SAA-C03 study plan: from hands-on to exam-ready
Updated
A dependable SAA-C03 study plan has four phases: build enough hands-on familiarity to reason about services, study the four design domains in weight order with daily scenario practice, consolidate the trade-offs you keep getting wrong, then rehearse with full-length timed mocks until you clear the pass mark consistently. Because SAA-C03 tests architectural judgment rather than recall, the plan front-loads doing over reading — and the timeline stretches or compresses depending on how much you already build on AWS.
Before you start: set the finish line and gauge your gap
Pick an exam window now, even loosely — a date turns study into a project. Then answer a short practice session cold, before studying anything. On an associate exam the baseline is especially useful: it tells you whether your gap is knowledge (you do not know the services) or judgment (you know them but pick the wrong one under a constraint), and those need different remedies.
Be honest about your hands-on level, because it sets the plan length more than anything else. Daily AWS builders can move fast; candidates whose experience is mostly conceptual should budget real time in Phase 1 rather than trying to skip it.
Phase 1 — Build hands-on intuition (do not skip)
Before deep study, make the core services concrete. On the AWS Free Tier, stand up the building blocks the exam obsesses over: an EC2 instance behind a load balancer in an Auto Scaling group, an S3 bucket with a couple of storage classes and a lifecycle rule, a small RDS or DynamoDB table, a VPC with public and private subnets and a NAT gateway, an SQS queue in front of a worker.
You are not learning to operate these at production quality — you are earning the intuition that makes scenario questions answerable. Every hour here pays back later, because "which option" questions about services you have actually touched stop being guesswork.
Phase 2 — Study the four domains in weight order (the bulk)
Work through the domains heaviest-weighting first — secure, then resilient, then high-performing, then cost-optimized. For each domain: read its lessons end to end, then drill scenario sessions scoped to that domain until your accuracy stabilises before moving on. Finishing a domain completely beats sampling everything shallowly, because half-known trade-offs are exactly what the distractor options are built from.
Keep a daily practice habit running across everything you have covered, mixing earlier domains back in so they stay warm. Read the explanation on every option, right or wrong — on a judgment exam the reasoning behind the second-best answer is the actual syllabus.
Phase 3 — Consolidate the trade-offs you miss
When all four domains are covered, switch from learning to repair. Rebuild practice sessions from your incorrect and flagged questions — the review pool is your personal weak-spot list — and pay attention to patterns: if you keep confusing EBS with EFS, or Multi-AZ with read replicas, that is a specific gap to close, not a vague weakness.
This is also cheat-sheet time: skim every key takeaway domain by domain, and treat anything unfamiliar as a signal to revisit that lesson. By the end of this phase the recurring decision pairs — compute, storage, database, decoupling, networking, cost — should feel automatic.
Phase 4 — Mock week: rehearse, then book
Sit a full-length timed mock under real conditions: one sitting, no notes, no pausing. The scenario stems are long, so this phase is as much about pacing and stamina as knowledge — practise flag-and-return so a single hard question never costs you three easy ones. The mocks use the exam's scaled scoring, so the pass line means what it will on the day.
After each mock, spend a session on its incorrect answers before sitting the next. Book the real exam once you clear the pass mark comfortably on consecutive mocks, and keep the daily practice habit until exam day.
Signals you are ready
Book the exam when all of these are true:
- Consecutive full-length mocks above the pass mark — a trend, not one lucky sitting.
- No design domain clearly trailing the others in your per-domain accuracy, especially not security.
- The recurring trade-off decisions feel automatic, not effortful.
- Your review pool of incorrect questions is shrinking week over week, not growing.