Establish Baselines, Targets, and Acceptance

What you will complete

Complete the Portfolio Measurement Plan with baseline, target, source, owner, window, acceptance test, and next-decision rule.

Portfolio rule: A projected benefit is an assumption until a defined baseline and post-change observation can be compared.

Measure the constraint

Choose measures tied to the workflow: time, delay, volume, error, rework, response, conversion, intervention, exception, adoption, cost, or customer experience. State the population, period, source, calculation, and owner. A metric without those fields cannot support a portfolio decision.

Distinguish planning from proof

A transparent estimate can support prioritization when its inputs and assumptions are visible. It does not become a savings claim. The Flowgrammer process guide's recovery range is explicitly a planning assumption. Use local evidence where available and record uncertainty.

Write acceptance before build

Define normal, incomplete, conflicting, duplicate, exception, integration-failure, and rollback cases. State the destination evidence that proves each result. Connect targets to continue, revise, pause, or stop decisions rather than celebrating activity alone.

Worked example: Harbourline Business Services

Fictional teaching example: Harbourline records fictional baselines for classroom use: lead-review delay, invoice rework count, and time spent assembling executive reports. These numbers are not customer data. Lead qualification receives a target and acceptance cases only after the owner confirms the baseline source. The chatbot remains unmeasured and cannot claim value.

Harbourline is not a Flowgrammer client. Its people, projects, weights, scores, costs, metrics, and outcomes are fictional learner inputs. Replace them with evidence from your organization and retain the evidence source beside each decision.

Workbook exercise

Complete section 5 of the 90-Day AI Automation Portfolio Workbook.

1. Choose one direct measure for each active initiative.
2. Record population, period, data source, calculation, and owner.
3. Label every estimate and sample metric as assumed or fictional where appropriate.
4. Write acceptance cases and continue, revise, pause, or stop thresholds.

Ask the affected owner to challenge one assumption. Record the original entry, revision reason, decision owner, and review date rather than silently replacing the history.

Common failure modes

• Using revenue without an attribution method.
• Presenting an estimate as measured savings.
• Measuring workflow runs but not destination outcomes.
• Choosing targets after seeing the result.

A portfolio should expose weak evidence and constrained capacity. Repair the responsible layer instead of increasing the score or compressing the schedule.

Capstone connection

Complete the Portfolio Measurement Plan with baseline, target, source, owner, window, acceptance test, and next-decision rule. The section must be specific enough for leadership, affected operators, and an implementation reviewer to inspect. It remains planning evidence rather than proof of implementation or outcome.

Check your application