The worst interview question isn't a bad question. It's a fine question the guest has answered in forty other interviews, delivered on autopilot, that produces a forty-first answer nobody wanted to hear again. Generic prep produces generic interviews — not because the host didn't try, but because "look them up before the call" isn't actually a research method.
Here's one that is.
Start by confirming you have the right person
This sounds obvious until you've spent an hour researching someone's LinkedIn only to realize, mid-interview, that it belongs to their namesake. Common names produce multiple plausible-looking profiles, and web searches for a person's name return whoever else shares it — students, other professionals, sometimes people in completely unrelated fields. Before you build questions on a fact, check that the fact is actually attached to your guest and not someone who happens to share their name.
Layer your sources — don't rely on one
A single LinkedIn profile tells you a career timeline. It doesn't tell you how someone talks, what they're currently excited or frustrated about, or what they've already said publicly that's worth building on. Good guest research pulls from several places and looks for where they overlap or contradict:
- Professional history (LinkedIn, a company bio) — the factual skeleton: roles, dates, education.
- Their own recent public writing (X, LinkedIn posts, a blog) — this is where tone and current thinking actually live, and it's the best source for a question only they could answer.
- Past interviews and talks — critical, and the step people skip. If a guest has done fifteen podcasts already, several of your planned questions have probably already been asked. Search for their name plus "interview" or "podcast" before you finalize your list.
- Recent news about them or their company — anything from the last few weeks keeps the conversation current instead of reading like it was prepped six months ago.
Build questions from specifics, not categories
"Tell me about your career" is a category. "You spent two years in application performance monitoring before moving into product analytics — what's the thing about why software feels slow that people in product roles still get wrong?" is a specific question built from an actual detail. The second version is harder to write and much easier for a guest to answer well, because it gives them something concrete to react to instead of an invitation to recite their bio.
A decent test: if the question could be asked word-for-word to five different people in the same industry, it's still too generic. Rewrite it until it only makes sense for this specific guest.
Write down what you don't know — and say so if it matters
Real research always has gaps: a guest with limited public presence, a recent job change with no announcement yet, a claim you can't verify from public sources. The instinct is to paper over the gap with a safe, generic question. Don't. Either leave it out, or turn the gap itself into the opening — "I could only find information about your work up to last year, so I want to just ask you directly: what have you been focused on since?" is a perfectly good question, and an honest one.
Flag what to avoid before you're in the room
Good research also means noticing what not to ask. A recent layoff at their company, a lawsuit, a controversial post — these show up in the same searches as everything else, and knowing about them ahead of time means you can consciously choose to raise them respectfully, or leave them alone entirely, rather than stumbling into them live because you didn't know.
Doing this properly takes real time — which is the actual problem
None of this is complicated. It's just slow: reading a LinkedIn profile, searching for past interviews, checking recent posts, and cross-referencing all of it against your agenda easily eats an hour or two per guest, and that's before you've written a single question. For a weekly show, that's most of a working day spent on research alone — which is exactly the part Prepod was built to take off your plate: it pulls from a guest's LinkedIn, X, YouTube history and current news in parallel, and hands you a set of talking points with the evidence for each one attached, so you're reviewing and sharpening research instead of doing it from a blank tab.
Whatever tool or method you use, the underlying discipline is the same: specific beats generic, layered beats single-source, and honest gaps beat confident guesses.