For CFOs, CEOs and owners
How to Evaluate AI ROI Before Approving a Pilot
Separate time saved from cash saved. Use five practical questions to evaluate an AI pilot, its full cost and the evidence needed before investing further.
For COOs, CEOs and operating leaders
ZTelligence · Published October 28, 2026 · Executive field guide
The team is busy, customers are waiting and another hire feels overdue. Hiring may be the right answer. First, find out whether the constraint is the amount of work or the way work moves.
More people can help with sustained demand. They may do little for a queue that sits behind one approver or a job that repeatedly arrives without the information needed to finish it.
Choose a recurring outcome: a customer quote, a new employee ready for work, or a completed job ready to invoice. Follow a small, representative sample through the actual process. Ask the people doing the work to show you where they wait, re-enter information or send something back.
Record both active working time and elapsed time. A task that takes twenty minutes of effort can still take several days to complete. Those are different problems, and the solution should reflect the difference.
Do this with the team, not as a ranking of individual performance. People often compensate for broken handoffs in ways that are invisible in a management report.
Illustrative example: a field-service company finishes jobs on time, but finance cannot bill until supervisors supply approved changes and completion details. A new billing employee may spend more time chasing the same missing records. A clearer completion checklist and named exception owner may address the underlying delay.
AI could help prepare a missing-information summary or draft a follow-up for review. It cannot make an unapproved change order approved, or replace an accountable operating decision. First agree which record is authoritative and who resolves an exception.
For each bottleneck, compare clearer responsibilities, fewer unnecessary approvals, better use of current software, targeted automation and additional people. Some constraints require expertise, physical capacity or customer judgment. Keep those visible rather than forcing every problem into an AI project.
Estimate the effort to change the process as well as the benefit. A small improvement in a frequent task may be more useful than an impressive demonstration in work that rarely occurs.
Choose one change and compare results with the baseline. Track end-to-end completion time, work waiting, rework, customer impact and the effort needed to supervise the new process. Check that the queue has not simply moved to another team.
Ask the manager what the team will do with any capacity released. Faster completion, fewer late evenings and room for more customers are distinct outcomes. Report the one you actually observe.
For a concrete cash-related example, read how to find completed work that has not become an invoice. To prioritize several bottlenecks, explore the Z Value Creation Scan.
For CFOs, CEOs and owners
Separate time saved from cash saved. Use five practical questions to evaluate an AI pilot, its full cost and the evidence needed before investing further.
For COOs, CEOs and operating leaders
A practical capacity review for growing businesses: find waiting, rework and unclear handoffs before deciding where hiring, process changes or AI can help.
For PE operating partners and portfolio company leaders
An executive framework for testing AI opportunities in a portfolio company: establish the baseline, validate a focused pilot and decide what deserves expansion.
For owners and CEOs
Find the decisions, relationships and recurring work that still depend on the founder. Start with one practical handoff.