Free
Test prices yourself.
$0forever
- Manual A/B split
- 1 product, 1 experiment
- Up to $500/mo in tested revenue
- Stripe, Polar, Lemon Squeezy, Dodo
Price testing software
Set a price range. Leverprice shows each visitor one price, counts only purchases your payment provider confirms, and moves traffic to the price that earns the most per visitor. Your checkout and your keys stay yours.
Leverprice is price testing software for SaaS and other software products. It runs on your server through open-source SDKs for TypeScript, Python, PHP, Ruby and Go, works with Stripe, Polar, Lemon Squeezy and Dodo Payments, and uses Thompson sampling to find the price with the highest revenue per visitor.
1.0
Keep your current price as the control and add two or three prices you'd be happy to charge. The SDK refuses anything outside your minimum, maximum and currency, even if a config says otherwise.
2.0
Each visitor gets one price and keeps it on every return visit. Crawlers and link previews always get your standard price and are never enrolled.
const quote = await leverprice.getPrice("pro", { request, userId: user?.id });
for (const cookie of quote.setCookie) headers.append("Set-Cookie", cookie);
render(quote.price); // 2900 = $29.00, the same number every time this visitor returns3.0
Pass the quote into the checkout you already have. Purchases come back through your provider's signed webhooks, so only money that really arrived counts, and refunds count against the price that sold them.
await stripe.checkout.sessions.create({
mode: "payment",
line_items: [{ quantity: 1, price_data: {
currency: quote.currency, unit_amount: quote.price, product_data: { name: "Pro" },
} }],
metadata: quote.metadata, // signed attribution, passed whole
success_url, cancel_url,
});4.0
Every few minutes the engine shifts visitors toward the price that earns the most per visitor. When one price is 95% likely to be best, or the rest are practically a tie, lock it in with one click. It is live for everyone in about a second.
We ran 960 simulated price tests on the production engine. A fixed even split keeps a third of your visitors on the worst price until you stop it. Leverprice doesn't.
40 days
median time to find a price that earns 10% more per visitor, at 300 sales a month.
0 of 870
decisions settled on the worst price. 94.5% picked the best one.
8–15%
of visitors saw the worst price during a test, instead of 33% in an even split.
No signed config within 500 ms, an outage or LEVERPRICE_DISABLED=1: visitors see your normal price and checkout keeps working.
Prices are decided on your server by open-source SDKs. No script on your pages, no flicker, nothing for ad blockers to break.
Leverprice never asks for an API key that can move money. A webhook signing secret can confirm a purchase and nothing else.
Configs are signed through an offline root key, versions only move forward, and every price is checked against your local limits.
A signed first-party cookie keeps each visitor on one price. Search engines see your standard price, so snippets and structured data stay consistent.
Claude, ChatGPT, Cursor and other MCP clients can create tests, read results and lock in winners as you, with an audit log and read-only tokens.
| Leverprice | Client-side test scripts | Feature flags + your code | |
|---|---|---|---|
| Where the price is decided | Your server | The visitor's browser | Your server |
| What it optimizes | Revenue per visitor, traffic moves to the leader | Usually conversions, even split | Whatever events you send |
| Revenue source | Signed webhooks from your provider | Page events | Events you send or import |
| Checkout price matches the page | Yes, from the same quote | You wire it up | You wire it up |
| Crawlers and previews | Always your standard price | Depends on the tool | You build it |
| If the service is down | Your normal price within 500 ms | Flicker or original page | Depends on your flag defaults |
Tool by tool, with prices and limits: the best price testing software in 2026.
Turn on regional prices and a visitor from India sees ₹999 where a visitor from the US sees $29. Countries are grouped into five tiers by World Bank price level, every price is charged in local currency with the endings people expect, and you can set any country by hand.
Pick a preset from gentle to full parity and adjust the share each tier pays. Exchange rates only move a price when they drift more than 10%.
The SDK reads the country header your host sets, never anything the browser sends, and refuses any currency or price outside the bounds you allow.
In a running test each visitor keeps one variant at their country's price. Read results by region and move a tier's share when it earns more. How regional pricing works.
Stripe, Polar, Lemon Squeezy and Dodo Payments for revenue. TypeScript, Python, PHP, Ruby and Go on the server. A no-code mode with hosted checkout links for sites without a backend.
Free: you split traffic between prices yourself. Pro: Leverprice finds the best price for you.
Test prices yourself.
$0forever
Leverprice finds your best price.
$199per year · about $17/mo
$29per month · cancel anytime
Pro forever. Only 50 spots.
$299once
50 of 50 left
A normal A/B test splits traffic evenly until you stop it and usually counts conversions. Leverprice measures revenue per visitor from purchases your payment provider confirms and moves traffic toward the best price while the test runs, so fewer visitors see a losing price.
No. Your server creates checkouts with your own keys, exactly as it does today. To verify revenue you connect your provider's webhook signing secret, which can confirm events but cannot move money.
Visitors see your normal price. The SDK waits at most 500 ms for a signed config, then falls back to the price you set in code, and it never throws because of us.
Each visitor keeps the same price on every visit through a signed cookie. Charge exactly the price you showed, and keep a short answer ready for support if someone compares notes.
It depends on your sales volume and how different the prices are. In our simulations of the production engine, a price that earned 10% more per visitor was found in a median of 40 days at 300 sales a month and 14 days at 1,000.
Random price tests are generally legal for software in the US, EU and UK, with rules on transparency and personalized pricing that are worth knowing. Our guide covers them; it is not legal advice.
Install the SDK, set the range you'd accept, and let real visitors decide. Your current price stays the control the whole time.