Free scoring matrix · score each use case, then see the order

AI Use-Case Prioritization Tool

For heads of AI and strategy leads choosing where to start. Score every candidate on identical criteria, adjust the weights to your business, and see the order and the reasons.

Built by XONIK. Nothing you enter leaves this page.

Line illustration of use cases scattered at first and ranked into one line, drawn as bars in falling length, cut once.

In brief

This tool helps you choose which AI use case to do first. You score each candidate on the same criteria, from the value of the decision each one improves to data readiness, ownership and risk if it is wrong, and you change the weights to suit your business. It ranks them, explains the leader and runs in your browser.

Score each from 0 to 5. Leave a name empty to leave that column out.

Criterion	Weight	Use case 1: name	Use case 2: name	Use case 3: name	Use case 4: name	Use case 5: name	Use case 6: name	Use case 7: name	Use case 8: name
Value of the decision it improves How to score this Judge the decision the use case improves, not the technology. Lowest score: the decision is rare or changes little when it is made better. Middle score: it is a recurring decision, and the gain is easy to describe but has not been measured. Highest score: the decision is made often, the person who owns it can name what a poor decision costs today, and an improvement would show in business results.	weight Value of the decision it improves	Value of the decision it improves, Use case 1: score	Value of the decision it improves, Use case 2: score	Value of the decision it improves, Use case 3: score	Value of the decision it improves, Use case 4: score	Value of the decision it improves, Use case 5: score	Value of the decision it improves, Use case 6: score	Value of the decision it improves, Use case 7: score	Value of the decision it improves, Use case 8: score
Data readiness How to score this Whether the data it needs exists, can be reached and can be trusted today. Lowest score: the data does not exist, or sits where no one will give access. Middle score: the data exists and access is possible, but it needs cleaning or has known gaps. Highest score: the data is held in one place you can reach, someone vouches for its quality, and you have already used it for something similar.	weight Data readiness	Data readiness, Use case 1: score	Data readiness, Use case 2: score	Data readiness, Use case 3: score	Data readiness, Use case 4: score	Data readiness, Use case 5: score	Data readiness, Use case 6: score	Data readiness, Use case 7: score	Data readiness, Use case 8: score
Ownership available How to score this Lowest score: nobody has agreed to own it. Middle score: a sponsor supports it, but the person who would adjust it, learn when it is wrong and switch it off is not named, or has no time. Highest score: one named person has agreed, has time set aside, and holds the authority to adjust it, trace a single answer back and halt it.	weight Ownership available	Ownership available, Use case 1: score	Ownership available, Use case 2: score	Ownership available, Use case 3: score	Ownership available, Use case 4: score	Ownership available, Use case 5: score	Ownership available, Use case 6: score	Ownership available, Use case 7: score	Ownership available, Use case 8: score
Effort to build into the workflow How to score this Score high when little has to change. Lowest score: it needs changes to several systems and to how several teams work before anyone can use it. Middle score: it fits one team's workflow with changes to one or two systems. Highest score: it fits into a tool the users already work in, needs little change to any system, and the people who would use it have helped define where it sits.	weight Effort to build into the workflow	Effort to build into the workflow, Use case 1: score	Effort to build into the workflow, Use case 2: score	Effort to build into the workflow, Use case 3: score	Effort to build into the workflow, Use case 4: score	Effort to build into the workflow, Use case 5: score	Effort to build into the workflow, Use case 6: score	Effort to build into the workflow, Use case 7: score	Effort to build into the workflow, Use case 8: score
Risk if wrong How to score this Score high when a wrong result is low risk. Lowest score: a wrong result could harm a person, break a rule or cost money that cannot be recovered. Middle score: a wrong result is costly and takes effort to reverse, but would be found before it spreads. Highest score: a wrong result is seen by a person before it counts and is corrected in minutes.	weight Risk if wrong	Risk if wrong, Use case 1: score	Risk if wrong, Use case 2: score	Risk if wrong, Use case 3: score	Risk if wrong, Use case 4: score	Risk if wrong, Use case 5: score	Risk if wrong, Use case 6: score	Risk if wrong, Use case 7: score	Risk if wrong, Use case 8: score
Time to first use How to score this Score high for sooner. Lowest score: no one can say when real users would first have it, or approvals and purchases stand in the way. Middle score: a first version for real users is possible within a planning cycle, subject to one approval. Highest score: a first version for real users could run on a live process within weeks, with the approvals in hand.	weight Time to first use	Time to first use, Use case 1: score	Time to first use, Use case 2: score	Time to first use, Use case 3: score	Time to first use, Use case 4: score	Time to first use, Use case 5: score	Time to first use, Use case 6: score	Time to first use, Use case 7: score	Time to first use, Use case 8: score
Strategic fit How to score this Lowest score: it is unrelated to anything leadership has said matters, and would be a one-off. Middle score: it supports a stated priority, but only indirectly. Highest score: it serves a priority leadership has stated in writing, and would be reused by the other use cases on this list, for example through the data or the way of working.	weight Strategic fit	Strategic fit, Use case 1: score	Strategic fit, Use case 2: score	Strategic fit, Use case 3: score	Strategic fit, Use case 4: score	Strategic fit, Use case 5: score	Strategic fit, Use case 6: score	Strategic fit, Use case 7: score	Strategic fit, Use case 8: score

How the ranking is produced

Name the use cases you are weighing, then give each a score on every criterion, anywhere from the lowest score up to the highest score on the scale you see. Higher always means better here, so give a high score for low effort and for a low risk of a wrong result. Each criterion has anchors that show what a low, a middle and a high score would mean in practice. Each criterion starts with a weight, with the value of the decision and the availability of an owner counting most, and any weight can be changed, down to zero if a criterion should not count. Each use case's weighted score multiplies every score by its weight, adds the results and divides by the total of the weights, which keeps the answer on the scale you used.

Use cases are listed highest first, with level scores sharing a rank. Against the runner-up, the leader is described by its margin, the criteria on which it is ahead and those on which the runner-up is ahead. Each use case is also shown with the two criteria that lifted its score furthest and the one that held it back. The starting criteria and weights come from XONIK's reading of what carries a use case into daily work, taken from the reports on why pilots stall and who owns an AI system after launch. They are not drawn from a study of other organizations.

What this ranking cannot tell you

The ranking reflects your own scores, and scores given in a meeting tend to follow the loudest voice. Ask several people to score separately, then compare their sheets. A use case that ranks first on a weighted average can still fail on one criterion, such as having no owner or no usable data, so read the individual scores as well as the total.

The tool does not size the value of any use case, check that its data exists, or know your constraints. It tells you which candidate your own scores favor and why, and it leaves the decision with you. Its order is not an audit, a certification or a benchmark for your organization. What you enter stays in this browser tab.

Choosing where to start, and the question the ranking cannot settle

An organization that looks at AI soon has a list of ideas, usually ordered by enthusiasm. The tool re-orders that list by what carries a use case into daily work: the value of the decision it improves, an owner, usable data, the effort to fit it in, the harm if it is wrong, how soon real users could have it, and its fit with stated priorities.

It ranks the candidates you give it and explains the order. It does not size the value of any of them, check that the data exists, or know your constraints. Whether to proceed stays with you.

Who scores, and how long does a sitting take?

Describe each candidate in a sentence before anyone scores it, so that everyone is scoring the same thing. A use case called customer service AI will be scored in many different ways. One described as drafting replies to a named kind of query, for review by a named team, will not.

Ask the heads of the functions affected, the technology lead and the sponsor to score separately. A sitting takes about as long as a working session once the candidates are described, and the argument afterward takes longer. Compare the separate sheets and look first at the criteria where people differ widely.

Criterion	Put it to	What earns a high score
Value of the decision	The person who makes that decision today	They can say what a poor decision costs now
Data readiness	The data owner	One reachable place, vouched for, already used
Ownership available	The sponsor, then the named owner	One person has agreed and has time set aside
Effort to build into the workflow	The people who would use it	It sits inside a tool they already work in
Risk if wrong	Risk or the head of the function	A person sees a wrong result before it counts
Time to first use	The delivery lead	Real users, a live process, approvals in hand
Strategic fit	Leadership	A priority stated in writing, and others reuse it

How do you read the order, the margin and the weakest criterion?

The result lists the use cases from highest to lowest, sets the leader against the runner-up, and names for each the criteria that lifted its score most and the one that held it back. Where scores are level they share a rank.

A narrow margin means the scores have not yet chosen for you, and the criteria where the runner-up leads are what to examine. A wide margin deserves a different test. Read the leader's lowest score. A use case can come first on average and still have no owner or no usable data, and in that case it is not first.

A use case that ranks first and has no owner is not first.

From the order to a decision

01
Take the lowest score to its owner
Ask the person who gave it what would raise it, and what that would cost.
02
Give the runner-up a fair test
If the runner-up leads on ownership or data, ask whether starting there and reaching the leader later is the safer path.
03
Decide how many to start
Starting every promising use case at once splits attention. Choose what the owners can carry.
04
Record why the others wait
A line each, with what would bring it back to the list.

When the tool fits, and when it does not

It fits when

You have a list of candidates, each described in a sentence
The people who would own and use them can take part

Look elsewhere when

You have only one candidate; the brief builder is the next step
The question is how the work should be done, which the build, buy or agent tool answers
You need the value of each candidate in money; use the business case calculator on each

A scored ranking trades depth for a clear order. It is meant to end the argument about where to start, and the evidence for each use case still has to be gathered.

What do people ask before using the AI use-case prioritization tool?

Why does the tool not rank the boldest use case first?

Because the boldest one is often not the one that reaches daily use. That one tends to have an owner with time for it, data that can be reached and a place in a workflow people already use. The weights start with the decision and the owner for that reason.

Does the tool work out how much each use case is worth?

No. It does not size the value of any use case, check that its data exists or know your constraints. It ranks the candidates you give it and explains the order. If you need the value of each in money, run the business case calculator on each one.

What if the top-ranked use case has no owner?

Then it is not first. A use case can come first on average and still have no owner or no usable data, so read the leader's lowest score. Take it to the person who gave it, ask what would raise it, and give the runner-up a fair test.

How should we describe each use case before we score it?

In one sentence: what it does, for whom, and who reviews it. Customer service AI gets scored in different ways, while drafting replies to a named kind of query, for review by a named team, does not. Then have the affected heads, technology lead and sponsor score separately.

Why AI pilots stall before productionThe hand-off between the team that proved a model and the team that must run it.Report AI StrategyWhere AI belongs in the business, what to fund first and what to leave alone.Service AI Use-Case Prioritization MatrixA weighted matrix that ranks AI use cases and names each one's weakest criterion.Download Enterprise AI Investment Decision BriefA brief that frames an AI investment for leadership in a few pages.Download

Want help choosing your first use case?

Describe what you are weighing in a few lines. A senior practitioner replies with a suggested first conversation.

Choose your first use case with us More free tools

AI Use-Case Prioritization Tool

Choosing where to start, and the question the ranking cannot settle

Who scores, and how long does a sitting take?

How do you read the order, the margin and the weakest criterion?

From the order to a decision

Take the lowest score to its owner

Give the runner-up a fair test

Decide how many to start

Record why the others wait

When the tool fits, and when it does not

It fits when

Look elsewhere when

What do people ask before using the AI use-case prioritization tool?

Want help choosing your first use case?