# Opportunity map — blank template and worked example

Benian Technologies · September 7, 2026
Source: https://benian.ai/free-audit
Use this yourself or with your team. The worked example below is fictional and is not a client result or a quote.

## Start with observed work

Business / team:
Prepared by:
Measurement period:
Decision date:

Track a normal working week before extrapolating. Note seasonal or unusually busy periods. Leave unknowns blank and name the person who can measure them.

| Candidate job | Trigger → finished result | Volume / month | Minutes / job today | Corrections or exceptions | Tools / data | Evidence source and date |
| --- | --- | --- | --- | --- | --- | --- |
| 1 | | | | | | |
| 2 | | | | | | |
| 3 | | | | | | |

For each job, ask: Is it repeated often? Are the rules clear? Is information available? Does it need AI judgment, a simpler rules-based workflow, or a person?

## Estimate the effort that would remain

Use averages across all jobs, including those handed to a person. Include review, corrections, exceptions, and maintenance. Do not assume every job becomes automatic.

| Candidate | Current minutes / job | Expected remaining minutes / job | Monthly volume | Hours released / month | Evidence for estimate |
| --- | --- | --- | --- | --- | --- |
| 1 | | | | | |
| 2 | | | | | |
| 3 | | | | | |

Hours released = (current minutes − remaining minutes) × monthly volume ÷ 60.
Keep a negative result: it means your estimate adds work. Distinguish time released from payroll or other costs actually removed.

## Put cost and uncertainty beside the benefit

| Candidate | Capacity value / hour | Monthly capacity value | Monthly running cost | One-time build + setup | Dependencies / uncertainty |
| --- | --- | --- | --- | --- | --- |
| 1 | | | | | |
| 2 | | | | | |
| 3 | | | | | |

Monthly capacity value = hours released × the hourly value you assign to that capacity.
Net capacity value = monthly capacity value − monthly running cost.
Capacity-value payback months = one-time cost ÷ positive net monthly capacity value.
If the net value is zero or negative, there is no positive payback under those assumptions. This calculation does not prove a cash saving. Record actual cash savings separately and say what spend would disappear.

Running cost inventory: licences, API or model usage, hosting, monitoring, support, and additional internal maintenance time. Note currency and whether taxes are included. Count each cost once.

## Worked example — fictional invoice-entry workflow

- Current workload: 100 jobs / month, 8 minutes each.
- Proposed workflow: prepare an entry, then a person reviews it.
- Expected remaining effort: 3 minutes / job, including corrections averaged across all jobs.
- Hours released: (8 − 3) × 100 ÷ 60 = 8.33 hours / month.
- Assumed capacity value: $30 / hour → $250 / month.
- Illustrative running cost: $80 / month → $170 net monthly capacity value.
- Illustrative one-time build + setup: $1,200 → 7.06 months of capacity-value payback.
- Cash saving: unknown; the example does not remove payroll or establish extra sales.

If review takes 6 minutes instead of 3, released time falls to 3.33 hours, valued at $100. Net capacity value falls to $20 and payback rises to 60 months. That sensitivity is a reason to test review time before committing.

## Select one small trial

Choose a job with accessible data, a clear owner, measurable results, and a reversible trial. A large theoretical benefit is less useful than a small benefit you can test.

- Selected job and reason:
- Smallest useful output:
- What stays manual:
- Inputs and permissions needed:
- Pilot owner:
- Test period and representative sample:
- Maximum spend / time allowed:
- Success measure and baseline:
- Acceptance threshold agreed before the trial:
- Stop condition and manual fallback:
- Date to review the result:

## Review before expanding

- Actual minutes per job after review and exceptions:
- Error / correction counts and source:
- Actual usage and support cost:
- Jobs requiring human intervention:
- Decision: expand / revise / stop
- Evidence and rationale:
- Next action, owner, and date:

Use the Excel workbook for an ongoing project record: https://benian.ai/resources/ai-project-checklist