one of several comparable products for the team.

Illustrative scenario

**Define the options**: The twelve-person support team needs a new ticket system before its current contract ends in March. Three products are serious candidates: Zendesk, Freshdesk and Jira Service Management.

**Weight the criteria**: Before anyone scores, the team agrees on four criteria and weights: integrations 5, implementation effort 4, usability 4, ongoing cost 3. That makes 80 points the maximum.

**Score the options**: Scores come from trial accounts and the vendors' price lists. On effort and cost, a 5 means little effort and low cost. Weighted totals: Freshdesk 71, Zendesk 66, Jira Service Management 56.

**Explain the front-runner**: Freshdesk leads by 5 points. Zendesk integrates better and is easier to use, but Freshdesk is quicker to set up and cheaper to run. Even with implementation effort weighted 2 instead of 4, Freshdesk still leads, 61 to 60.

**Check the risks**: The reports the team lead needs every week were not part of the trial. A two-week pilot tests them before the contract is signed. If they fall short, Zendesk is the fallback.

How you could share it

> I recommend Freshdesk as our new ticket system. We compared three tools on integrations, implementation effort, usability and cost, and set the weights before scoring. Freshdesk has 71 of 80 points, Zendesk 66. Zendesk integrates better, but Freshdesk is quicker to set up and cheaper to run. Before we sign, a two-week pilot checks whether its reports cover what we need.

### Choosing what to do first

Decide which of several improvements gets the team's limited time.

Illustrative scenario

**Define the options**: The team has about three weeks this quarter for improvements. Three candidates compete for it: automating the manual release steps, fixing the flaky tests, and upgrading the web framework, whose security updates end in September.

**Weight the criteria**: She sets four criteria with the team: time saved per week (weight 4), fewer errors that customers notice (5), effort (3) and risk of waiting (4). The maximum is 80 points.

**Score the options**: Time saved is based on CI logs and the release checklist: manual releases cost about six hours a week, rerunning flaky tests about four. Weighted totals: release automation 55, flaky tests 52, framework upgrade 44.

**Explain the front-runner**: Release automation leads by only 3 points, and one changed score could flip the order: with a 5 instead of a 4 for time saved, the tests would reach 56. She checks that score against the logs. Six hours beats four, so the order holds.

**Check the risks**: Only one engineer knows the release script well, so a second engineer pairs with him on the work. The framework upgrade cannot wait past September, so she schedules it for July.

How you could share it

> My proposal for this quarter's improvement time: we automate the release steps first. It scored 55 of 80 points, just ahead of fixing the flaky tests at 52. The difference is time: according to the CI logs, manual releases cost us about six hours a week, flaky tests about four. The framework upgrade scored lowest, but its security updates end in September, so I've planned it for July.

### Changing how the team works

Choose a way of working that affects everyone on the team.

Illustrative scenario

**Define the options**: Other teams send the six-person platform team about 25 requests a week through chat. Everyone gets interrupted, and answers take anywhere from an hour to four days. He compares three setups: a weekly rotation where one person handles all requests, fixed office hours twice a day, and a queue that he assigns himself.

**Weight the criteria**: Before the options are discussed in detail, the team agrees on criteria and weights: focus time for the team 5, response time for other teams 4, a fair share of the load 3, setup effort 2. The maximum is 70 points.

**Score the options**: Everyone scores alone first. Where scores are more than one point apart, the team discusses the reasons and agrees on one number. Weighted totals: rotation 59, office hours 55, queue 43.

**Explain the front-runner**: The rotation answers requests within a few hours and still keeps five of six people free of interruptions. Office hours protect focus best but score low on response time: a request that arrives just after a slot waits half a day.

**Check the risks**: Whoever is on rotation does almost no project work that week, and newer team members know little about older systems. For their first turn, they pair with someone experienced. The team reviews the setup after six weeks, once everyone has had a turn.

How you could share it

> Starting Monday, one of us handles all incoming requests for a week, in turn. We compared three setups on focus time, response time, fair load and setup effort, and the rotation came out ahead with 59 of 70 points. Office hours protect focus even better, but other teams would sometimes wait half a day. For your first turn, you pair with someone who has done it before. After six weeks, when everyone has had a turn, we review it together.

The numbers are illustrative. Each total is weight times score, added up across all criteria, which is how Reflacto calculates it. Each message names the recommendation, the margin, the reason and the main risk. Keep the full matrix ready for questions.

### Vague or specific criteria?

With a vague criterion, everyone scores their own idea of it. A specific one says what a good score means, so two people scoring the same option end up close.

-   Vague – “Cost”

    Specific – “Licence and setup cost over two years: 5 means under €15,000, 1 means over €40,000”

-   Vague – “Usability”

    Specific – “A new agent can answer a typical ticket after one hour of training”

-   Vague – “Good for the team”

    Specific – “Hours per week each person can work without interruptions”

With vague criteria, disagreement disappears into the total. With specific ones, you see where people scored differently and can ask why. If a criterion such as values or trust can't be made specific, another method may fit better.

### Template: share your recommendation

Copy the template and replace each bracket with your own case.

I recommend \[option\]. We compared \[number\] options on \[criteria\] and set the weights before scoring. \[Option\] has \[points\] of \[maximum\] points, \[runner-up\] has \[points\]. It leads because of \[the criteria that make the difference\]. The main risk is \[risk\]; we handle it by \[mitigation\].

#### Check before you share it

-   Define the options: Are all options serious and at a comparable level?
-   Weight the criteria: Were the weights fixed before you scored the first option?
-   Score the options: Can you name the evidence behind each score?
-   Explain the front-runner: Would one changed score flip the order? If so, say so.
-   Check the risks: Does the front-runner's main risk have a mitigation and an owner?

## Questions Reflacto asks

The decision matrix has no single author. Three questions follow its classic steps: list the options, name the criteria, then weight them and score each option. Reflacto added two: why the top option deserves to lead, and what could go wrong with each option.

### Operational decision (team)

Choose between several viable options and explain the choice.

1.  What options are you considering? (Classic matrix)
2.  What criteria are most important for this decision? (Classic matrix)
3.  Rate each option against each criterion (Classic matrix)
4.  Based on your scoring, which option ranks highest and why? (Added by Reflacto)
5.  What are the key risks for each option? (Added by Reflacto)

After the last question, Reflacto shows your options ranked by weighted total and a decision plan with concrete next steps.

## Questions about using the decision matrix

### Set up the matrix

#### How many criteria should I use?

Usually four to seven. Too few let one factor decide; too many blur the differences between options.

You could say

> “Integrations, implementation effort, usability and ongoing cost. Is anything missing that we'd regret later?”

If you need more than seven, check whether you are making two decisions at once, such as which tool and when to switch. Each criterion should be specific enough that two people would score it alike. [Vague and specific criteria side by side](https://www.reflacto.com/frameworks/decision-matrix#vague-or-specific-criteria)

#### How do I weight criteria without engineering the outcome?

Set criteria and weights before you score any option. Otherwise the matrix tends to justify the answer you already prefer.

You could say

> “Before we look at the tools: what matters most, and how much?”

Agree on the weights with the people affected and write them down. If you change them later, say so and explain why. [Example: changing how the team works](https://www.reflacto.com/frameworks/decision-matrix#example-request-handling)

### Score the options

#### Should I score alone or with the team?

With the people who know the options: each person scores alone first, then you compare.

You could say

> “Everyone scores on their own first. Where we're more than one point apart, we talk about why.”

Scoring alone first keeps the first number someone says from steering the rest. A big gap between two people usually points to a vague criterion or missing information, not to a math problem.

#### What if two options are almost tied?

Find the single score that would flip the order, and check that score against evidence.

You could say

> “One more point on time saved and the flaky tests would win. What do the CI logs say?”

If the order still depends on a guess, the matrix has done its job: it shows you which question to settle. Reflacto flags exact ties in the result. [Example: choosing what to do first](https://www.reflacto.com/frameworks/decision-matrix#example-priority)

### Use the result

#### What if the result feels wrong?

Don't quietly change the scores. Name the missing concern, such as risk, reversibility, capacity or trust, add it as a criterion and recalculate.

You could say

> “The numbers favor option B, but I'm uneasy. Which criterion are we missing?”

A gut feeling often points to a criterion you left out. If you add one, show the new result next to the old one so nobody suspects you tuned it.

#### How do I explain the decision to others?

Name the recommendation, the margin, the criteria that made the difference and the main risk.

You could say

> “Freshdesk has 71 of 80 points, Zendesk 66. It's quicker to set up and cheaper to run.”

Keep the full matrix ready for questions instead of presenting every score. [Template: share your recommendation](https://www.reflacto.com/frameworks/decision-matrix#template)

### About the method

#### When is a decision matrix the wrong tool?

When you have only two options, when the choice is about values, in a genuine emergency, or when someone else owns the decision.

For two options, a simple comparison of pros and cons is enough. If someone else decides, clarify ownership and alignment first. [Limits of the method](https://www.reflacto.com/frameworks/decision-matrix#limitations)

#### Is the decision matrix an established method?

Yes, as a general practice without a single author. Weighted scoring appears in engineering standards such as Germany's VDI 2225 and in quality-management literature.

Three of Reflacto's five questions follow the classic steps: options, criteria, and weighted scores. Reflacto added the question on why the front-runner leads and the risk check for each option. [Origin and sources](https://www.reflacto.com/frameworks/decision-matrix#provenance)

## Origin and Reflacto's adaptation

Decision matrices have no single owner. Weighted scoring appears in engineering standards such as Germany’s VDI 2225 and in quality-management literature.

Also known as: weighted decision matrix, weighted scoring model

### What Reflacto changed

**From the classic decision matrix**: From the classic decision matrix: the options, weighted criteria, and a score for every option on every criterion, added up to a weighted total.

**Added by Reflacto**: Reflacto asks why the top score deserves to lead and adds a risk check for each option, so ties and weak evidence stay visible.

[See every question Reflacto asks](https://www.reflacto.com/frameworks/decision-matrix#questions)

### Primary sources and further reading

-   [ASQ: what is a decision matrix? – Steps for choosing criteria, weighting, scoring, and reading the result.](https://asq.org/quality-resources/decision-matrix)
-   [Systematic review of sensitivity analysis in multi-criteria decisions – Peer-reviewed review of how rankings can change when criteria weights change.](https://www.sciencedirect.com/science/article/pii/S156849462300933X)
-   [VDI 2225: technical and economic evaluation (de) – Explains the German engineering standard for weighted scoring of alternatives.](https://www.ressource-deutschland.de/leitfaden-re/methoden/technisch-wirtschaftliche-bewertung)

Editorially reviewed by [Daniel Schmeiß](https://www.reflacto.com/about)

Last substantive review: September 29, 2026

## Related reflection methods

Compare another method that approaches a similar situation from a different angle.

-   [Pros and cons: weigh the consequences – You need to weigh a consequential decision and explain the reasoning. You collect the arguments for and against, then judge how strong and how likely each one is. This makes the trade-offs visible and the final call easier to justify.](https://www.reflacto.com/frameworks/pros-cons)
-   [GROW Model – You feel stuck on a complex problem or are helping someone move forward. GROW guides you from the Goal through the current Reality and the Options to a Way Forward you commit to. It works well for your own thinking and as a coaching structure for others.](https://www.reflacto.com/frameworks/grow)

[Compare all methods](https://www.reflacto.com/frameworks)

---

Original page: https://www.reflacto.com/frameworks/decision-matrix