When the interviewer pushes back
Treat tough follow-ups as teamwork: admit the weak spot, then fix it.
What they’re listening for
- • Do you stay calm when someone challenges you?
- • Can you find the weak spots in your own design?
- • Do you change your mind when there's a good reason?
When an interviewer asks “what if this fails?”, they're usually not saying your design is bad. They want to see how you think about failure. The best answers agree with the problem first, then explain the fix and what it costs.
Stiff vs. natural
Interviewer: “What happens if this database goes down?”
Sounds stiff
“It won't go down. We are using a reliable cloud database, so that will not be a problem.”
Sounds natural
“Yeah, right now that's a single point of failure. If it goes down, nobody can create links, and redirects only work for whatever's already in the cache. So the first thing I'd add is a replica in another zone with automatic failover. We might lose a few seconds of writes during a failover, which I think is fine for a link shortener. If this were payments, I'd want something stricter.”
- “Yeah, right now that's a single point of failure” admits the weak spot right away.
- It describes what users would actually see.
- It names the cost of the fix (a few seconds of writes).
- Comparing it to payments shows judgment.
Interviewer: “Why not use a NoSQL database here?”
Sounds stiff
“No, SQL is better. I have always used Postgres and it works very well.”
Sounds natural
“Honestly, that could work too. Our main query is “look up this code,” which is a simple key-value lookup, and a lot of NoSQL stores are great at that. I picked a relational database because most teams already know how to run one, and we're only talking a few terabytes. If we expected a lot more data, I'd be happy to switch. Were you thinking about scale?”
- “That could work too” shows you're open, not defensive.
- You give a real reason for your choice, and say when you'd change it.
- Asking a question back turns push-back into a conversation.
Thinking through a failure together
Interviewer
“Okay, say one link goes viral. Millions of clicks in a few minutes.”
You
“Ah, the fun scenario. This is the part where someone's pager goes off at 3am. So the good news is it's one link, which means it's one cache key. The cache should handle that fine, as long as the link is already in it.”
One light joke, then straight back to the problem.
Interviewer
“And if it isn't in the cache yet?”
You
“Then the first wave of requests all miss at the same time and hit the database together. Hmm, that's a real problem. One fix is to let only one request load it from the database while the others wait for that result. Some people call it request coalescing. I'd also let the CDN cache popular redirects, so most clicks never reach us at all.”
“Hmm, that's a real problem” is honest and calm.
Interviewer
“Nice, I like that.”
Handling push-back
- “Yeah, that's a weak spot right now.”Agreeing before you fix it.
- “Good question. Let's break it.”Welcoming the challenge.
- “What users would see is…”Describing the impact.
- “The trade-off there is…”The cost of the fix.
- “That could work too. The reason I picked … is…”Defending a choice, politely.
- “You've convinced me. Let's change it.”Changing your mind gracefully.
Humor and life talk
Works well
- • A quick joke about pagers or 3am alerts, then back to the problem
- • Admitting your first version was too simple (“Okay, version one just fell over”)
Skip it
- • Joking that the failure “won't happen”
- • Sarcasm when the interviewer challenges you
Example: “Okay, so version one of my design just fell over. Good thing it's only a whiteboard.”
Now say it out loud
Rehearse with the AI interviewer and get feedback on your answers.