I Built a Pricing Experiment Tool for Solo Founders — Here's What I Learned
Six months ago, I talked to 40 solo founders about pricing. Every single one said the same thing:
"No. I'm scared of losing customers."
I built PricingSim to solve that.
What PricingSim Does
Connect your Stripe, Gumroad, or Shopify store — the app analyzes 90 days of sales, estimates your price elasticity using Bayesian inference, suggests 2–3 conservative test prices with confidence scores, and generates a live A/B experiment page. When the experiment completes, one click applies the winner or rolls back to your original price.
Why Traditional A/B Testing Fails for Solo Founders
Traditional A/B tests need 200–1,000 conversions per variant for significance. At 50–100 monthly sales, that's a 2-year wait. Bayesian inference answers the right question instead: "What's the probability Price B generates more revenue?" — answerable with 20–40 data points.
Three Things I Learned Building It
Conservative defaults win. Adding hard caps (max 2.5× current price) and a downside floor (p05 ≥ 95% of current revenue) increased usage significantly. Founders don't trust aggressive suggestions.
Real-world CSVs are messier than you think. Gumroad changed their export format twice in 2024. We switched to fuzzy column detection — matching by substring, not exact header names.
Supabase RLS is excellent but requires planning. Getting Row Level Security right at the start saved an entire category of data isolation bugs.
The Stack
- Next.js 15 (App Router, TypeScript)
- Supabase (PostgreSQL + Auth + RLS)
- Vercel (deployment + AI Gateway)
- Stripe (checkout + billing portal)
- Engine: Pure TypeScript — Normal-InvGamma conjugate model, zero external deps
See the technical deep dive for the full engine implementation with code.
Try it free (no credit card): pricingsim.com