A free mock interview with no signup, by role and level — here's what that costs us
Google's Interview Warmup proved people want to try before they register. We built the same idea, but scoped to a specific role and seniority level instead of generic questions — and here's what giving away the demo actually costs.
Google shut down Interview Warmup's growth a while back, but the idea behind it was right: most people will not create an account to find out if a tool is worth using. They'll try it, or they won't try anything.
So the demo on Prepair works the same way. Pick a role — QA, backend in a specific language, frontend, whatever's closest to what you actually do — pick a seniority level, answer one real question, see the score. No email, no card, no "sign up to see your results."
What's different from a generic warmup tool is the scoping. "Practice interview questions" with no role and no level attached tends to produce either something too easy to be useful or too generic to match what you'll actually be asked. A senior backend question and a junior QA question have almost nothing in common, and a tool that doesn't ask which one you are is guessing.
What giving this away actually costs
I want to be specific here rather than vague, because "free" gets thrown around without anyone saying what it means for the person offering it.
Every demo answer gets graded by a model call. That's a real cost per attempt, not a rounding error — it's the same cost as a paid interview, just with nobody paying for it. The honest number is smaller than most SaaS discourse implies, but it isn't zero, and pretending otherwise is how free tiers quietly die a year later.
The bet is that a fraction of people who try the free question come back for the second one, hit the free-tier limit, and some smaller fraction of those people decide three interviews a month isn't enough and pay for more. That's the whole model. If nobody converts, the free demo is a cost center with no return, and eventually it either gets cut or gated — which is exactly the trajectory that killed the tool this idea is modeled on.
What I got wrong the first time
The first version of this demo asked for an account before showing a single question. I built it that way because that's the default — every SaaS onboarding tutorial says to capture the email early. It's also exactly backwards for proving the thing works, which is the entire point of a demo.
The fix wasn't complicated: move the account wall to after someone sees their score, not before they see a question. The awkward part was admitting the original version existed for weeks before I noticed how much friction it was creating, because I never used my own product as a stranger would.
Where this actually goes
If you're deciding between five tools before spending any money on any of them, this is meant to be the fastest one to rule in or out — one real question for your actual role and level, answered honestly, no account. If it's not useful after that, you've lost two minutes, not an email address you now have to unsubscribe from.
More of the questions people ask about this kind of practice are answered in the AI mock interview FAQ.
Try the role-and-level demo — pick what you actually do, answer one question, see the score.