The case at a glance
- Product: AI mock interviews with live feedback, sold as day-sized passes
- Launch: January 2026 → $10K MRR by July 2026 (~7 months)
- Pricing: $12 day pass, $19 for 72 hours, $39/mo (minority buyer)
- Team: 1 solo founder, no prior audience
- Distribution: Programmatic SEO + niche forums (Blind, 1point3acres, LeetCode) + AI UGC on TikTok
- Seasonality: Two clear spikes tied to spring and summer hiring cycles
Stack. Next.js web app, OpenAI/Anthropic model calls, Stripe for micro-charges, programmatic SEO landing pages per company + role combo, and an AI UGC pipeline producing situational short-form video. Total infra spend stays under $200/mo until well past $5K MRR. No installs, no onboarding, buy-and-use-tonight UX.
What he's actually selling (and why subscription thinking breaks it)
The product isn't AI interview coaching. The product is removing panic before a $180K salary conversation tomorrow morning.
That reframe changes everything. Candidates aren't optimizing LTV. They have one specific Meta loop on Thursday. A $12 charge that removes the sweating converts at rates SaaS founders dream about, because the alternative is showing up unprepared to a six-figure decision. Twelve dollars versus six figures. That ratio is the entire business.
This is why every "improvement" pushing buyers toward a monthly plan quietly destroys the economics. The buyer doesn't want another subscription. They want tomorrow to go well. The operator himself frames one-time revenue as "less stable" in his own post. It isn't. It's honestly priced at the exact moment of maximum willingness to pay, which is a structurally stronger business than pretending someone needs interview prep eleven months a year.
The mechanism works because three things landed at once: AI good enough to simulate a real loop, programmatic SEO cheap enough to own the long tail of company-specific queries, and AI UGC that solves the demo problem invisibly (you show a candidate mid-panic, not a feature list). That window opened in 2025.
Sell the hour, not the software. Price at the moment of maximum fear, not maximum retention.