Books / Learning comic
Program Management
Theme: Business & Management
CartNest's eight-week sale deadline launches a twelve-chapter story about delivering useful checkout improvements responsibly. Maya and her colleagues define shopper outcomes, validate baselines, choose safe end-to-end slices, and connect requirements, code, tests, and release evidence. Planning exposes limited specialist capacity and a payment-service dependency, prompting explicit tradeoffs instead of unsupported promises. Sprint execution brings a real integration defect, a corrected status report, and an urgent button request that is deferred rather than displacing essential verification. The team eventually completes the candidate, controls customer exposure through staged rollout, and measures mature outcomes. Completion improves but misses its target, while payment reliability and speed succeed. The program closes with honest results, continuing operations, cross-training, and bounded discovery, with AI consistently treated as human-reviewed assistance.
Start reading →View in the interactive libraryAbout this book
CartNest wants faster, more reliable checkout before a sale eight weeks away. This twelve-chapter fictional story follows Maya, Ravi, Priya, Elena, Jules, Sam, sponsor Ada, and their delivery partners from that urgent request through measured outcomes. The opening distinguishes programs, projects, products, and teams before replacing feature-count success with a specific shopper journey: domestic mobile-web guest card payments. An agreed charter defines eligibility, reproducible baselines, completion, technical failures, speed, and support demand, alongside non-negotiable charge-safety, security, and accessibility guardrails. The team turns the outcome into small end-to-end stories: retry after confirmed technical failure, resolve unknown payment status, and show clear confirmation. Story mapping preserves safeguards while excluding wallets, loyalty redesign, international expansion, and unnecessary visual polish. Connected lifecycle records link goals, requirements, acceptance examples, designs, changes, tests, builds, deployments, and monitoring. Drafts remain distinct from evidence. Estimates account for engineering, verification, operations, capacity, and scarce skills; story points, person-days, elapsed duration, forecasts, and commitments are deliberately kept separate. A late Payments sandbox breaks the initial timing assumption, so the team withdraws its forecast rather than disguising the gap. Provider and consumer owners agree versioned behavior for pending, success, safe failure, repeat requests, callbacks, and reconciliation. Unknown status never authorizes a blind retry. Ada approves reduced confirmation polish without inventing critical-path savings or extra capacity. Developers select a feasible sprint goal and a smaller pending-to-success slice, retaining a complete Definition of Done. During execution, stale pending data overwrites confirmed success despite a green coding task, prompting a named defect, staffed repair, and evidence-backed rerun. Status reporting and change control test the same discipline. An AI draft based on stale records incorrectly declares completion and a guaranteed launch; Maya replaces it with current observations, conditional forecasts, risks, assumptions, and a concrete staffing request. Ada's urgent copy-reference button appears to fit aggregate capacity but exceeds Sam's specialist verification allocation. The team uses the existing selectable reference and readback process instead, recording deferral without authorizing implementation. Later planning uses actual effort and fresh capacity. Completed story checks, integration, accessibility, security, support, monitoring, and rollback evidence establish readiness, while a demonstration or merged change alone does not. Production release proceeds in bounded stages using the tested artifact and controlled configuration. Low traffic and immature support data cause holds at five and twenty-five percent rather than automatic expansion. Full exposure follows additional evidence, then a fixed fourteen-day non-sale window and outcome-maturity periods. Completion rises from 60% to 64%, missing the 66% target; technical failures reach 1.8%, confirmation p95 reaches 2.8 seconds, and support remains within its guardrail. The team reports observed improvement without claiming causal proof or invented financial return. Ada accepts closure with residual risks and the benefit gap assigned. Operational owners retain monitoring and pause authority, specialist cross-training receives reserved capacity, and a two-person-day discovery study explores the next investment. Across these stages, approved AI tools use sanitized inputs, while people verify sources, arithmetic, decisions, and final claims. The story ends with measured progress and accountable continuation rather than declaring deployment the final success.
Published chapters
- 01 - We need this before the sale! - Program Management Comic
- Program Management - Chapter 02 - What Does Success Actually Mean
- Program_Management_03_We_Cant_Build_Everything
- Program_Management_04_Where_Does_The_Work_Live
- Program_Management_05_Can_You_Give_Me_A_Date
- Program_Management_06_Our_Team_Is_Ready_Their_Team_Isnt
- Program_Management_07_Which_Plan_Can_We_Responsibly_Promise
- Program_Management_08_What_Can_We_Achieve_This_Sprint
- Program_Management_09_The_Board_Is_Green_Why_Are_We_Stuck
- Program_Management_10_One_Urgent_Change
- Program_Management_11_Done_Demonstrated_or_Released
- Program_Management_12_Did_the_Program_Actually_Help