What to Automate First: The Impact/Effort Matrix
Direct answer
List your candidate processes, then score each one on impact (time saved, errors avoided, revenue protected) and effort (tools needed, process stability, review cost). Automate the high-impact, low-effort quadrant first. Two quick wins in 30 days build the team's confidence for bigger projects later.
Why prioritization decides the outcome
McKinsey Global Institute estimated that generative AI and existing automation technologies could automate work activities that absorb 60% to 70% of employee time. The potential is enormous, and that is exactly the trap: with so many options, companies pick the flashy project instead of the useful one. McKinsey's State of AI 2025 survey found only 38% of organizations have scaled AI beyond pilots. Stuck pilots usually trace back to the first project choice.
The matrix, made practical
| Quadrant | What it looks like | Examples |
|---|---|---|
| High impact, low effort | Frequent, rule-based, easy to check | Lead follow-up, reminders, meeting notes, standard replies, report drafts |
| High impact, high effort | Big win, needs redesign first | Full customer service agent, quote generation, CRM cleanup pipelines |
| Low impact, low effort | Small fixes when idle | Formatting documents, renaming files, simple data copies |
| Low impact, high effort | Skip it | Rare processes that still change every week |
How to score impact
Rate each candidate on three numbers: minutes spent per week, error cost when it goes wrong, and whether it touches customers. A process that eats 5 team-hours weekly and shows up in every quote cycle scores high. One that happens twice a quarter scores low, however annoying it feels.
How to score effort
Ask three questions. Is the process documented and stable? Are the inputs clean (consistent formats, single source)? Can a human review outputs in under a minute? Three yeses means low effort. Two noes means fix the process first.
The pattern that works
Start with two quick wins in the first quadrant across two different teams, ship them in 30 days, and publish the time saved. Then take one high-impact, high-effort project with the proof from the first two. This sequence also follows the research: most value concentrates in customer operations, marketing and sales, software engineering, and R&D (McKinsey, 2023), so businesses in services will usually find their first win in lead handling or customer replies.
Mistakes that stall automation programs
- Automating a broken process. Clean it first; automation locks in whatever exists today.
- Picking by department politics instead of scored impact.
- Building the biggest project first. Long timelines kill momentum and budget patience.
- No baseline. Record current time and error rates before launch, or nobody can prove the win.
Related reading
- What Is AI Automation: A Guide for Business Owners
- AI ROI: The Real 2026 Numbers
- Before you automate: how to organize your business so AI works
Frequently asked questions
How do I know if a process should be automated?
Run the impact/effort grid: high-frequency, rule-based tasks with clear inputs and outputs are strong candidates. Rare, judgment-heavy, or politically sensitive work stays human-led.
What are the easiest processes to automate first?
Lead follow-up, appointment reminders, meeting notes, standard customer replies, report generation, and data entry between systems. They are frequent, rule-based, and easy to verify.
How long should a first automation take?
A good first project lands in production within 30 days, including testing. If a candidate needs major process redesign first, queue it after two quick wins.
When should I not automate a process?
When the process changes weekly, when ownership is unclear, or when errors cost more than the time saved. Stabilize the process first, then automate.
Want a scoped automation plan with estimated impact before you commit? Talk to inicIA.
Talk to inicIA on WhatsAppBy inicIA
