Why do interviewers penalize candidate questions that a two-minute search of public pages already answers?
answer
- Think about what the closing window is for
- Ask who else could have answered it
- A public page already says this
- Cut it, then convert it
- Cite the fact, ask the tradeoff
basics
~20 sA question the careers page already answers spends your closing minutes proving you did not prepare. Panels read it as low interest and weak filtering, and it burns your one chance to ask something only an insider knows.
solid answer
~40 sThe window at the end of a round is short and it is the only moment you get an insider's private knowledge. A question whose answer sits on the job description, the product pages or the company's own engineering blog wastes that, and it tells the panel two things at once: you did not read, and you could not tell a strong question from a weak one. The fix is a discard pile. I prepare six questions and deliberately cut the ones a two-minute search answers, which usually leaves two. The cut ones are not wasted: each one converts by citing the public fact in half a sentence and then asking for what nobody publishes — the tradeoff behind it, or what living with it costs the team week to week.
go deeper
Before every round, run each of your questions through one filter: could a public page answer this? If yes, cut it. Walking in with two questions nobody outside the company could answer is enough.
Be ready to explain why panels weight this so heavily: it reads as both preparation and filtering, and it burns a short window that only an insider can fill. Show the conversion move, not just the cut.
Demonstrate that you can convert on the spot — take a published fact, name it in half a sentence, and ask for the tradeoff behind it. Be willing to cut your own favorite question when it turns out to be published.
Own the tradeoff between a question that displays your reading of the public material and one that gathers information you actually need to decide. Both are legitimate; be able to say which one you spent the window on and why.
## What "search-answerable" actually means A candidate question is search-answerable when the answer sits on a page anyone can open in roughly two minutes: the job description you were sent, the careers page, the product or pricing pages, the company's own engineering blog, a public docs site or status page. On an infrastructure team round, questions like *"What does the platform team build?"*, *"Which regions do you run in?"* and *"How big is the engineering organization?"* are all in that class. The test is not "is this fact obscure to me" but "is this fact published". A candidate who has never worked on infrastructure may genuinely not know what an internal deployment platform is — but if the company describes theirs publicly, asking what it is is still a preparation failure, not a curiosity signal. ## Why the panel weights it so heavily Three things happen at once when a search-answerable question lands. 1. **Preparation signal.** The panel infers how much of your evening this application got. Interest that does not survive two minutes of reading does not read as interest. 2. **Filtering signal.** Interviewers are also judging whether you can separate a good question from a weak one — the same judgment they will rely on when you triage an incident or review a design. A candidate who asks the weakest question on their own list has shown their filter. 3. **Opportunity cost.** A peer engineer on an infrastructure team knows things nobody publishes: what the last migration actually cost the on-call rotation while it was in flight, which parts of the platform people quietly route around, how a change to the deployment path gets agreed. You get one short window with that person. Spending it on a published fact is the expensive part. ## The discard pile Treat the cut list as a real working artefact rather than a mental note. A workable ratio: prepare six questions, discard four because a two-minute search answers them, and walk in with the two that survive plus one spare in case a survivor gets answered during the round. A prepared list for an infrastructure round, with the cuts marked: | Prepared question | Verdict | |---|---| | What does the platform team build? | discard — the careers page says | | Which regions does the platform run in? | discard — published | | Is there one shared deployment path for all services? | discard — the engineering blog describes it | | How many engineers are on this team? | discard — the posting says | | When a service owner needs a change to the deployment path, does your team do that work or do they? | survives | | What did the last big migration cost the on-call rotation while it was in flight? | survives | Keeping the pile matters because the cut question usually contains a strong question inside it. ## Converting instead of deleting The conversion is mechanical: cite the public fact in half a sentence, then ask for the part that cannot be published — the tradeoff behind it, the cost of living with it, or what it changed for the people who maintain it. - Discarded: *"Do you have one shared deployment path?"* - Converted: *"The public write-up describes one shared deployment path for every service. What does a team do when their service does not fit that path?"* The converted version proves you read the material and asks for something only someone inside can answer. That is the whole trick, and it works on almost every cut question. ## The line that ends a round badly The extreme version is the one interviewers repeat to each other afterwards: *"So... what does your company actually do?"* — the question that ends a round badly. It is rarely asked out of contempt; it is usually a candidate who applied broadly, ran out of prepared material and reached for something safe. It is not recoverable inside the round, which is why the prevention work happens the night before. ## When a fact question is legitimate Three honest exceptions: - **The fact is not actually published.** Internal team boundaries, who owns which service, how work reaches the team — these are usually invisible from outside, and asking is correct. - **You cite the source and ask for the update.** "The posting describes this role as split between platform work and support rotations — is that still the shape of it?" That is verification, not ignorance. - **The fact is material to your decision and this interviewer owns it.** Confirming something you will decide on is worth a plain question, as long as you address it to someone who can answer. ## If you realize mid-question Do not abandon the sentence — a stalled question is worse than a weak one. Land it, then add the insider half immediately: "...and more usefully, what does that look like on a bad week?" The recovery costs you a few seconds and turns a flat question into a real one.
- Is it ever fine to ask about something the job description already states?Yes, when you cite it and ask for what it cannot say. "The posting describes this role as half platform work and half support rotation — is that still the split in practice?" is verification, and it reads as preparation. What fails is asking for the published fact itself as though you had never opened the page.
- How many of your prepared questions should survive your own discard pile?Roughly two out of six. Prepare six, cut the four a two-minute search answers, and carry the two survivors plus one spare in case a survivor gets answered mid-round. Two strong insider questions comfortably fill a closing window; six weak ones do not, and the panel remembers the weakest one.
- How would you convert a discarded question about our public deployment tooling into a stronger one?I would name the published fact in half a sentence and then ask for the unpublished half: "The write-up describes one shared deployment path for every service — what happens when a service does not fit it?" That asks about tradeoffs and exceptions, which is exactly the knowledge that only lives inside the team.
saying these in an interview costs you the question
- Asking what the company's main product is at the end of a round
- Believing any question is better than no question at all
- Requesting facts the job description states, purely to seem engaged
- Assuming a panel cannot tell prepared curiosity from unprepared filler
- Skipping preparation on the theory that questions can be improvised