Most SAA-C03 study plans you will find online are just the four exam domains split across however many weeks, with reading assignments attached to each one. That approach produces candidates who can define what a service does and still struggle on exam day, because SAA-C03 rarely asks "what does this service do" — it asks "which of these four plausible options actually satisfies this specific constraint," and that judgment only comes from having weighed real tradeoffs yourself, at least once, in an actual AWS account.
So this plan is built around building things, in roughly the order the exam domains build on each other — IAM and networking first, because nearly everything else depends on understanding them, then compute and storage, then the higher-level patterns (high availability, decoupling, cost) that combine all of it. Ten weeks assumes 6 to 8 hours a week; compress or extend based on your actual pace, but do not skip the hands-on weeks to save time — that is the part of this plan doing the real work.
Week-by-week schedule
IAM, and getting your hands actually dirty
Skip the temptation to start with EC2 — start with IAM, because every other week depends on understanding users, groups, roles, and policies cold. Create a free-tier account if you have not already, set up a non-root IAM user with MFA, and write two or three custom IAM policies by hand rather than using the visual editor. Read AWS's own IAM documentation on the difference between identity-based and resource-based policies; SAA-C03 tests this distinction directly.
EC2, EBS, and instance purchasing options
Launch an EC2 instance manually (not via a template) at least twice, once with a public subnet and once attached to a security group you configure yourself. Study the EBS volume types (gp3, io2, st1, sc1) and when each is appropriate — this shows up constantly in "which storage solution is most cost-effective" questions. Cover On-Demand, Reserved, Spot, and Savings Plans pricing models; a surprising number of Domain 4 questions are just this table in disguise.
VPC networking, the domain most candidates underrate
Build a VPC from scratch: public and private subnets, an internet gateway, a NAT gateway, and route tables that actually route correctly between them. This is tedious and worth doing anyway, because VPC questions on the real exam assume you understand why a private subnet needs a NAT gateway to reach the internet, not just that it does. Cover VPC peering, Transit Gateway, and the difference between a security group and a network ACL — stateful versus stateless is a recurring trap.
S3 in depth: storage classes, lifecycle, and access control
S3 shows up in more questions than any other single service. Cover every storage class (Standard, Intelligent-Tiering, Standard-IA, One Zone-IA, Glacier Instant Retrieval, Glacier Flexible Retrieval, Glacier Deep Archive) and be able to justify a choice based on access frequency and retrieval time, not just memorize the names. Set up a lifecycle policy that transitions objects between classes automatically. Cover bucket policies versus IAM policies versus ACLs — knowing which one to reach for in a given scenario is tested directly.
Databases: RDS, Aurora, DynamoDB
Launch an RDS instance with Multi-AZ enabled and, separately, create a read replica — then look at what actually differs between the two failover behaviors, since this exact distinction is a recurring exam trap (Multi-AZ failover is automatic; read replica promotion is manual). Cover DynamoDB's on-demand versus provisioned capacity modes and when each fits. If you have time, stand up an Aurora cluster and compare its replication story to standard RDS.
High availability, fault tolerance, and disaster recovery
This week is almost entirely Domain 2 (Resilient Architectures, 26% of the exam). Study the four DR strategies AWS names explicitly — backup and restore, pilot light, warm standby, multi-site active-active — and be able to rank them by cost and recovery time. Configure an Auto Scaling group with a scaling policy tied to a CloudWatch alarm; watch it actually scale. Cover Elastic Load Balancer types (ALB, NLB, GWLB) and when each applies.
Serverless, messaging, and decoupling patterns
Deploy a small Lambda function and trigger it from an S3 event and, separately, from an SQS queue. Cover SQS standard versus FIFO queues, SNS fan-out patterns, and where EventBridge fits versus SQS/SNS. These services appear more in SAA-C03 than candidates expect, since decoupled architecture is a recurring "design for resilience" pattern across multiple domains.
Cost optimization: the domain everyone under-studies
Domain 4 is only 20% of the exam, which is exactly why candidates skip it and lose easy points. Study Cost Explorer, Trusted Advisor, and the tradeoffs between Reserved Instances, Savings Plans, and Spot Instances in more depth than week 2 covered. Cover S3 storage-class cost tradeoffs again from a pure pricing angle, not just an access-pattern angle. This is a short, high-leverage week.
Full-length timed practice exams, review only
Stop learning new material. Take a full 65-question, 130-minute timed practice exam under real conditions — no notes, no pausing. Review every question you missed and, just as importantly, every question you got right for the wrong reason (a lucky guess between two plausible answers is a gap, even if the answer was correct). Repeat with a second full attempt later in the week.
Targeted review and booking the exam
By now your practice scores should be consistently above 85 percent across full attempts. Spend this week drilling whatever domain is still your weakest rather than reviewing everything evenly. If you are hitting 85%+ consistently, book the exam for the end of this week or the start of the next — waiting longer past this point tends to cost more in forgotten detail than it gains in extra review.
How to know you are actually ready
Not "I read every domain" — that is a poor readiness signal for this exam specifically. The real signal is scoring consistently above 85 percent across at least four to six separate full-length, timed, 65-question practice exams. Consistency matters more than a single good score, because which scenarios you happen to draw on any given practice run affects the result more on SAA-C03 than on a shorter, more recall-based exam like Cloud Practitioner.
CertSharp's 500-question SAA-C03 bank is built for exactly this kind of repeated full-length practice — large enough that you are not seeing the same 65 questions on every attempt, with an explanation on every question covering not just the right answer but why the plausible-looking wrong ones fail the specific constraint in the scenario.
Frequently asked questions
Can I pass SAA-C03 in less than 10 weeks?
Yes, if you already have real AWS experience — some candidates with a year or more of hands-on work compress this into 3 to 4 weeks of focused review rather than fresh learning. The 10-week timeline in this plan assumes you are starting with limited hands-on AWS time and need to build that experience, not just review it, which is the slower part for most candidates. If you already hold Cloud Practitioner and have touched EC2, S3, and VPC in a real project, you can likely cut this plan to 6 weeks by skipping the earliest foundational weeks.
Do I really need to do hands-on labs, or can I just read and use practice questions?
You can attempt it that way, and some candidates do pass, but it is measurably harder and the failure mode is specific: reading-only candidates tend to know what a service does in isolation but struggle when a question asks them to weigh two services against each other under a constraint, which is most of what SAA-C03 actually tests. An hour spent building something in the AWS free tier teaches tradeoffs that no amount of reading replicates, because you run into the actual friction points — IAM permission errors, a security group misconfiguration, a Multi-AZ failover you have to trigger and watch happen — that a study guide can only describe secondhand.
How many practice questions should I do before booking the real exam?
Enough to complete at least 4 to 6 full-length, 65-question timed practice exams with a score consistently above 85 percent, not just one good run. SAA-C03's scenario density means a single strong practice score can be partly luck of the draw — which specific scenarios you happened to get — so consistency across multiple full attempts is a more reliable readiness signal than any single score.
What if I fail my first attempt after following a plan like this?
Use AWS's post-exam section-level score breakdown to identify which of the four domains you scored weakest on, then spend the mandatory 14-day retake wait drilling that domain specifically rather than restudying everything at the same intensity. Most second-attempt failures happen because candidates re-read the same material broadly instead of targeting the actual gap the score report identified.
Start Building Your Practice Score
30 free SAA-C03 questions to see where you actually stand — free account, no credit card. $11.99 for lifetime access to all 500, or $9.99/month Pro for every CertSharp certification.