Score your mock round.
A 24-point rubric across the seven phases of a 45-minute system-design round. Score yourself 0–3 per phase; the deep dive is double-weighted because that's the phase the interviewer remembers. Your total maps to a calibrated hire / no-hire band. Scores save locally to your browser.
01 — Clarify
02 — Capacity math
03 — API + schema
04 — High-level architecture
05 — Deep dive
×2 weight06 — Scaling + failure modes
07 — Trade-offs + follow-ups
How to grade honestly
The rubric only works if you score what you actually said in the round, not what you meant to say. The most useful score is the one a friend who watched the round would give you. If you're scoring yourself, try this trick: re-read your scratch notes / whiteboard from the round, and grade based purely on what's written there. If the phase-01 questions weren't written down, they didn't happen as far as the grader is concerned.
A second trick: score immediately after the round, not the next morning. Reconstructed memory inflates scores; the same is true of the inflation in your own recall.
What the bands mean
The bands are calibrated against feedback from senior engineers we've coached through real loops. They aren't FAANG-published rubrics — those are proprietary. They are, however, consistent with what experienced interviewers grade against at most companies running the 7-phase shape.
| Score | Band | Reading |
|---|---|---|
| 0–8 | No hire | Score below 9/24 usually means you skipped phases or didn't reach phase 05. Focus next round on pacing — drop detail in earlier phases to make it to deep dive and scaling. |
| 9–13 | Lean no-hire | You hit the right phases but underweighted the deep dive. The grader's memory of you is "did the structure, lacked specificity". Practise picking and pushing on one component. |
| 14–18 | Mixed signal | Solid round. The hire decision will depend on your other rounds in the loop. Common gap at this band: failure modes underdeveloped or trade-offs vague. |
| 19–21 | Lean hire | Strong round. You owned trade-offs and went deep. Tiny gaps remain — usually in phase 07 self-awareness or in the proactivity of the deep-dive pick. |
| 22–24 | Strong hire | This round on its own pushes the loop toward yes. The remaining advice is don't blow another round — pace your energy. |
Using the rubric over multiple rounds
One round's score is a data point. Five rounds of scoring is a trend. The most useful pattern: track which phases consistently score 1 or below. Those are your study targets. Improving phase 05 from 1 to 2 lifts your total by 2 points (because of the weight) — usually enough to move a band.
Don't obsess over hitting 24/24. The ceiling is rarely worth the effort. 19+ is reliably "strong hire"; 15–18 is "good enough that other rounds decide". Optimise to consistency in the middle band rather than to a single peak.
Caveats
This rubric is for the system-design round specifically. Coding rounds have different rubrics (correctness, complexity, clarity, edge cases); behavioural rounds have wildly different ones (STAR completeness, impact specificity, ownership signal). Don't generalise the numbers.
The deep-dive double-weighting reflects our experience that this is the phase that decides most rounds. Some companies (Meta is one) weight all phases equally — if you know your loop does, treat the deep dive as a 1× weight and the max becomes 21.