What manual QA actually costs (we did the math)

We talked to about 50 startups over the past year. Mostly seed to Series B. All of them had some version of the same setup for testing signup and onboarding flows.

The pattern: one or two QA engineers manually testing critical flows before each release. They maintain a list of test accounts. They have a shared phone for SMS verification. They log in with real credentials, walk through the flow, eyeball the result, and write up bugs in Jira. The whole thing takes a few hours per release.

Sounds manageable. Here's where the math gets bad.

A mid-level QA engineer costs $85K-$120K/year depending on market. If they spend even 30% of their time on manual signup and identity-related testing, that's $25K-$36K/year — on one person. Most teams have two. So you're at $50K-$72K/year on manual testing that could be automated. And that doesn't count the opportunity cost of what those engineers could be building instead.

Then there's the release velocity problem. Manual testing is a bottleneck. If QA needs half a day to verify flows before a deploy, and you deploy twice a week, you've lost a full day of engineering time per week just waiting.

The reliability problem is worse. Manual tests miss things. Not because the testers are bad — because humans get tired, skip edge cases, and can't realistically test 15 OAuth providers on every release. Automated tests catch regressions that manual testing misses.

We built a calculator for this (it's on the pricing page). For a team of 3 engineers deploying twice a week, the average time saved with Zantiq is about 40 hours per month. At a blended rate, that's roughly $6,000-$8,000/month in reclaimed engineering time. A Zantiq plan costs a fraction of that.

I'm not saying fire your QA team. I'm saying let them work on the hard stuff — exploratory testing, edge cases, test architecture — and let the AI handle the repetitive signup flows that nobody wants to do anyway.