Every department in your company is spending on AI. The question is whether any of them can prove what they're getting for it.
What Does AI Spend Look Like by Department?
The typical pattern across mid-market companies:
- Engineering — highest total spend ($5K-$20K/mo). Copilot, Cursor, API calls. Highest utilization rate (70-90%). Lowest waste.
- Marketing — moderate spend ($2K-$8K/mo). Jasper, ChatGPT, Midjourney, Copy.ai. Highest waste rate (30-50%) due to tool overlap and low adoption.
- Sales — moderate spend ($2K-$6K/mo). Gong, ChatGPT, outreach AI. Mixed utilization.
- Operations/Admin — low visible spend ($500-$2K/mo). Highest shadow AI risk — expenses are hidden in miscellaneous categories.
How Do You Benchmark Department AI Spend?
Grade each department on four metrics: utilization rate (% of licenses actively used), waste rate (% of spend with no measurable output), spend per employee, and ROI clarity (can they articulate what the tools produce). The department with a D grade is where you start cutting — or investing in training.