AI Use Case Prioritization
& AI Strategy Roadmap Workshop
Fewer AI pilots. Better bets. A roadmap your leadership team owns.
AI use case prioritization is the process of deciding which AI opportunities to pursue first, based on business value, feasibility and risk. A prioritization and roadmap workshop helps a leadership team turn a long, scattered list of AI ideas into two or three priority use cases, each with a named owner, a baseline and a measurable outcome, sequenced into a roadmap the organisation can actually deliver.
The problem
Too Many AI Pilots, Not Enough AI Value
Most organisations do not have an AI ideas problem. They have an AI choices problem. Every function has a list of things AI could do, vendors arrive with demos every week, and pilots multiply faster than anyone can evaluate them. The result is activity without impact.
In July 2024, Gartner predicted that at least 30% of generative AI projects would be abandoned after proof of concept by the end of 2025. The reasons it named were poor data quality, inadequate risk controls, escalating costs and, tellingly, unclear business value. Gartner also put the cost of different generative AI deployment approaches at between $5 million and $20 million.
Sources: Gartner (July 2024); BCG, AI Adoption in 2024 (October 2024).
What the leaders do differently
Fewer Bets, Bigger Returns
The clearest evidence that prioritization matters comes from BCG's 2024 survey of 1,000 CxOs and senior executives across 59 countries. The 26% of companies BCG classed as AI leaders pursued, on average, only about half as many AI opportunities as their less advanced peers, and expected more than double the return on investment.
BCG also found that AI leaders generate 62% of their AI value from core business processes rather than support functions. In other words, they do not just pick fewer use cases. They pick use cases closer to how the business makes money.
- 62% from core business processes, such as operations, sales and product
- 38% from support functions
What is the 10-20-70 rule for AI?
The same BCG research is the source of the 10-20-70 rule. Companies leading on AI put roughly 10% of their effort into algorithms, 20% into technology and data, and 70% into people and processes. BCG also found that around 70% of AI implementation challenges sit in that people-and-process layer. For prioritization, the implication is direct: a use case's feasibility depends far more on process ownership and change capacity than on the model.
Our point of view
Prioritization Is a Leadership Decision, Not an IT Backlog
Most AI portfolios are assembled bottom-up, one enthusiastic request at a time. These are the beliefs that keep organisations stuck.
"More pilots means more learning."
Beyond a handful, extra pilots dilute attention, data work and change capacity. Learning comes from finishing and measuring, not from starting.
"Start with what the tool can do."
Tool-first thinking produces impressive demos with no owner. Start from the decisions and workflows that constrain your strategy, then ask where AI changes them.
"AI value can't be measured upfront."
If a use case cannot name the outcome it should move and its current baseline, it is not ready to be prioritised. That discipline alone removes most weak ideas.
Thinking in decisions
Start From the Decision, Not the Technology
Economists Ajay Agrawal, Joshua Gans and Avi Goldfarb of the University of Toronto argue that AI is, at its core, a technology that makes prediction cheap, and that its value shows up in the decisions those predictions improve. In a 2018 Harvard Business Review article, they introduced the AI Canvas, a simple tool for breaking any decision into the components AI needs.
We use a version of this thinking early in every prioritization workshop, because it forces each AI idea to answer a concrete question: which decision does this improve, and how will we know?
| AI Canvas element | The question it asks |
|---|---|
| Prediction | What do we need to know, or estimate, to make this decision better? |
| Judgment | How do we weigh the costs of being wrong in each direction? |
| Action | What will we do differently once we have the prediction? |
| Outcome | How will we know whether the decision was right? |
| Input | What data is available at the moment the decision is made? |
| Training | What historical data, paired with outcomes, do we have to learn from? |
| Feedback | How will results feed back to improve the system over time? |
Elements as described by Agrawal, Gans and Goldfarb (2018). Question wording is our own adaptation for leadership workshops.
The canvas is particularly useful for separating the parts of a decision that can be delegated to AI (prediction) from the parts that must stay with people (judgment and accountability), which is the same distinction at the heart of the 10X Humane AI Lens.
Our method
How We Prioritize AI Use Cases: Value, Feasibility and Risk
Every candidate use case is scored on three dimensions, by the people who will own the outcome, not just by the technology team.
- Value: how much the use case moves a business outcome that matters, such as revenue, cost, cycle time, quality or customer experience, and how close it sits to the core of the business.
- Feasibility: whether the data, tools, skills, process ownership and change capacity exist to deliver it in the next two quarters.
- Risk: what could go wrong for customers, employees, data, compliance or reputation, tested through the 10X Humane AI Lens.
Value and feasibility place each use case on a simple matrix. Risk then acts as a filter: a high-value, high-feasibility idea with unacceptable risk is redesigned or dropped, not fast-tracked.
Strategic bets
Worth pursuing deliberately. Invest in data and capability, and stage the work.
Priority use cases
Start here. Two or three, each with an owner, a baseline and a target.
Park
Revisit only if strategy, data or tools change materially.
Quick wins
Useful for building confidence and skills, but don't mistake them for strategy.
An illustrative scoring example
The table below is a hypothetical example of how a mid-sized services company might score five candidate use cases on a 1–5 scale. Real scores depend entirely on your strategy, data and context.
| Candidate use case (hypothetical) | Value | Feasibility | Risk | Decision |
|---|---|---|---|---|
| Drafting first-pass client proposals | 4 | 5 | Low | Priority |
| Forecasting demand for delivery staffing | 5 | 3 | Medium | Strategic bet |
| Summarising internal meeting notes | 2 | 5 | Low | Quick win |
| Automated screening of job applicants | 3 | 4 | High | Redesign before proceeding |
| Fully automated pricing decisions | 4 | 2 | High | Park |
Notice that the two highest-risk ideas involve decisions about people and prices, where errors are costly and accountability matters most. That is where human judgement should stay firmly in the loop, even when the technology looks ready.
Turning this into an AI use case prioritization template
A practical prioritization template needs only a few columns, as long as every row is filled in honestly by the people who will own the outcome:
- Use case and the decision or workflow it changes
- Business outcome and current baseline: the metric it should move, and where it stands today
- Value score (1–5) and feasibility score (1–5), with a one-line reason for each
- Risk rating after applying the 10X Humane AI Lens
- Owner, target, stop criteria and decision: priority, strategic bet, quick win, redesign or park
Leadership teams that attend our workshop receive our full scoring template. Ask us for a copy if you would like to see it.
Rules of thumb
What Is the 30% Rule for AI?
The "30% rule for AI" is widely searched, but it has no single authoritative definition. Different writers use it to mean different things: some describe AI handling around 30% of the work while people keep the higher-judgement 70%; others describe the reverse split; none we have found traces the rule to an original study.
The best-documented "30%" in this space is Gartner's July 2024 prediction that at least 30% of generative AI projects would be abandoned after proof of concept by the end of 2025, largely because of poor data, weak risk controls, rising costs and unclear business value.
That is the logic behind our prioritization method: rather than applying a fixed percentage, every use case is judged on the outcome it should move and on what must stay human, which is exactly what the 10X Humane AI Lens is designed to test.
Where to look first
Common AI Use Cases Worth Evaluating
These are among the most common AI use cases organisations evaluate today. Common does not mean right for you: each still needs scoring against your strategy, data and risk.
Document and content drafting
Proposals, reports, policies and marketing content, drafted by AI and refined by people.
Knowledge search and summarisation
Answering questions from internal documents, policies and past work in seconds.
Customer service assistance
Drafting responses, triaging tickets and surfacing answers for service teams.
Forecasting and planning
Demand, staffing and cash-flow forecasts that support better planning decisions.
Document and process automation
Extracting and checking data from invoices, contracts and forms.
Software and analytics support
Helping teams write code, build queries and turn data into dashboards faster.
For examples across nine business functions, see our guide to AI for business leaders.
What the workshop covers
Inside the Prioritization & Roadmap Workshop
Five working sessions in which your leadership team builds the portfolio and roadmap itself, using its own strategy, data and constraints.
Opportunity Discovery Across the Value Chain
A structured scan of where AI could change decisions and workflows in each function, starting from strategic priorities and current constraints rather than tools.
Exercise: building a long-list of AI opportunities by functionDecision Mapping
Translating each opportunity into the decision it improves: the prediction needed, the judgement that stays human, the action and the outcome that proves success.
Exercise: mapping priority decisions with an AI Canvas-style templateScoring Value, Feasibility and Risk
Scoring candidates as a leadership team, surfacing disagreements early, and applying the 10X Humane AI Lens to every idea that touches customers, employees or sensitive data.
Exercise: scoring and placing use cases on the prioritization matrixPortfolio Balancing and Sequencing
Choosing two or three priority use cases, balancing confidence-building wins with strategic bets, and sequencing work against data readiness and change capacity.
Exercise: agreeing the portfolio and what the organisation will stop doingRoadmap, Owners and Measures
Turning priorities into a roadmap with named owners, baselines, success measures, stop criteria and a review rhythm for the leadership team.
Exercise: building the 30-day action plan and the 6–12 month roadmapOutcomes
What Your Leadership Team Walks Away With
Decisions, not just ideas, agreed and owned by the people accountable for results.
An AI opportunity long-list
Opportunities across functions, each linked to the decision or workflow it would change.
A scored, risk-checked portfolio
Every use case placed on the matrix and tested for risk, with clear reasons for each decision.
2–3 priority use cases
Each with a named owner, a baseline, a target outcome and explicit stop criteria.
A roadmap and review rhythm
A 30-day action plan inside a 6–12 month roadmap, with a cadence for reviewing progress.
Where this fits
Part of Our Leadership Programmes and Advisory
Use case prioritization is built into our leadership programmes and is the core of our Executive Alignment advisory. It can also run as a focused workshop for an existing leadership team.
AI for Executive Teams
For leadership teams prioritising AI together and agreeing owners, guardrails and a shared roadmap.
Explore the programme →AI for Leaders
For CXOs and senior leaders who need to evaluate AI opportunities and vendors with confidence.
Explore the programme →Executive Alignment Advisory
A structured advisory engagement across 10 focused working sessions to set AI direction, priorities and governance in depth.
Learn about our advisory →FAQ
Frequently Asked Questions
Common questions about AI use case prioritization and AI roadmaps.
What is AI use case prioritization?
AI use case prioritization is the process of deciding which AI opportunities an organisation should pursue first. It usually involves scoring candidate use cases on business value, feasibility and risk, then selecting a small number to deliver, each with an owner and a measurable outcome.
How do you prioritize AI use cases?
Start from business priorities and the decisions or workflows that constrain them. List candidate use cases, define the outcome each should move, score them on value, feasibility and risk, place them on a value-feasibility matrix, and choose two or three to pursue first. Revisit the portfolio regularly as results come in.
What is an AI strategy roadmap?
An AI strategy roadmap is a sequenced plan for delivering an organisation's priority AI use cases over time. It is different from the "AI learning roadmaps" for engineers that often appear in search results. A good roadmap names owners, baselines, success measures and stop criteria, accounts for data readiness and change capacity, and includes a regular leadership review.
How many AI use cases should a company start with?
Usually two or three. BCG's 2024 research found that companies leading on AI pursued about half as many opportunities as their peers and expected more than double the return. A small, well-owned portfolio is easier to deliver, measure and learn from.
Why do AI pilots fail to scale?
Common reasons include unclear business value, no named owner or baseline, poor data quality, weak risk controls and rising costs. Gartner cited several of these when predicting that at least 30% of generative AI projects would be abandoned after proof of concept by the end of 2025.
Who should attend a prioritization workshop?
The leaders who will own the outcomes: typically the CEO or business head, functional leaders, and the heads of technology, data and risk. Prioritization works best when the people accountable for results make the choices together.
Do we need a data or AI team in place first?
No. The workshop helps you decide where AI should create value before you invest further. Feasibility scoring will show where data, skills or partners are needed, so investment follows priorities rather than the other way round.
What is the AI use case prioritization framework?
An AI use case prioritization framework is a structured way to compare AI opportunities. Most frameworks score each use case on business value and feasibility, often plotted on a two-by-two matrix, with risk applied as a filter. Our version adds the 10X Humane AI Lens to test what must stay human.
What is the 10-20-70 rule for AI?
The 10-20-70 rule comes from BCG's research on AI adoption: companies that get value from AI put about 10% of their effort into algorithms, 20% into technology and data, and 70% into people and processes.
What is the 30% rule for AI?
The 30% rule for AI is an informal rule of thumb with no single authoritative definition; it is usually used to describe how work might be split between AI and people. The best-documented 30% figure is Gartner's prediction that at least 30% of generative AI projects would be abandoned after proof of concept by the end of 2025.
What are 5 common use cases for AI?
Five of the most common business use cases are document and content drafting, knowledge search and summarisation, customer service assistance, forecasting and planning, and document or process automation such as invoice processing. Each should still be scored against your own strategy, data and risk.
Is there an AI use case prioritization template?
A useful template lists each use case with the decision it changes, the outcome and baseline, value and feasibility scores, a risk rating, an owner, a target and stop criteria. Leadership teams attending our workshop receive our full scoring template.
Sources
- Gartner, Gartner Predicts 30% of Generative AI Projects Will Be Abandoned After Proof of Concept by End of 2025 (July 2024)
- BCG, AI Adoption in 2024: 74% of Companies Struggle to Achieve and Scale Value (October 2024)
- Agrawal, Gans and Goldfarb, A Simple Tool to Start Making Decisions with the Help of AI, Harvard Business Review (April 2018)
Stop Collecting AI Pilots.
Start Choosing.
Tell us where AI activity stands in your organisation today. We'll recommend a prioritization and roadmap workshop shaped around your leadership team.
Request a Workshop Consultation