LEARN. PRACTICE. GROW.1. Start from the decision and its useful deadline
In a fictional middleware migration, a reservation can be cancelled without cost until Thursday at 12:00. The decision needs eight working hours of review, available from 09:00 to 17:00 with no holidays. Delivering the packet on Wednesday at 16:00 permits completion only on Thursday at 16:00: one hour on Wednesday and seven on Thursday. Delivery on Wednesday at 12:00 allows five hours that day and three on Thursday, finishing at 12:00. This calculation uses the case’s explicit calendar. An analysis completion date may satisfy a document schedule yet arrive too late to influence the decision. Record the decision, its last useful date, minimum evidence, time needed to assess it and decision owner. Account for time zones when teams work in different countries; do not resolve them by assuming everyone uses the same local time.
2. Represent dependencies and availability
Two people start independent tasks at 10:00. Confirming the inventory takes two hours; analyzing exceptions takes three. Synthesis takes one hour and starts only after both finish. Its earliest completion is 14:00, not 16:00: the initial tasks overlap. If a half-hour review can start only at the 15:00 session, earliest approval is 15:30. With a single person whose exclusive attention each task requires, overlap is no longer allowed. Distinguish precedence, effort and resource availability. Plan near-term work in greater detail and identify what may change after new evidence. This supports controlled replanning without pretending that a sequence drawn before discovery already contains every answer. Do not confuse a technical dependency between requirements with the availability of the people analyzing them.
3. Specify what an approval permits
A conditional approval may authorize a pilot for version 3 and a defined group, provided an owner confirms reconciliation before Friday at 16:00. Under that contract, the signed document does not authorize version 4, every customer or confirmation after the deadline. Model the condition as a checkable rule: scope, version, evidence, accepting person and temporal validity. Distinguish funding approval, agreement on requirements and operational acceptance; each answers a different decision. An editorial update with no impact may follow a simple process, but its classification must also follow agreed governance. If a condition ceases to hold, record the need for a new decision by the appropriate authority. Do not replace that decision with a note that artificially retains approved status.
4. Apply bounded delegations
In the example, the operations owner is absent. Written delegation lets a deputy accept pilots for up to 30 users provided the recovery requirement is unchanged. A 24-person pilot with the same requirement fits; a 24-person pilot with a changed recovery requirement does not. Headcount and decision type are independent limits. Experience does not automatically create authority, but a valid delegation should not be ignored merely because the principal is absent. If the case requires two independent approvers, one person holding two roles does not count as two people. Plan who may decide, how conflicts are resolved and what happens when nobody with authority is available. These are local governance examples, not universal CBAP requirements.
5. Preserve context as information circulates
An evidence file may contain restricted data. The committee needs the assessment outcome, but not every member may open detailed records. Plan an approved view showing version, scope, conclusion, limitations and a reference for an authorized reviewer without improperly distributing data. A link does not establish recipient access or that a screenshot remains current. Retain the relationship between a decision and the information actually reviewed. A conclusion about one consumer on version 3 does not automatically transfer to another consumer or version. For relevant changes, assess impact before reusing the conclusion. A summary view can be sufficient for the decision if policy and purpose permit; distributing the whole repository is not an objective in itself. If evidence must be inspected again, retain the applicable version in an authorized location. Keeping only a filename or cryptographic digest of a deleted file does not retain its content. A digest helps compare a candidate copy but does not recover information that no longer exists.
6. Measure analysis work with useful denominators
The team returned 12 of 40 packets in one window and 15 of 100 in the next. Absolute counts increased, but the proportion fell from 30% to 15%. That describes the two samples. Attributing the difference to a new checklist would require assessing changes in scope, difficulty, criteria and team, among other relevant explanations. Define the observed unit before comparing: returned packets, corrected requirements and rework hours are different measures. If a packet returned three times, a proportion of affected packets counts it once; a correction-cycle measure may count three events. Use confirmed causes to guide improvements. Delayed reviewer assignment calls for a different response from contradictory rules discovered during review. The first may call for reassigning reviewers; the second may change how rules are confirmed. For an effort measure, classify why work occurred. If nonoverlapping categories total 80 drafting hours, 20 planned-review hours, ten defect-correction hours and ten approved-new-requirement hours, total effort is 120. The defect-correction fraction is 10/120, about 8.33%. A legitimate scope change does not become a defect merely because it followed the first version.
7. Choose investigation that can change the decision
There are 90 minutes of expert time before signing a non-cancellable solution contract. A 60-minute session resolves whether the solution meets a mandatory recovery limit; reviewing a visual model takes 30; confirming a color preference takes 30. The visual model can be revised later without material cost. The recovery constraint can eliminate the option before signature. Under these assumptions, reserve the 60-minute session first and use the remaining 30 on the task best supporting the immediate decision. If valid evidence already established recovery, repeating the session might cease to be a priority. Investigation value depends on what remains unknown and which decisions it can change. Do not assign automatic priority to the largest document, the most senior stakeholder or the task with most missing pages.
8. Exercise: build a plan that enables a decision
Produce a table with five columns: decision, evidence, dependencies, owner and deadline. The 24-user pilot retains the recovery requirement; the deputy has the delegation described in this lesson. Inventory and exception analysis start at 10:00 with different people and take two and three hours. Synthesis takes one hour after both. The approval session starts at 15:00 and lasts 30 minutes; the reservation expires at 16:00. The packet contains restricted records and policy permits an approved summary view. Solution: synthesis finishes at 14:00, the decision can finish at 15:30, delegation covers the pilot and the view must retain version, scope and limitations. Schedule feasibility does not establish that the pilot will be approved: evidence may justify refusal. For the international meeting, distinguish “ready for decision”, “conditionally approved” and “authorized to execute”, explaining each state’s concrete condition. International variant: a session at 18:30 on a UTC+8 clock corresponds to 10:30 UTC; another at 12:00 UTC+1 corresponds to 11:00 UTC. With 30 minutes each and 30 minutes of synthesis after both, delivery can occur at 12:00 UTC without reserve. These offsets are fixed in the exercise; for a real calendar, confirm offsets on the meeting dates.
Exercise files
Original Python 3.10+ exercise without dependencies: compatibility, block costs, working calendar and international sessions. Includes editable data, instructions and worked results.
A packet delivered on time for a committee may arrive too late to cancel a reservation. The useful deadline depends on the commitment and review time.
Common pitfalls
Planning only documents; adding tasks that run in parallel; forgetting reviewer availability; extending approval scope; counting roles as independent people; comparing totals without denominators; attributing causation without evidence.
Related topics: Change strategy · Analysis governance
The analysis plan should get sufficient information to a person with authority while the decision can still have its intended effect.
References
- Business Analysis Planning and Monitoring task cards · Public task card v2.0, copyright2025
- Plan Business Analysis Approach · Public task card v2.0, copyright2025