Editorial policy

We are a small team building interview practice tools. This page sets out how our content is produced, where AI is involved, what we will not publish under any circumstances, and how to get something fixed. Last reviewed 2026-08-14.

How AI is used, stated plainly

Most of the 86,616-question bank was generated in bulk by a language model and then imported. The Machine Learning and LLM banks are the exception: those were written by hand, one question at a time, because subjects where a wrong answer sounds reasonable are exactly where bulk generation fails. Explanations, hints and answer feedback in the product are model-generated at request time.

We are telling you this rather than describing the bank as expert-authored, which would read better and would not be true. What we do commit to is that every question ships with an explanation of why the correct answer is correct — 100% coverage — and that errors get fixed at the source when you report them.

What we will not publish

These are not aspirations. Where a rule is enforced by a test, the test is named, and it fails the build rather than a reviewer noticing.

• ✕Invented reviews, ratings or testimonialstests/seo.spec.ts asserts no page emits aggregateRating or Review structured data.
• ✕Fabricated authors, credentials or years of experienceWe publish no author bylines, because we will not invent people to sign them.
• ✕Company attribution we cannot supportWe do not display any employer attribution on questions, and no page claims a named company asked one. An earlier version of this site did show such a tag; the underlying data did not support it, so the tag was removed rather than dressed up.
• ✕Statistics without a measurement behind themEvery figure on /methodology and /research is generated from the database by a script in tools/, never typed by hand.
• ✕Mass-generated thin pagesTopic pages require at least 15 distinct questions after deduplication; the rest are noindex.
• ✕Bought links, link exchanges or private blog networksNo paid placements. Nothing on this site is a paid link.

Where our numbers come from

Any statistic we publish about the question bank is measured against the live database by a generator script and written into a data file the pages import. Nobody types a number into a page. The practical effect is that our published figures cannot quietly drift from reality, and when they change we have to re-run the generator to update them — including when the new number is worse.

The clearest example is on our methodology page: it reports both the row count and how many of those rows are genuinely different questions, and it carries a correction where an earlier version of that page got the measurement wrong in our own disfavour.

Scores are practice feedback, not decisions

Interview scores and hire-band recommendations are rehearsal feedback. They are not a hiring decision, they are not shared with any employer, and no employer receives your session. Where the interview did not gather enough evidence to support a recommendation, the report withholds one instead of producing a confident guess.

Reporting an error

Send it through feedback. Include the question text or the page URL. Wrong answers and misleading explanations are treated as the highest priority, because a confidently wrong explanation is worse than no explanation at all — someone will walk into an interview carrying it. Fixes are applied to the question bank, so they reach every page the question appears on rather than the one you happened to be looking at.

Related: methodology · privacy · about