AWS SAA Exam Prep: The Pragmatic SAA-C03 Study Plan

The AWS Certified Solutions Architect – Associate (SAA-C03) exam is less about memorizing AWS service facts and more about making the right architectural trade-off under time pressure. Most questions give you a scenario, a set of constraints, and four or five plausible answers. The correct answer is rarely the only technical option. It is the one that best balances security, resilience, performance, and cost.

What the exam actually tests

According to the AWS Certified Solutions Architect – Associate Exam Guide, SAA-C03 covers four domains:

Design Secure Architectures: about 30%

Design Resilient Architectures: about 26%

Design High-Performing Architectures: about 24%

Design Cost-Optimized Architectures: about 20%

You will see 65 questions with a 130-minute timer. Some questions have multiple correct responses. The exam uses a scaled score from 100–1000; the commonly reported passing score is 720. AWS does not publish the exact raw cut score.

Question stems tend to look like: 'A company is running a three-tier web application...' or 'A solutions architect needs to design a solution that minimizes operational overhead...' The keywords in those stems matter more than the service catalog in your head.

The high-yield services to master

You do not need deep expertise in every AWS service. You need strong decision trees for a core set of primitives.

Prioritize these:

Networking: VPC, subnets, route tables, NAT Gateways, VPC endpoints, Security Groups, NACLs, Transit Gateway, VPN, Direct Connect

Compute: EC2, Auto Scaling, Lambda, ECS/EKS at a high level, Elastic Beanstalk basics

Storage: S3 storage classes and lifecycle policies, EBS volume types and snapshots, EFS, S3 Glacier, Storage Gateway, AWS Backup

Databases: RDS, Multi-AZ, read replicas, DynamoDB, DynamoDB Accelerator (DAX), ElastiCache, Aurora

Messaging and decoupling: SQS, SNS, EventBridge, Kinesis Data Streams

Identity and security: IAM policies, roles, Organizations, SCPs, KMS, CloudTrail, CloudWatch, AWS Config, GuardDuty at a high level

Edge and DNS: Route 53 routing policies, CloudFront, Global Accelerator, S3 Transfer Acceleration

Cost and governance: Savings Plans, Reserved Instances, Spot Instances, Compute Optimizer, AWS Budgets, Cost Explorer, Trusted Advisor

The key skill is choosing between similar services. For example:

SQS vs. Kinesis: SQS for decoupling and buffering at lower throughput; Kinesis for ordered, replayable streaming at higher scale.

NAT Gateway vs. VPC Endpoint: NAT Gateway for outbound internet access from private subnets; VPC Endpoint for private AWS API access without internet egress.

Security Group vs. NACL: Security Groups are stateful and instance-level; NACLs are stateless and subnet-level.

RDS Multi-AZ vs. Read Replica: Multi-AZ is for high availability and failover; read replicas are for scaling reads and offloading reporting.

S3 Standard vs. Intelligent-Tiering vs. One Zone-IA: choose based on access patterns, durability requirements, and tolerance for zone loss.

A realistic 6–8 week study plan

If you already work with AWS, six weeks at about 60–90 minutes per day is realistic. If you are starting from scratch, add two weeks of hands-on fundamentals first.

Weeks 1–2: Core primitives

Learn the core services deeply. Build a simple VPC with public and private subnets. Launch EC2 instances behind an Application Load Balancer. Move static assets to S3 and front them with CloudFront. Write an IAM role with least-privilege policies. You should understand why the pieces exist, not just how to click through the console.

Weeks 3–4: Architecture patterns

Focus on high availability, disaster recovery, decoupling, caching, and serverless design. Work through patterns such as:

Web tier with ALB, Auto Scaling, and RDS Multi-AZ

Event-driven fan-out with SNS to SQS queues

Serverless APIs with API Gateway, Lambda, and DynamoDB

Hybrid connectivity with VPN and Direct Connect

Multi-Region routing with Route 53 failover or latency policies

When you study a pattern, write down the failure mode and the cost or operational impact of the fix.

Week 5: Security and cost hardening

Spend one week on IAM policies, resource policies, key management, Organizations, SCPs, CloudTrail, and monitoring. Then shift to cost. Know when Spot Instances make sense, how Savings Plans differ from Reserved Instances, and how to optimize S3 storage spend.

Weeks 6–7: Practice exams and gap review

Use scenario-based practice exams. Tutorials Dojo, AWS Skill Builder, and official AWS practice exams are useful. Do the first one untimed and open-book. Review every answer, including the ones you got right. Build a mistake log grouped by domain and service.

In the final week, take two or three timed exams. Aim for a consistent 80% or higher in timed conditions. Do not chase a perfect score. The goal is to expose weak decision trees before exam day.

How to use practice exams

Practice tests are diagnostic tools, not just scoreboards.

After each exam:

1. For every wrong answer, write a one-line rule: 'Use VPC endpoints for private S3 access from a VPC when data cannot traverse the public internet.'

2. If you missed a question because you did not know the service, add it to a flashcard or note.

3. If you missed a question because you misread the constraint, record the keyword you ignored: 'least cost,' 'most secure,' 'most operationally efficient,' 'single Region.'

4. Revisit the official AWS documentation or AWS Well-Architected Framework only for the gaps, not as a start-to-finish read.

Many test takers find real exam questions slightly less difficult than the hardest third-party practice questions, but more ambiguous. Treat every practice question as a tiny design review.

The architecture judgment loop

SAA-C03 rewards a repeatable assessment loop:

What is the requirement?

What are the constraints?

Which services can satisfy the requirement?

What is the failure mode?

What is the cost and operational overhead?

Which option best aligns with the AWS Well-Architected Framework?

If you can run that loop quickly, you can answer most questions without memorizing deep service limits.

What to deprioritize

Do not spend hours on:

Exact service quotas, except for commonly relevant limits like Lambda's 15-minute timeout, S3 object size and multipart upload thresholds, and EBS volume sizes.

Deep machine learning service internals. SAA-C03 may mention SageMaker, Rekognition, or Comprehend, but the expected depth is usually architectural integration, not model training.

Developer tools such as CodeDeploy, CodePipeline, and CodeCommit beyond their basic role in CI/CD.

Every EC2 instance family. Know the broad categories: general purpose, compute optimized, memory optimized, storage optimized, and burstable.

Exam-day tactics

Read the last sentence of the question first. The actual ask is often there.

For multiple-response questions, know exactly how many options you need to select.

Eliminate obviously bad answers before comparing the remaining two.

Use the flag-and-return feature. Do not let one hard question consume five minutes.

If two answers both seem correct, look for the constraint that differs: cost, operational overhead, security, or time to recover.

Final word

SAA-C03 is a judgment exam disguised as a multiple-choice test. The people who pass are not the ones who memorized the most service FAQs. They are the people who built a small set of strong mental models and learned to apply them under pressure. Study the core services, drill the trade-offs, and use practice exams to find and fix the gaps.

At Sapior, we take the same approach with developer tools: learn the primitives, understand the constraints, and then compose with confidence.