Decision Matrix and After-Action Review: Two Templates
A decision matrix is a grid that scores several options against weighted criteria and ranks them by total, so a choice can be compared and defended on one page.
Teams reuse two templates more than any other: the matrix before a decision, and the after-action review once it has played out. This page gives the sections of each in order, the wording for each section, and a worked example you can copy.
Quick answer: what is a decision matrix?
To make a decision matrix, list your options across the top and your criteria down the side, weight the criteria, score every option, and multiply each score by its weight. The highest weighted total is the option that best fits the priorities you wrote down first.
A decision matrix, also called a weighted scoring model or a decision grid, is a table that compares options on the same criteria. Each criterion carries a weight showing its importance, and each option receives a score on each criterion. Weights multiplied by scores and summed give every option a single total. The method turns a loose discussion into an explicit trade-off, so a committee argues about weights and evidence rather than about whose favourite wins. It structures a judgement. It does not make one.
The idea is old. In a letter to Joseph Priestley dated 19 September 1772, Benjamin Franklin described dividing "half a sheet of paper by a line into two columns" and calling the technique "Moral or Prudential Algebra", then striking out reasons of equal weight until the balance showed. A modern decision matrix is that method with the weights made numeric.
The five parts of a decision matrix
Every decision matrix has the same five parts. Miss one and the grid either cannot be scored or cannot be defended.
Options. The alternatives on the table, one column each. Three to five is the working range. Include the do-nothing option as a column, because it is the one a committee quietly prefers and rarely sees scored. Example: Vendor X, Vendor Y, Vendor Z.
Criteria. The factors the decision turns on, one row each. Five to seven is enough. Each must be something you can score from evidence, and no two should measure the same thing twice. Example: fit to requirements, total cost, time to implement, security, vendor support.
Weights. How much each criterion matters relative to the others, set before any option is scored, summing to 100 percent. Weights are the decision's real content: they are the argument about priorities, made visible. Example: fit 30, cost 25, time 20, security 15, support 10.
Scores. A rating for each option on each criterion, on one fixed scale such as 1 to 5, with a written anchor for what a 1, a 3 and a 5 mean. Without anchors, a score is a mood. Example: on cost, 5 means lowest three-year cost, 1 means highest.
Weighted total. Each score multiplied by its weight, summed down the column. It is the only arithmetic in the method, and it is the number that makes the matrix feel objective. Treat it as a ranking aid, not a verdict. Example: 4 times 0.30 plus 3 times 0.25, and so on down the column.
The trade-off: five parts is light on purpose. Every part you add, such as sub-criteria or probability adjustments, makes the grid look more rigorous and harder to challenge, which is exactly the wrong direction for a method whose value is being questioned openly.
Decision matrix example: choosing a reporting platform
One example, carried end to end. A finance team must pick one of three vendors for a reporting platform. The weights were agreed and signed off first, then each vendor was scored from 1 (weak) to 5 (strong) against written anchors. The scores below are illustrative inputs, not market data.
| Criterion (weight) | Vendor X | Vendor Y | Vendor Z |
|---|---|---|---|
| Fit to requirements (30%) | 5 | 3 | 4 |
| Total three-year cost (25%) | 2 | 5 | 3 |
| Time to implement (20%) | 3 | 5 | 4 |
| Security and compliance (15%) | 4 | 3 | 5 |
| Vendor support (10%) | 4 | 2 | 4 |
| Weighted total | 3.60 | 3.80 | 3.90 |
The arithmetic, shown for Vendor Z: 4 x 0.30 gives 1.20 for fit. 3 x 0.25 gives 0.75 for cost. 4 x 0.20 gives 0.80 for time. 5 x 0.15 gives 0.75 for security. 4 x 0.10 gives 0.40 for support. The five lines sum to 3.90. Vendor Y sums to 3.80 and Vendor X to 3.60.
Z ranks first, but the margin is 0.10 on a 5-point scale, which is thin. So run the check every matrix needs: change the weights once and see whether the winner survives.
| Weight set | Vendor X | Vendor Y | Vendor Z | Leader |
|---|---|---|---|---|
| Base weights (30 / 25 / 20 / 15 / 10) | 3.60 | 3.80 | 3.90 | Z |
| Fit raised to 40, cost cut to 15 | 3.90 | 3.60 | 4.00 | Z |
Vendor Z leads under both sets of weights, so the ranking is robust to that change. But the runner-up changes: Vendor Y is second on the base weights, Vendor X second once fit is raised. The honest conclusion for the slide is that Z is the preferred option, X and Y are close behind, and the decision depends on how much the team really values fit over cost. That sentence is what the matrix produced.
The trade-off: a clean winner invites the team to stop thinking. Present the sensitivity row next to the totals, so the reader sees how much of the answer is the scores and how much is the weights.
Decision matrix template: the sections, in order
Copy the block below into a document, a spreadsheet or a slide. The order matters: weights are fixed at step 3, before any score exists.
Decision matrix template (copy and fill in)
Step 9 is the one most templates omit, and it is the slide title. State the recommendation, the reason, and the assumption it rests on.
Decision matrix types: weighted, unweighted and Pugh
The grid has neighbours that get confused with it. Pick by what you are choosing between and how much you need to defend the result.
| Variant | Scoring | Output | Best for | Main risk |
|---|---|---|---|---|
| Weighted decision matrix | Weights and 1 to 5 scores | A weighted total per option | Choosing between three to five real alternatives | False precision from decimals |
| Unweighted matrix | Scores only, equal importance | A simple count or sum | Fast screens where criteria are roughly equal | Hides that criteria are not equal |
| Pugh matrix | Each option against one baseline: better, same, worse | A net count of pluses and minuses | Engineering and design concept selection | Depends entirely on the baseline chosen |
| Must-have gate, then matrix | Pass or fail screen first, then scoring | A shortlist, then a ranking | Procurement and tenders with hard requirements | Gates set too tight leave nothing to score |
The Pugh matrix is named for Stuart Pugh, an engineering design professor at the University of Strathclyde, who set it out in his book Total Design. It rates each concept against a baseline, which is quick and avoids false precision, but it tells you nothing about how much better an option is. For a related grid that plots options on two chosen axes rather than scoring them, see the guide to a 2x2 positioning matrix.
The trade-off: the weighted matrix is the most defensible and the most gameable. The Pugh matrix is harder to game and harder to defend to someone who wants numbers.
When to use a decision matrix
Use one when three conditions hold. There are at least three genuine options. Several criteria pull in different directions, so no option wins on everything. And more than one person has to agree, or the decision has to be explained later to someone who was not in the room.
Typical consulting uses: vendor and system selection, site and market-entry shortlists, build versus buy versus partner, and ranking a long list of initiatives for a roadmap. In each, the matrix does the same job. It forces the criteria and their relative importance into the open before the preferences harden, and it leaves a record a steering committee can challenge cell by cell.
When a decision matrix misleads
A matrix flatters its conclusion: a number looks like evidence. Five failure modes account for most bad matrices.
Weights set after the favourite is known. If the team already prefers Vendor Z, the weights drift toward the criteria Z wins on. Fix and sign off the weights before anyone sees a score. That one rule removes most of the manipulation.
Double-counted criteria. Listing both time to implement and speed to value, or both cost and budget fit, gives one underlying factor twice the weight you meant to give it. Check each pair of rows for overlap.
Compensation hides a fatal flaw. A weighted sum lets a 5 on cost cancel a 1 on security. If a criterion is a hard requirement, it is a gate, not a row. An option that fails a gate should never reach the total.
Decimals imply precision nobody has. A gap of 0.1 on a 5-point scale is inside the noise of the scores themselves. The sensitivity test above shows the same thing: the winner held, the margin did not. Report ties honestly.
Scores from one person. A single scorer imports one set of biases. Have two or three people score independently, then discuss only the cells where they differ by two points or more.
Franklin saw the limit in 1772. He wrote that although "the Weight of Reasons cannot be taken with the Precision of Algebraic Quantities", considering each reason separately and comparatively made him less likely to take a rash step. The matrix is a discipline for the comparison. It is not a measurement.
What is an after-action review?
An after-action review (AAR) is a short, structured conversation held soon after an event, comparing what was meant to happen with what did, so the next attempt is better. Its four questions are: what was supposed to happen, what actually happened, why was there a difference, and what should we sustain or improve.
The US Army developed the method after studying the 1973 Yom Kippur War and institutionalised it at the National Training Center at Fort Irwin, whose first training rotations ran in 1981 and 1982. It is documented in the Army's Training Circular 25-20, A Leader's Guide to After-Action Reviews, dated September 1993. Business teams use it after a launch, a bid, a close, a deal or any project phase, and it pairs with the decision matrix naturally: the matrix records what you expected, the review records what happened.
The case for doing it is the shape of project outcomes. Bent Flyvbjerg and Alexander Budzier, writing in Harvard Business Review in September 2011, studied 1,471 IT projects and found an average cost overrun of 27 percent. One in six was a black swan, with a cost overrun averaging 200 percent and a schedule overrun of almost 70 percent. The average hides a tail, and a review is how a team finds out early which tail it is in.
After-action review template: the sections, in order
Seven sections, one page. Each has a job and a wording you can lift. Hold the review within a week of the event, while memory is accurate.
1.Purpose and scope
One line on the event under review and the date range. Wording: "This review covers the vendor selection from kickoff on 3 March to contract signature on 12 May."
2.What did we expect to happen?
The intended outcome and the plan, as written before the event. Wording: "We expected to select a vendor in eight weeks within a budget of the approved figure, with finance as the sole decision owner."
3.What actually happened?
Facts only, in time order, with no blame and no adjectives. Wording: "Selection took fourteen weeks. Security review began in week nine, after scoring had closed."
4.Why was there a difference?
Root causes, not symptoms. Ask why until the answer is a process or a decision, not a person. Wording: "Security was not a scored criterion, so it was not scheduled."
5.What do we sustain?
The practices that worked and should be kept on purpose. Wording: "Fixing the weights before scoring stopped a late argument over priorities."
6.What do we improve?
Specific changes, each with an owner and a date. Wording: "Add security as a gated criterion in the template. Owner: procurement lead. Due: next intake."
7.Follow-up
Where the actions will be tracked and when they will be checked. Wording: "Actions reviewed at the next steering committee on the agreed date."
Ground rules that make it work: everyone in the room took part in the event, rank stays outside the door, the facilitator asks and does not answer, and the output is actions, not minutes. Keep sections 3 and 4 apart. Mixing what happened with why it happened is how a review turns into an argument.
After-action review template (copy and fill in)
After-action review example: the vendor selection
The same finance team reviews its reporting platform decision. The matrix above was the plan. This is the review, filled in.
Expected
Select a vendor in eight weeks. Finance owns the decision. Budget as approved.
Actual
Fourteen weeks elapsed. Security review began in week nine. The final contract came in above the approved budget after a late integration add-on.
Why the gap
Security and integration cost were not criteria in the matrix, so neither was scheduled or priced during scoring.
Sustain
Weights were fixed and signed off before scoring. No late reweighting happened.
Improve
Add a must-have gate for security. Add integration cost as a scored line. Owner: procurement lead, before the next intake.
Notice the loop. The review found that security and integration cost were missing criteria, and the fix goes straight back into the decision matrix template: a gate and a new row. The two templates are one cycle, expectation then outcome, and each review sharpens the next matrix.
The trade-off: a review is only as honest as the room. Where people fear the consequences, sections 3 and 4 fill with safe generalities. Hold it away from the line manager, or have a neutral facilitator run it, and publish the actions rather than the transcript.
How do you put a decision matrix on a slide?
Do the scoring in a spreadsheet, where the formulas live, then show the conclusion on one slide.
Title the slide with the recommendation, not the method: Vendor Z is preferred, with a narrow margin over Y and X.
Show criteria as rows with weights in the left column, and options as columns. Keep it to five to seven rows so it reads at projector distance.
Show scores as plain numbers, and highlight only the weighted total of the leader with the one accent colour.
Add the sensitivity result as a single line under the grid: the leader holds when fit and cost weights change by 10 points.
Put the assumption the choice rests on in the footnote, and move the full score sheet to the appendix.
The review slide follows the same discipline: expected against actual in two columns, causes in one line each, actions with owners. If the output of a workshop is a pile of notes, the guide to turning meeting notes into editable slides covers the conversion, and a project charter is the natural home for the success criteria a later review will check against. For the people side of the same programme, a stakeholder map shows who has to agree to the weights.
Can Oria build the decision matrix slide for you?
Yes, and that is the narrow thing it is for. Oria is built for corporate documents and consulting presentations: board packs, steering committee decks, diligence exhibits, dense frameworks held to an enforced corporate template. It reaches you two ways: a PowerPoint add-in running in the task pane on Windows, macOS and PowerPoint for the web, and a connector for Claude and ChatGPT over MCP, so the slide can be built from the chat you are already in. The grid lands as native PowerPoint objects on your master, so weights and scores stay editable.
One-prompt decision matrix slide
The trade-off: Oria is built for the corporate environment and loses on highly visual work. Founder fundraising and pitch decks, launch and campaign decks, student presentations and marketing one-pagers want visual impact more than defensible content, and a design-led tool such as Gamma, Canva, Pitch or Beautiful.ai will beat it there and look better doing it. It also will not choose your weights, which is the judgement this page is about.
Which tool should you use for a decision matrix?
It depends on whether the grid is still being argued or is ready to present.
| If you need | Use | Why |
|---|---|---|
| A live scoring session with a team editing the same grid | A shared spreadsheet | Weights, scores and totals are formulas. Excel or Google Sheets recalculates as the room argues, which a slide cannot do. |
| A matrix that has to look designed for an external audience | Canva or Pitch | Where the job is visual impact rather than defensible content, a design-led tool will beat a corporate template and look better doing it. |
| The result as a native, editable slide inside your firm template | Oria | Built for board packs and steering committee decks: the criteria grid, weights and highlighted winner come out as real PowerPoint objects on your master. |
| Charts and waterfalls in the same pack on a team licence | think-cell | A mature add-in with an established enterprise footprint. It owns the chart exhibits around the matrix rather than the matrix itself. |
| A formal tender with scoring rules and audit requirements | Your procurement platform | Regulated tenders need a documented, repeatable score sheet held in the system of record. A slide is the summary of it, not the record. |
Common decision matrix and review mistakes
Frequently asked questions
What is a decision matrix?
A decision matrix is a grid that scores a set of options against a set of criteria so a choice can be compared on one page. Options run across the top, criteria down the side, and each cell holds a score. In the weighted version each criterion also carries a weight, and the weighted totals rank the options. It structures a judgement. It does not replace one.
How do you make a decision matrix?
List three to five options and five to seven criteria. Assign each criterion a weight, summing to 100 percent, and agree them before scoring. Score each option from 1 to 5 against written anchors. Multiply each score by its weight, sum each column, and rank the totals. Then change the weights once and see whether the winner holds.
What is a weighted decision matrix?
A weighted decision matrix gives each criterion a weight that reflects its importance, then multiplies every score by that weight before totalling. It differs from an unweighted matrix, where every criterion counts equally. Weighting is what makes the method useful when cost matters more than speed, or the reverse, and it is also where most of the manipulation happens.
What is the difference between a decision matrix and a Pugh matrix?
A weighted decision matrix scores every option on an absolute scale such as 1 to 5. A Pugh matrix rates each option only against one baseline as better, the same or worse, then counts the net result. Pugh is faster and suits concept selection in design. The weighted matrix gives finer ranking but demands more defensible scores.
When should you not use a decision matrix?
Skip it when one criterion is decisive, when an option fails a hard requirement, or when the real disagreement is about values rather than scores. A matrix cannot settle a priorities argument. It can only record the outcome. It also adds little to a choice between two near-identical options where a short conversation would do.
What is an after-action review?
An after-action review is a structured conversation held soon after an event to compare what was meant to happen with what did. The US Army formalised it in the 1980s and describes it in Training Circular 25-20. It asks four questions: what was supposed to happen, what actually happened, why the difference, and what to sustain or improve.
How is an after-action review different from a post-mortem?
A post-mortem is usually held after something failed and tends to hunt for cause. An after-action review is held after any significant event, including successes, and focuses on learning rather than fault. The format is the same four questions each time, which is what makes the lessons comparable across projects.