Three days before my on-site at a mid-sized software company, I bombed a mock interview so badly that the interviewer asked, gently, whether I'd slept. I hadn't. I'd spent the previous week memorizing algorithm patterns and exactly zero minutes practicing how to say anything out loud. The rejection email came nine days later, which stung less than the realization that I'd prepared for the wrong test entirely.
That's the trap with technical interview prep. Most candidates treat it like an exam to cram for. It isn't. It's a performance under observation, and the observation is half the grade. Here's what actually moved the needle for me, and what I'd tell you to skip.
Key Takeaways
- Preparation beats talent by a wider margin than most candidates admit—but only if you rehearse out loud, not just in your head.
- Learn the five C's (competence, culture, communication, consistency, curiosity) and map every answer to at least one of them.
- The 30-60-90 rule turns a vague "tell me about yourself" into a structured pitch for your first three months on the job.
- Apply the 80/20 rule ruthlessly: a handful of core topics will carry most of your technical rounds.
- Expect 3 to 5 rounds, live coding, and at least one round where they watch how you think, not just what you know.
What a technical interview actually tests
Here's the thing nobody tells you upfront: the interviewer already believes you can code. Your résumé cleared that bar. What they're measuring now is how you behave when a problem doesn't cooperate—because that's what the job is, every single day.
I've sat on both sides of the table now, and the pattern holds. Strong candidates narrate their thinking. Weak candidates go silent, panic, and produce nothing. Same skill level, wildly different outcomes.
What are the 5 C's of interviewing?
The five C's are competence, culture, communication, consistency, and curiosity. Competence is whether you can solve the problem. Culture is whether the team wants you around at 4 p.m. on a bad day. Communication is how clearly you explain a decision you've already made. Consistency means your answers don't contradict each other across rounds. Curiosity is the one candidates forget—asking a real question about their architecture or their on-call rotation signals that you're interested in the work, not just the offer.
I started scoring my own mock answers against all five. My communication score was embarrassing for the first two weeks. Competence was fine. That gap is exactly why good engineers get rejected.
The 80/20 rule, applied to your prep calendar
What is the 80/20 rule in interviewing? It's the principle that roughly 80% of your results come from 20% of your efforts—so you identify the small set of topics and behaviors that dominate your specific interview loop, and you drill those until they're automatic, instead of spreading yourself thin across everything.
When I was prepping for a backend role, I mapped the loop: two coding rounds, one system design, one behavioral. Coding was 50% of the weight, so it got 50% of my time—not 90%. I'd been over-indexing on algorithms because they felt measurable. Design and behavioral rounds are harder to self-assess, so I'd been avoiding them. Classic mistake.
The 20% that mattered for me:
- Two patterns in system design (caching strategy, and how to shard a database when traffic spikes)
- Three behavioral stories I could reshape for almost any prompt
- One habit: talking through my reasoning before writing a single line
Everything else—the obscure graph algorithm, the framework trivia—was noise. I cut it and my pass rate on mocks went from roughly a third to well over half in about six weeks.
What is the 30-60-90 rule in an interview?
The 30-60-90 rule is a framework for describing what you'd accomplish in your first 30, 60, and 90 days in the role. Interviewers ask about it because it separates candidates who've thought about the job from those who just want a job. Your first month is learning—the codebase, the team, the unwritten rules. Days 30 to 60 are about contributing small, safe improvements. By day 90, you should be owning something real.
I keep a rough version of this in my head before every final round. When they asked "where do you see yourself in six months," I had a concrete answer ready. The interviewer nodded and moved on—which, honestly, is the best reaction you can hope for.
What should you expect in a technical interview?
Expect three to five rounds, and expect them to test different things. The typical structure looks something like this:
| Round | Format | What they're really checking |
|---|---|---|
| Screening | 30–45 min call, one problem | Can you write working code at all |
| Live coding | 45–60 min, shared editor | How you think when someone's watching |
| System design | 45–60 min, whiteboard or doc | Can you reason about scale and tradeoffs |
| Take-home | 2–6 hours, your own time | Code quality, testing, documentation |
| Behavioral | 45 min, conversation | Whether they want to work with you |
Not every company runs all five. Smaller teams often skip the take-home; larger ones sometimes add a pair-programming session where you work alongside an engineer on a real ticket. The one constant: at least one round where you can't hide behind a keyboard. Prepare for that one hardest.
How to survive live coding without freezing
The silence is what kills you. I've watched brilliant people go quiet for four minutes and lose the room. The fix is almost insultingly simple: say what you're about to do, then do it.
- Restate the problem in your own words and confirm you understood it
- Ask one or two clarifying questions (input size, edge cases, constraints)
- Sketch the approach in plain language before typing
- Write the code, narrating the tricky parts
- Test it out loud with a small example
That's it. No secrets. I practiced this loop on ten different problems before my last loop, and the difference was night and day—my interviewer later told me my communication was the deciding factor.
Building a prep plan that actually holds up
Six weeks is a reasonable runway if you're working full time. Shorten it to three if you're between jobs and can put in real hours; stretch it to ten if you're switching fields entirely. I've done both a rushed two-week sprint (bad idea—I was exhausted by round two) and a comfortable eight-week build, and the longer timeline produced a noticeably calmer version of me.
A rough weekly shape I've used:
- Five hours of problem practice, spread across the week, not crammed into one sitting
- Two mock interviews, recorded if possible—the playback is brutal but useful
- One hour reviewing your own past projects, so you can talk about tradeoffs you actually made
- Thirty minutes writing down questions to ask them
Real talk: the recording part hurt. I heard myself say "um" eleven times in four minutes. But you can't fix what you won't watch.
Which tools earned their keep
A cheat sheet you build yourself beats any downloadable PDF, because the act of writing it is the studying. Mine was two pages: pattern names, when to reach for each, and one snippet per pattern. I rewrote it from memory every Sunday. By week four I barely needed the paper.
For practice platforms, anything that gives you a shared editor and a timer works. The specific site matters far less than whether you're doing it under mild pressure with someone watching.
The part everyone underestimates
Two days before my successful interview, I stopped studying entirely. I slept nine hours, went for a walk, and read nothing technical. My brain needed the rest more than the last five algorithm problems.
That's the piece most guides skip. You can over-prepare into a state where you're too wound up to think clearly, and I've been there—staring at a whiteboard, knowing the answer, unable to reach it because my hands were shaking.
So here's what I'd leave you with. Preparation isn't about knowing everything. It's about building enough familiarity that the unfamiliar stops scaring you. The candidates who get offers aren't always the smartest ones in the room. They're the ones who walked in calm, said what they were thinking, and treated the whole thing like a conversation between two people trying to figure out if they should work together.
Which, when you strip away the whiteboards and the timers and the rejection emails, is all it ever was.