A common way to fail an AWS exam goes like this. You watch 30 hours of course video, take notes, do a couple of practice tests and feel ready. Then an early question asks you to pick between a NAT gateway and a VPC endpoint for private S3 access, and you realize you watched that exact lesson three weeks ago and kept none of it.
The fix usually isn't a better course. It's how you handle everything after pressing play.
If you're searching "how to study for AWS certification" right now, you'll find plenty of lists of courses and practice exams. Those are fine. This post is about the part they skip, which is keeping the material in your head long enough to use it on exam day.
The problem with binge-watching a course
Most AWS courses are long. Associate-level ones commonly run 20 to 40 hours, and a lot of people get through them the way they'd get through a TV series: a few hours a night, at 1.5x, ticking off sections.
That feels like progress, and it's a bad way to remember anything. By the time you hit the CloudFront section, the IAM policy evaluation logic from week one is mostly gone. Memory researchers have known this since Hermann Ebbinghaus tested himself on nonsense syllables in the 1880s. Without review, what you learn drops off fast in the first few days.
AWS exams punish this harder than most, because the questions are scenario-based. The Solutions Architect Associate exam (SAA-C03) gives you 65 questions in 130 minutes, and many of them are "a company needs X with constraint Y, which option is most cost-effective?" You can't answer those by vaguely recognizing service names. You need to remember the tradeoffs.
Watch less per sitting, and stop after each section
I cut my sessions down to about an hour, and I stopped at the end of each section to close the video and write down what I'd just learned. From memory, without looking.
This was annoying at first. After a 25-minute lesson on S3 storage classes I'd sit there and only manage three lines. But the gaps showed me exactly what to rewatch, and the act of pulling it back out of my head is what made it stick. It's called active recall, and it does more for retention than rereading your notes ever will.
My notes looked like this:
S3 Intelligent-Tiering: moves objects between access tiers automatically based on usage. Good when access patterns are unknown. Small monthly monitoring fee per object.
Short, in my own words, and written as something I could be quizzed on.
Turn the tradeoffs into questions
The real exam material isn't "what is SQS." It's "when would you pick SQS over SNS over Kinesis." So I wrote my review questions around comparisons:
- When is a VPC gateway endpoint better than a NAT gateway for reaching S3?
- What's the difference between Multi-AZ and read replicas in RDS, and which one helps with read-heavy load?
- When does Aurora Serverless make more sense than a provisioned cluster?
I kept these in a plain list and answered a handful each morning before work. If I got one wrong, it went back to the top of the list.
Space out the reviews
Answering the same questions every day is a waste once you know them. What works better is coming back at growing intervals: the next day, a few days later, a week after that, then two weeks. Each time you successfully recall something, the gap before you need to see it again gets longer.
That's spaced repetition. You can do it with Anki, a spreadsheet, or a paper notebook. I wrote up how the scheduling math works in How to Implement Spaced Repetition (SM-2) in JavaScript if you want to see the guts of it.
For a six to eight week study plan, this is what keeps week one material alive in week seven.
Use practice exams as a diagnosis
I treated my first practice test as a test of the material. That was the wrong way round. A practice exam is most useful as a map of what you've forgotten.
After each one, go through every wrong answer and every lucky guess, and write a new review question for each. A few practice tests in, you'll have a stack of questions built from your own weak spots, and that stack is worth more than any section of the course.
Where the scattered stuff ends up
The course is rarely the only thing you study. I also watched re:Invent talks on YouTube, saved a few AWS Architecture Blog posts, and listened to a couple of podcast episodes about serverless patterns on walks. Those all went into different apps and I never looked at them again.
That's the problem I built Recl to fix. You share a video, podcast, reel or article to it from your phone or the Chrome extension. It writes a summary of the key ideas, with timestamps for videos, sorts the save into a topic, and brings it back 1, 3, 7, 14, 30 and 60 days after you saved it.
So the re:Invent talk on DynamoDB single-table design I saved in week two showed up again in week three and week four, instead of sitting in Watch Later forever. I also used the Ask feature a lot. Typing "what did I save about cross-region replication?" gave me an answer drawn from my own saves, with the sources, which beat scrubbing through a 50-minute video for the one slide I remembered.
It doesn't replace writing your own questions. That part you still have to do yourself.
A study week that worked for me
Weekdays: about an hour of course material, stopping after each section to write notes from memory, plus 10 to 15 minutes of review questions in the morning. Saturday: a practice exam or a set of practice questions, then turning every mistake into a new question. Sunday: off, mostly.
It's slower than binge-watching. But watching the course once, properly, beats watching it twice badly, and you walk into the exam recognizing the scenarios instead of guessing at them.
I build Recl, an AI learning app that saves your videos, reels, podcasts and articles, summarizes them, sorts them into topics, and uses spaced repetition so you remember what you learned.
Top comments (0)