Driving the round
Driving the round means setting the agenda in a design interview: propose a plan, budget the clock, narrate while you draw, pick the hard part, and close with what you'd change next. It shows the judgment the round tests.
What it is
Driving the round is taking charge of how a design interview spends its time. You propose the order of work, say how long each part gets, narrate the board as it changes, choose the part worth going deep on, and close by saying what you would change next. The interviewer still asks questions and can redirect you, but between their questions you are the one moving the conversation forward.
Meta's guide to its software engineering interviews puts "drive the discussion" on its list of tips for the design round, and says the interviewer is trying to learn whether you have built large systems before, which shows in how you anticipate and route around the usual problems. It also asks you to make it a conversation and ask clarifying questions, while still pushing toward a good solution. It describes the round as 45 minutes of talking and drawing on a whiteboard, with the interviewer typically keeping the last five minutes for your questions.
| Habit | What you do | Why it helps the interviewer |
|---|---|---|
Plan aloud | Before drawing anything, say the order you'll work in and roughly how long each part gets. | They can correct the plan in the first minute instead of the last ten. |
Budget the clock | Name a time for each part and move on when it's spent, even if the part isn't perfect. | There isn't time for each detail of a design in one round; a budget is how you act on that. |
Narrate the board | Say what each box is for and why the arrow goes where it does, as you draw it. | They hear the reasoning, not just the result, and can step in early. |
Choose the focus | Name the hardest part of the problem and spend most of the time there. | Picking the hard problem is itself a signal of experience. |
Check in and close | Pause at each transition to offer a redirect, and end with what you'd harden next. | They get the depth they want, and a view of how you'd keep building. |
Why narration matters. A design round is spent talking and drawing, and the two should happen together. When you draw in silence and explain afterward, the interviewer gets a finished picture with no view of the choices behind it, and redirecting you means interrupting a drawing rather than joining a conversation. When you say "this is the write path, and I'm putting a queue here because the fan-out is the expensive part" as the box appears, they hear the decision at the moment you make it, and they can ask about the alternative while it is still cheap to change.
Why the close matters. A common piece of preparation advice is to revisit systems you have built and ask what you would change if you did them again. Our advice: end the round the same way, about the design you just drew. Name the weakest part, what you would harden first, and the condition that would make you do it (a traffic figure, a new requirement, a failure you haven't handled). It shows you know where the design is thin, which is more convincing than claiming it is finished.
Driving pairs with the other habits of the round: naming the bottleneck first is how you choose the deep dive, and walking one request is the easiest way to narrate a board you have just drawn.
When to use it
Use it when the prompt is broad. Meta's own example is "how would you architect the backend for a messaging system?", which it says sounds vague at the start, and it suggests opening with requirements: how many people, how many messages, what delivery latency, how messages are stored, and whether clients are pushed to or poll. Those first questions are already the first act of driving, because you are deciding what the round is about.
Use it again at each transition. The three we'd watch are the end of requirements (keep going or start drawing?), the end of the high-level sketch (which part to open up?), and the last ten minutes (another component or a close?). At each one, say what you plan to do next and offer the interviewer the wheel for a second: "I'd go into the feed fan-out now because that's where the scale bites. Unless you'd rather I cover storage first?"
| Part of a 45-minute round | Our suggested share | What ends it |
|---|---|---|
Requirements and scale | About 5 minutes | You can state the two or three numbers the design depends on. |
Plan and high-level sketch | About 10 minutes | Each main component is on the board with one sentence of purpose. |
Deep dive on the hard part | About 20 minutes | You've compared at least two options and chosen one. |
Close: what you'd change next | About 3 to 5 minutes | You've named the weakest part and the trigger for fixing it. |
Their questions | The last 5 minutes | The interviewer usually keeps these for the candidate. |
The split above is our suggestion; the 45-minute length and the five minutes at the end are the figures cited above. Adjust it to the round you are in, and say your budget aloud so the interviewer can tell you if their round is shaped differently.
When not to
Driving is not the same as holding the floor. There are moments when the right move is to hand the wheel over.
- The interviewer redirects you. The round is meant to be a conversation, and the interviewer's questions are their half of it. When they point at a component, go there, even if your plan had it later. Steering back to your agenda spends their time on your preference.
- They ask a narrow question. "Why a queue and not a direct call?" wants that answer in a sentence or two, with the trade-off, and then a pause. Our advice: answer before resuming the tour of the board.
- You don't understand the problem yet. A plan made before you know what you are building is a guess, and you may have to throw it away. Ask the clarifying questions first; the plan comes after the requirements.
- The interviewer sets the order. If they open each part themselves, driving means answering well and signaling where you would go next, not proposing a different order.
- The hint lands twice. If the interviewer has nudged you toward the same part twice, we'd treat that as the deep dive they want. Take it instead of defending your own choice of focus.
Compared
| Moment | Letting the interviewer lead | Driving | When letting them lead is right |
|---|---|---|---|
Opening | Start drawing the familiar boxes and see where it goes. | State the plan and the time budget, then ask if it fits their round. | The interviewer opens with a plan of their own. Follow it and say where you'd like extra time. |
The board | Draw the design, then explain it when it's complete. | Say what each piece does as you add it, and the choice behind it. | A quick private sketch of two options for yourself, before you pick one to narrate. Keep it to seconds, not minutes. |
Choosing the deep dive | Wait for "what would you like to go into?" | Name the hardest part and why, then go there. | They name the part. Go there: it is the part they chose to spend their time on. |
The clock | Finish each part when it feels finished. | Move on when the budget is spent, even with loose ends, and say what you're skipping. | You are three sentences from resolving a point they asked about. Finish it, then move. |
The close | Recap each component. | Name the weakest part and what you would change with more time. | There is under a minute left and they have asked a direct question. Answer that. |