Prioritize your AI projects in 2 hours with the Impact / Complexity matrix
Faced with the proliferation of possible AI use cases, 78% of SMBs struggle to prioritize and end up launching too many projects in parallel — or none at all. The Impact / Complexity matrix is a simple, proven tool that lets you classify your ideas into 4 quadrants: Quick Wins (high impact, low complexity), Strategic projects (high impact, high complexity), Incremental improvements (low impact, low complexity), and Bad ideas in disguise (low impact, high complexity). In a 2-hour workshop, you obtain a 12-month roadmap with clear priorities.
The problem
After an initial AI audit or an exploration phase, SMBs and mid-market companies often find themselves with a list of 10 to 20 AI project ideas. Customer service chatbot, accounting automation, demand prediction, sentiment analysis on customer reviews, marketing content generation... The possibilities seem endless, but resources (budget, skills, managerial bandwidth) are limited.
Without a prioritization method, companies fall into three classic traps:
- The "sexy" project syndrome: the company chooses the most technically impressive project (generative AI, computer vision) rather than the most profitable one. The result: a spectacular POC that is unusable in production.
- Spreading too thin: to avoid upsetting anyone, management launches 5-6 projects in parallel with insufficient budget for each. None reaches the critical mass to generate value.
- Inertia: faced with the difficulty of choosing, the company launches nothing and falls behind its competitors. Based on our field observations, the time between the first AI reflection and the first deployed project often exceeds a year in SMBs.
The consequence is concrete: based on our client feedback, SMBs that methodically prioritize their AI projects achieve significantly higher ROI than those that proceed opportunistically.
The AI solution
The Impact / Complexity matrix is a visual prioritization tool that positions each AI project on two axes. It turns a disorganized list into a structured action plan. Here are the three key components of the method.
Business impact scoring
Each project is scored from 1 to 5 on four impact criteria: time savings (hours/week saved), financial gain (additional revenue or avoided costs), quality improvement (error reduction, customer satisfaction), and competitive advantage (market differentiation). The impact score is the weighted average of these four criteria.
Complexity scoring
Each project is scored from 1 to 5 on four complexity criteria: data availability and quality, technical skills required, integration with the existing IT system, and change management needed. A score of 1 means "feasible in 2 weeks with current resources," a score of 5 means "requires new skills and infrastructure."
The 4 action quadrants
Quick Wins (high impact, low complexity): launch immediately, target 0-3 months. Strategic projects (high impact, high complexity): plan with dedicated resources, 6-12 month horizon. Improvements (moderate impact, low complexity): integrate gradually. Avoid (low impact, high complexity): don't invest, reassess later.
Implementation
Here is the step-by-step method to build your Impact / Complexity matrix and derive an operational AI roadmap from it. Plan for a half-day workshop with key stakeholders.
List and describe the use cases (30 min)
Gather all the identified AI use cases (audit, brainstorming, field feedback). For each, write a half-page sheet: description of the current process, envisioned AI solution, expected benefit, data required. Aim for 8 to 15 use cases for a productive exercise. Invite the relevant business managers to ensure relevance.
Score collectively (1h30)
In a workshop, score each use case on the 8 criteria (4 impact, 4 complexity). Use simultaneous voting (each participant scores independently, then discuss gaps > 2 points). This method avoids the halo effect and authority bias. Calculate the average impact and complexity score for each use case. Position them on the 2×2 matrix.
Build the 12-month roadmap (1h)
Turn the matrix into a time-based action plan: Q1 (months 1-3) = 2-3 Quick Wins to create momentum and prove value; Q2 (months 4-6) = 1-2 tactical projects with intermediate impact; Q3-Q4 (months 7-12) = 1-2 strategic projects with high impact. Assign a business sponsor and an estimated budget to each project. Schedule quarterly reviews to adjust. Discover our AI roadmap support offering to be guided through this exercise.
Results
Frequently asked questions
How many AI projects should an SMB roadmap contain?
A pragmatic AI roadmap for an SMB contains 6 to 12 projects over 12 to 18 months, split into 3 horizons: 2-3 quick wins (0-3 months), 3-4 tactical projects (3-9 months), 2-3 strategic projects (9-18 months). Beyond 12 projects, the plan becomes unrealistic for an organization of 50 to 500 employees.
How do you score the impact of an AI project without historical data?
Use the proxy method: estimate the volume of tasks involved (number/week), the average time per task, and the realistic automation rate (30-60% for a first project). Multiply to get the gain in hours/week. Even if approximate, this estimate is far more useful than intuition alone.
Should you redo the Impact / Complexity matrix regularly?
Yes, review your matrix every quarter. Priorities evolve with the market, feedback from the first deployed projects, and the team's growing skills. A project deemed complex in Q1 may become simple in Q3 thanks to the experience gained.
What is the difference between an AI roadmap and an AI master plan?
The AI master plan is a high-level strategic document (vision, governance, overall budget) spanning 2-3 years. The AI roadmap is a detailed operational action plan over 12-18 months with prioritized projects, milestones and allocated resources. The roadmap flows from the master plan.
For technical profiles
Comparison of tools for building and tracking an AI roadmap:
| Criterion | Spreadsheet + manual matrix | Notion / Airtable | Dedicated tool (Productboard / Aha!) |
|---|---|---|---|
| Cost | Free | 0-20 euros/month | 50-200 euros/month/user |
| Matrix visualization | Basic XY chart | Kanban view + timeline | Native matrix + roadmap |
| Collaboration | File sharing | Real time | Real time + workflows |
| Automated scoring | Manual formulas | Formulas + filtered views | Native scoring + weighting |
| Progress tracking | Manual | Statuses + automations | Gantt + dependencies |
| Suitable for | First exercise, < 10 projects | SMBs, 5-15 projects | Mid-market, structured AI program |
Recommended scoring template: create a spreadsheet with the following columns: Use case name | Description | Business sponsor | Impact score (1-5): time gain, financial gain, quality, competitive advantage | Complexity score (1-5): data, skills, integration, change | Final score (Average impact / Average complexity) | Quadrant | Horizon (Q1/Q2/Q3-Q4) | Estimated budget | Status.