Short answer
It's too cheap if more than 65.5% of your buyers would still pay $29, because past that point $29 earns more before fees. Leverprice tests it on your real visitors: start free with a manual split on one product, test wide price gaps because small ones take months at indie volume, and upgrade when the lift pays.
- Raising $19 to $29 loses money only if more than 34.5% of buyers leave: 1 - 19/29 = 0.345.
- In our simulation of the production engine, a product with 30 sales a month needed a median of 230 days to find a price that earned 20% more per visitor.
- Free covers one product and one experiment with a manual split, up to $500 a month in tested revenue.
- Crawlers and link previews, including the bots that unfurl launch posts, always get your standard price and never count.
- Regional prices give each country its own price, set by hand or from 5 tiers by World Bank price level, and a running test keeps each visitor on one variant at that variant's price for their country.
Example: you sell a tool for $19 a month, 40 people buy it each month, and you picked $19 because $29 felt greedy. If $29 kept 70% of those buyers, the month's new customers would bring in $812 instead of $760, and each of them would be worth 53% more for as long as they stay.
Cheap prices don't generate complaints, so nothing ever tells you to change them. Leverprice runs the test on your real traffic: each visitor sees one price from a range you choose, and the price that earns the most per visitor wins.
Why is price testing hard for indie hackers?
Mostly traffic. A product with 2,000 visitors and 40 sales a month can't run the fixed 50/50 test a growth team would, because at that volume a fixed split can take years to answer. In our simulation of the production engine, which shifts traffic toward the better price as it learns, a product making 30 sales a month needed a median of 230 days to find a price that earned 20% more per visitor. At 100 sales a month it took 68 days (full simulation).
Time is the other limit. You're the whole team, so a test has to install in an afternoon and must never break checkout. Leverprice is a few lines of server code, and on any failure your visitors get your normal price.
What should an indie hacker price test first?
Your headline price, with wide gaps between the options. Small differences take too long to show up at indie volume, so make each step big enough to matter.
1. Your headline price, two steps up
Example: you charge $19, so test $19, $29 and $39. Run the break-even first: $29 earns more than $19 as long as fewer than 34.5% of buyers walk away, because 1 - 19/29 = 0.345 (before fees). The price increase calculator does this for any two prices.
2. A one-time product
Example: a SaaS boilerplate or a template pack at $99, $149 or $199. One-time sales are the cleanest thing to test: the first payment is the whole sale, and Stripe Checkout, Polar, Lemon Squeezy and Dodo Payments all confirm it through a signed webhook.
3. Your price outside your home market
If you sell worldwide, a single dollar price is a different ask in every country. Regional prices set a price per country the way App Store Connect does: countries fall into 5 tiers by World Bank price level, and a preset decides what share of the standard price each tier pays, unless you set a country's price by hand. Example: under the balanced preset (100/85/70/55/40%), a $19 plan costs $19 in tier 1 and $7.60 in tier 5, before conversion to the local currency and rounding to a familiar price.
In a running test, each visitor keeps their variant at that variant's price for their country, and you can read results by region with a suggested tier share. How to set regional prices goes deeper.
How does Leverprice fit an indie stack?
It runs on the hosting you already use, and no Leverprice script ships to the browser. The TypeScript SDK works on Vercel and Netlify functions, Cloudflare Workers, Bun, Deno, AWS Lambda and Node 20+, and there are SDKs for Python, PHP, Ruby and Go; the Next.js guide walks through a full setup. A Carrd or Framer landing page with no backend can use no-code checkout links instead.
Payments stay with your provider: Stripe, Polar, Dodo Payments or Lemon Squeezy. You paste a webhook signing secret, never an API key, and all four verify one-time sales (on Stripe, through Checkout). For subscriptions, Polar verifies the first payment and Dodo Payments verifies one-time payments only; a Stripe subscription's first payment is reported with trackPurchase() and shows as unverified. Lemon Squeezy stops processing payments on March 31, 2027 (announcement); Leverprice verifies its sales until then, and you connect the new provider after you move.
If you build with Claude Code, Codex or Cursor, copy the integration prompt from the dashboard's Install page into your agent; it names the secret key by its environment variable and never contains it. Later, npx leverprice setup connects the same agents to Leverprice's MCP server, so you can read results or lock in a winner from your editor.
How do you set up your first price test?
- Create a free project and put the test key in
LEVERPRICE_SECRET_KEYon your server. - Paste the Install prompt into your coding agent, or install the SDK yourself and add the product with
fallbackPrice: 1900,minPrice: 1900andmaxPrice: 3900. - Show
quote.priceon your pricing page and append everyquote.setCookievalue. Ask for the price again at checkout, chargequote.priceinquote.currencyand passquote.metadatawhole. - Paste your provider's webhook signing secret into Payments.
- Start the test, buy once in test mode to check the purchase arrives, then switch to the live key.
Which Leverprice plan fits an indie hacker?
Start on Free. It runs a manual A/B split on one product and one experiment, up to $500 a month in tested revenue, with all four providers. You choose the split and read the results yourself.
Pro adds automatic optimization: the engine moves traffic toward the better price as evidence comes in, so fewer of your scarce visitors see the losing one. It costs $29 a month or $199 a year (about $17 a month) and removes the revenue cap. Lifetime is $299 once for everything in Pro, limited to 50 spots, which is less than two years of Pro billed yearly ($398). Paid periods are non-refundable, so try Free or monthly first.
Frequently asked questions
Can I price test with only a few hundred visitors a month?
You can run it, but expect a long wait and only large differences to show. At a 2% conversion rate, 1,500 visitors a month is about 30 sales, the smallest volume in our simulation, where a 20% better price took a median of 230 days to find. With less traffic, narrow the range with a survey and test two prices far apart; price testing with low traffic covers the options.
Will launch traffic from Product Hunt or Hacker News skew the test?
Bots won't: crawlers and link previews never get a test price or count as visitors. Launch-day humans do get enrolled and can behave differently from your usual traffic, so start the test a week after launch or let it run long enough that the spike is a small share.
What happens if Leverprice goes down during my launch?
Your normal price shows. If the SDK gets no signed config within 500 ms, getPrice() returns your fallbackPrice and doesn't throw, and LEVERPRICE_DISABLED=1 switches the test off entirely.
Do buyers who saw a lower price keep it?
During the test, yes. Each visitor keeps one price through a signed first-party cookie, and signed-in users keep it on every device through the user id you pass. When you lock in a winner, everyone gets the winning price within about a second.