A controller automates three-way match in accounts payable using RPA, migrates the general ledger to a SaaS enterprise resource planning (ERP), and pilots a blockchain freight-payment system in the same fiscal year. Each project demands a different governance model, and the exam rewards the candidate who can name the technology, its benefit, and its trade-off in one breath.
SDLC is the disciplined sequence finance follows when building or buying any major system. The five phases run in order, and each ends with a deliverable that gates the next phase.
Systems analysis. Define the problem, document current-state workflows, and gather user requirements. The output is a requirements specification.
Conceptual design. Translate requirements into a logical model: data flows, entity relationships, screen logic, and reporting needs. This is technology-agnostic.
Physical design. Choose hardware, software, database platform, and network architecture. Specify file layouts, program modules, and security controls. Build or configure to spec.
Implementation and conversion. Test (unit, integration, user acceptance), train users, migrate data, and cut over.
Common mistakes
- Confusing RPA with AI. RPA executes pre-written rules and breaks when inputs change. AI learns patterns and adapts. The exam will describe a judgment-based task and ask which technology fits, looking for "AI", not "RPA".
- Naming the wrong SDLC phase as most expensive. Implementation feels expensive, but operations and maintenance is the longest and largest cost bucket over the system's life.
- Treating SaaS as cheaper in all scenarios. SaaS is cheaper upfront, but at roughly 5 to 7 years of subscription fees, total cost often crosses the on-premises equivalent. The trade is speed and flexibility, not always cost.
Bottom line
- SDLC has five phases: systems analysis, conceptual design, physical design, implementation and conversion, operations and maintenance, with operations and maintenance the longest and largest lifetime cost
- Business process analysis must precede automation: simplify the workflow first, because automating a broken process only delivers faster failure
- RPA automates rules-based, high-volume, structured-data tasks; benefits are speed, accuracy, lower cost, full audit trail, and 24/7 throughput
- AI extends RPA into pattern recognition and judgment, while OCR converts unstructured documents into bot-readable data
Exam shortcut
When a question describes a finance task and asks which technology applies, key on input structure and rule stability. Structured, stable, rule-based equals RPA. Unstructured or judgment-based equals AI. Document conversion equals OCR. Multi-party shared record equals blockchain. When a question compares deployment options, default to SaaS if speed and standardization are emphasized and to on-premises if customization, data residency, or competitive-differentiator language appears.
The full lesson (about 2,045 words, 14 min read) adds 2 worked examples, all 6 common mistakes, a self-check, free in the app.
Learning objectives
- 1F3
Browse all free CMA Part 1 lessons or jump into free CMA Part 1 practice questions.