Plan benefit and effort measurement

1. Specify the measure supporting the decision

In fictional fund-reconciliation automation, the sponsor wants to release at least 35 monthly hours for exception analysis. Before the pilot, the BA must turn that intent into a verifiable measure: which activity becomes unnecessary, for which population, over which period, at what quality, and which additional work enters the calculation.

“Time saved” can mean less customer waiting, fewer human work minutes, or a forecast valued in euros. These quantities are not interchangeable. This example measures human work within the same month and organizational scope while maintaining required quality; it does not measure processing duration or salary reduction.

A measurement contract records the unit, population, reference, formula, justified exclusions, data origins and handling of incomplete observations. It includes who collects, who confirms meaning and who decides when a gap remains. The 35-hour target is a fictional condition supplied for the exercise, not a PMI or bank rule.

When detailing the formula, validate with Business, APS and finance whether each term represents the same effect and whether a value already includes effort you are about to subtract.

2. Count additional work without duplication

The pilot releases 60 gross hours. Monitoring reporting contains 16 hours and rework reporting contains 12. Reconciliation by person, task and interval confirms that six hours are the same work episodes recorded in both categories.

Distinct effort is 16 + 12 − 6 = 22 hours; net gain is 60 − 22 = 38. Adding categories without reconciliation would give 32 net hours and could change the decision against the 35-hour target. Identity concerns human work: two people working simultaneously produce two effort contributions even if they share one calendar hour.

Conversely, renaming the same episode does not create new effort. If overlap were still unknown but could range from zero to twelve hours, net gain would lie between 32 and 44. These are deterministic bounds constructed from supplied data, without probabilities.

Proving at least three overlapping hours would suffice to demonstrate the 35-hour target if the policy requires only that bound rather than an exact value. Retain the origins of both lists and the reconciliation rule for later review.

3. Distinguish combined outcomes and marginal contributions

For the same horizon and population, a model forecasts that initiative A alone releases 40 hours and B alone releases 30. Together they release 50 because they replace some of the same work. Adding standalone forecasts gives 70, which does not represent the combined portfolio.

If B is already in the reference, adding A increases the modeled outcome from 30 to 50: a marginal contribution of 20. If A is already in the reference, adding B increases it from 40 to 50: ten. The two marginal contributions do not uniquely decompose the 50 hours and should not be added as though they explained the entire outcome.

Credit allocation is another decision requiring an agreed convention; these figures impose neither a 20/10 split, a 40/30 split nor budget-proportional allocation. In the measurement plan, identify which initiatives belong to the reference and how interactions and scope changes will be observed.

A measured combined result can support the enterprise total without identifying each project’s causal contribution. Also distinguish a scenario forecast from an actual measurement: executing a formula does not establish that initiatives produced those effects.

4. Preserve relationships among assumptions

A forecast contains only two admissible scenarios for next month. Under low adoption, gross gain is 60 hours and additional effort is 20, giving 40 net. Under high adoption, gross gain is 100 and effort is 50, giving 50 net.

This model’s net range is 40 to 50. Calculating 60 − 50 = 10 and 100 − 20 = 80 creates combinations excluded by the model. Those extremes would be admissible bounds only if the combinations were also possible; appearing conservative does not automatically make them better.

Record the dependency instead of retaining only each column’s minimum and maximum. Without scenario probabilities, there is no basis for calling 45 hours the expected value. If stakeholders later admit a third scenario with 60 gross and 50 additional hours, the boundary includes ten net hours and the decision may change.

That change requires revisiting the assumption, version and forecast communication. Model bounds do not guarantee the actual next month: assumption quality and coverage of plausible situations still require assessment.

5. Collect information that can resolve the decision

Before measurement begins, define which information permits analysis to finish. A report containing 100 operations may not identify effort overlap; a smaller record linked to tasks may resolve precisely that gap. Data quantity and decision usefulness are different.

In the example with 60 gross hours and 16/12 additional hours, the target is 35 and gain lies between 32 and 44. An authorized check confirms at least four overlapping hours without determining the total: the bound becomes 36 to 44, sufficient to demonstrate the threshold.

The fictional policy requires proof of the bound rather than a point estimate. The plan must retain that distinction rather than present 36 as an exact outcome. If the objective also requires knowing capacity available for work schedules, greater precision may remain useful.

Investigation cost does not disappear: record it within the appropriate horizon and measure, specifying whether it enters the same operational target or a separate investment decision. Do not subtract euros from hours or mix one-time collection effort with monthly effort without an explicit rule.

6. Plan evidence for every mandatory condition

A fictional expansion decision simultaneously requires at least 35 net monthly hours and recovery evidence for the expansion scope. Gain may be demonstrated while recovery remains unexecuted. No average readiness percentage replaces these two conditions.

The target measures recurring operation and excludes one-time analysis effort, which is recorded separately. Suppose only six work hours are available before the committee. Two authorized activities exist: reconciling overlap requires two hours and resolves the benefit bound; executing and reviewing recovery requires four and resolves the technical condition if its outcome is compliant.

An additional report on average value requires four hours but leaves recovery unknown. The two-plus-four-hour plan covers both gaps within available capacity, assuming the supplied owners and dependencies. Planning the activity guarantees neither its outcome nor authorization to expand.

If recovery fails, communicate the discrepancy and next action; if there is insufficient time to obtain both pieces of evidence, expose the limitation and revisit timing or scope with the designated authority. Schedule pressure gives no team an automatic waiver.

7. Prepare observation and action after handover

The measurement contract must survive the project team. APS may provide task and operational records, the benefit owner may confirm whether released hours still serve the need, and finance may classify expenditure effects. Plan frequency, authorized access, category reconciliation, definition controls and authority to act.

A tooling change that records the same work under two categories may alter reporting without changing actual work. A process change may instead genuinely increase effort. Preserve the ability to distinguish these situations.

Use a versioned reference, period description and indication of missing evidence. If a metric changes from net to gross, do not automatically join both series as though they measured the same thing. In an international meeting, begin with the decision and scope: “The lower bound meets the benefit target for this period; recovery evidence remains outstanding.”

Then identify an owner, deadline and completion condition for each action. A training platform does not certify an actual operation’s readiness.

8. Exercise: write the plan and recommendation

Before opening the worked reasoning, prepare a measurement contract and committee note. Use 60 gross hours, 16 monitoring hours, 12 rework hours and initially unknown overlap from zero to twelve. The target requires at least 35 hours; recovery is mandatory and has not been executed.

You have six hours for the two- and four-hour activities described above, with no resource contention or other dependencies in this case. Solution: the initial bound is 32 to 44, so gain has not yet been demonstrated. Two-hour collection confirms at least four overlapping hours and establishes 36 to 44; it supplies no exact value.

Recovery takes the other four hours and yields a compliant result for the specified version and scope. Both conditions are now demonstrated in the exercise, but the authority still needs to decide. Write what changes if recovery fails or confirmed minimum overlap is two: the first creates a technical discrepancy; the second gives a 34-to-44 bound and gain remains inconclusive.

In the downloadable material, change the data and compare outputs with your prediction. Keep calculation, observed evidence, model limits and human decision separate.

Common pitfalls

Add duplicated effort; combine incompatible assumptions; sum marginal contributions as a total; confuse a sufficient bound with an exact measurement; plan an activity as though its result were already confirmed.