AI-assisted decision matrix
Compare real options with weighted criteria and evidence, let AI challenge your reasoning, and make a decision you can explain — and still call your own.
This free workbook takes one decision from its question and non-negotiables, through weighted scoring and recorded evidence, to a documented choice with its trade-offs and next action.
7 worksheets · Up to 6 options and 10 criteria · 4 AI challenge prompts
The problem
Big decisions, small amounts of structure.
Most decisions that matter are made from a list, a hunch and a conversation. AI makes it tempting to skip even that, and ask a model which option to pick.
A pros and cons list where every point counts the same
The option someone already preferred, justified afterwards
“Which should I choose?” typed into a chatbot
Estimates quoted as if they were facts
A decision nobody can explain six months later
Definition
What is a weighted decision matrix?
A table that scores each option against the criteria that matter, with each criterion weighted by how much it should count. It turns a vague comparison into explicit trade-offs.
It supports judgement. It doesn’t prove there is one objectively correct answer.
Use it when
- Several real options exist
- Multiple criteria and trade-offs matter
- The decision deserves a documented rationale
Avoid it for
- Trivial or obvious choices
- Decisions with only one viable option
- Specialist legal, medical, safety or regulated analysis without qualified review
How it scores
Five rules that keep the numbers honest.
A small example
The leader isn’t the decision.
From the workbook’s fictional example: a small consultancy choosing between three software options.
Which business software option should a small consultancy adopt for the next 12 months?
Criteria & weights
Evidence
Option B has the shortest setup time, but the evidence for long-term flexibility is incomplete.
Result
Option C ranks first after scoring. The owner still reviews the trade-offs and the AI challenge findings before deciding.
How to run it
Six steps, one decision.
Criteria and weights come before scoring, and scoring comes after the evidence. The order is what stops the numbers being reverse-engineered from a preference.
- 01
Set up the decision
Frame one decision, list the viable options, define the non-negotiables and record your assumptions.
- 02
Choose criteria and weights
Define what matters, weight it to 100%, and write criterion-specific scoring guidance.
- 03
Record the evidence
Facts, estimates, sources, confidence and the rationale behind each score.
- 04
Score the matrix
Score each option 1–5 against each criterion. Eligibility and ranking calculate themselves.
- 05
Let AI challenge it
Four prompts test the criteria, the evidence, the leading option and the trade-offs.
- 06
Decide and document
Record the final choice, explain any override, and set the immediate next action.
Behind every score
Know which numbers rest on what.
The evidence register records the finding, its type, source and confidence, whether it’s an assumption, the rationale for the score, and what you still don’t know.
Requirement status
Evidence warning
Evidence types
The important part
AI as the reviewer, not the decision-maker.
Four prompts point AI at the weak spots in your reasoning. Under each one you record the finding, whether it matters, and what changed as a result.
Criteria review
Looks for missing, overlapping or mis-weighted criteria before you rely on the model.
Evidence challenge
Finds unsupported claims, weak sources and assumptions presented as facts.
Decision red team
Builds the strongest reasonable case against the option currently in the lead.
Trade-off analysis
Explains what you gain and give up with each option, without picking one for you.
The decision
Overrides are allowed. Unexplained ones aren’t.
The summary shows the highest-scoring eligible option next to the one you actually chose. If they differ, it asks why.
Then it records the immediate next action, and the trigger that should make you reopen the decision.
Decision rationale
- 1Why this option was selected
- 2Most important advantages
- 3Trade-offs accepted
- 4Main risk
- 5Important assumption
- 6Remaining uncertainty
The workbook
Decide with the reasoning written down.
Everything above is on this page for free. The workbook does the part a webpage can’t: check the weights, apply the requirements, calculate and rank, and carry it all into a decision you can explain later.
ai-assisted-decision-matrix-workbook.xlsx
XLSX · 7 worksheets
Enter your email and we’ll send the workbook straight to your inbox. Free, no account, no card.
We may email you occasionally for feedback, or to tell you about new Cognardo resources. Unsubscribe any time.
Where it helps
Any choice worth explaining later.
Choosing software or vendors
Compare real options on what matters to you, not on feature lists.
Hiring and partners
Weigh candidates or agencies consistently, with the evidence written down.
Strategy and priorities
Choose between initiatives competing for the same budget.
Client recommendations
Show a client exactly how a recommendation was reached.
Personal and career decisions
Structure a hard choice without pretending it is only maths.
Revisiting a decision
A review trigger tells you when the decision should be reopened.
Keep it where you work
Keep the challenge prompts ready for the next decision.
The four review prompts work on any decision, which means you will want them again. Workbench is our browser extension for exactly that — keep prompts organised, fill the variables in a form, and run them without leaving ChatGPT, Claude or Gemini.
FAQ
Questions, answered.
What is a weighted decision matrix?+
A table that scores each option against a set of criteria, where each criterion carries a weight reflecting how much it matters. Multiplying scores by weights gives each option a comparable total. It makes trade-offs explicit, but it is a support for judgement, not a replacement for it.
Can AI make the decision for me?+
It should not. AI is useful as a reviewer: finding gaps in your criteria, challenging weak evidence, arguing against the leading option and explaining trade-offs. The decision, and the accountability for it, stay with you. The workbook records your final choice separately from the calculated result for exactly that reason.
How should I weight the criteria?+
By how much each one should influence this particular decision, set before you score anything. The weights must total 100%. If changing a weight would flip the result, that is worth knowing, and the AI challenge prompts help you test it.
What is a non-negotiable requirement?+
Something an option must meet to be considered at all, such as a budget ceiling or a legal requirement. One confirmed “does not meet” makes the option ineligible, however well it scores elsewhere. “Unknown” marks it for review rather than ruling it out.
What if I disagree with the top-scoring option?+
You can choose differently. The Decision Summary shows whether your choice matches the highest-scoring eligible option and asks for an override reason when it does not. A documented override is a legitimate outcome. An unexplained one is the problem.
Is AI-generated information evidence?+
Not automatically. The evidence register labels it as AI-generated so it stays distinguishable from facts, estimates and stakeholder input. Check any important claim before it influences a score.
When should I revisit a decision?+
When the review date arrives or the reconsideration trigger you recorded happens: a key assumption turns out to be wrong, a cost changes, or a new option appears. Because the reasoning is written down, revisiting it is quick.