---
title: "Six efficiency indicators to measure before and after AI"
slug: tablero-eficiencia-operacion
lang: en
situation: medir
format: plantilla
minutes: 25
updated: 2026-10-19
---

# Six efficiency indicators to measure before and after AI

> For managers about to bring AI into a process who want to be able to show, at 90 days, what changed and what did not.

## When to use it

- You are about to start a pilot and still do not have the before number.
- You are asked whether the tool "is working" and all you have is the team's opinion.
- The results look good on one indicator and you want to check whether another got worse.

## The sheet

Process: [ ] · Unit processed: [invoice, request, ticket, contract] · Baseline date: [ ] · Measurement date: [ ]

| Indicator | Formula | Before (baseline) | After (90 days) | Data source | Owner |
|---|---|---|---|---|---|
| Cycle time | Time from when the unit comes in until it is closed (average). | [ ] | [ ] | [system or log] | [ ] |
| Cost per unit processed | Total process cost for the month ÷ units processed in the month. Includes licenses. | [ ] | [ ] | [ ] | [ ] |
| Rework / error rate | Units corrected or returned ÷ units processed. | [ ] | [ ] | [ ] | [ ] |
| Human review hours | Hours per month spent reviewing, validating or correcting. | [ ] | [ ] | [ ] | [ ] |
| Volume per person | Units processed in the month ÷ people working in the process. | [ ] | [ ] | [ ] | [ ] |
| Internal or external customer satisfaction | Average rating on one fixed question, with the same scale before and after. | [ ] | [ ] | [ ] | [ ] |

Illustrative example: "Purchase request cycle time: before 6 days, after 4. Review hours: before 20 h/month, after 34. The time went down, but review went up."

## How to fill it in 25 minutes

1. Define the unit before the indicators. If the unit changes between before and after, the comparison is useless.
2. Pull the baseline from a system or log, not from memory. If none exists, measure for two weeks before starting.
3. Use the same formula, the same source and the same period in both measurements.
4. Name an owner per indicator: whoever pulls the data, not whoever presents it.
5. The most common mistake: measuring only cycle time and presenting the improvement without looking at the other five rows.
6. Read the whole table before concluding. One indicator improving while another worsens is a finding, not a success.

## The line that is almost always missing

Human review hours. If they go up, the savings may have moved rather than disappeared: the team stops doing the work and starts reviewing what the tool does. Compare the hours freed with the new review hours before reporting any savings.

## If this becomes a big decision

If the results are going to support a larger investment or a structural change that carries your name, the free tools at vicsol.org/decision-clarity/ help you structure it. They do not replace a formal review.

---

Vic Sol · Decision Desk · vicsol.org/resources/ · Free to use with attribution.
This sheet structures a conversation; it is not legal, financial or audit advice.