how to price your product

Hey,

for most indie vibe coded products, picking a price point is more about looking at a competitor and undercutting a little bit, vs. using a battle tested formula that ensures you’re able to acquire users profitably, manage float, and re-invest.

In this email, I break down how to calculate a pricing band that will allow you to successfully go to market, scale, and be competitive.

A price works when what a customer pays you over their lifetime comfortably beats what it cost to get them. The standard bar for “comfortably” is a 3:1 ratio of lifetime value to acquisition cost; the median healthy B2B software company sits right around there.

On a napkin, that’s:

Four numbers. Price is what you charge per month. Margin is what’s left after your cost to serve one user (near 100% for plain software, very much not for AI products, more below). Months-they-stay is retention. CAC is what one paying customer costs to acquire. Everything else in pricing content, the psychology, the charm prices, the tier games, is decoration on top of this equation. The rest of this issue is filling in the four numbers cheaply and then solving for price.

CAC feels unknowable before launch. It isn’t. Two ways to ballpark it this week:

1. Free version (20 minutes): open Google Keyword Planner and look up the cost per click on the searches your buyer would type. That CPC is the top of your funnel.

2. Paid version ($100-200, one weekend): put up a landing page with a price on it and run a small Meta ads test at your audience. You get a real CPC, a real signup rate, and a first read on whether anyone clicks “buy” at that price. Same page, two ad sets, two different prices, and it quietly becomes a willingness-to-pay experiment too.

Then walk the funnel. Say clicks cost $1.50, 5% of visitors start a trial, and 20% of trials convert to paid. That’s $30 per trial and $150 per paying customer. Until you have your own data, plug in the public benchmarks: opt-in free trials convert around 4-6% at the median, and card-required trials around 25-35%, freemium runs 3-5% free to paid, and on mobile a hard paywall converts trials to paid at roughly 11% versus 2% under freemium. Those spreads are not rounding errors: the same product can convert five times better in a different funnel shape. Pick yours deliberately.

Classic pricing advice assumes serving one more user costs you nothing. If your app calls a model on every action, that assumption is dead, and it decides your pricing model for you.

The arithmetic, using mid-2026 API prices (a typical action burning 6,000 input and 2,000 output tokens costs about 3.5 cents): a median user doing 120 actions a month costs you about $4.20. A power user doing 600 costs $21. On a $20 flat plan you are profitable on the median user and underwater on exactly the users who love your product most.

So match the model to the cost:

• Recurring cost gets a recurring price. Pieter Levels sells Nomad List lifetime memberships because the average member churns out in about three months anyway, so the math favors him. PhotoAI, where he pays for GPUs on every generation, runs on subscriptions and does about $105K a month, $80K of it profit, per his own writeup. Same builder, opposite models, both chosen by cost structure.

• One-time pricing needs near-zero marginal cost. Tony Dinh’s TypingMind charges once and has users bring their own API keys, which moved the token bill off his books entirely: $22K in the first week, no inference risk.

• Credits when usage varies 10x between users. Base44 metered credits from day one and banked $189K profit in a single month while paying its own inference bills, then sold to Wix for $80M six months after launch. Cursor went the other way, sold “unlimited,” and spent July 2025 apologizing and refunding surprise charges. If a company valued near $10B can’t make flat unlimited survive token costs, your side project can’t either.

If you keep one rule from this section: cap included usage around 3x your median user, and never print the word unlimited above a metered cost.

Back to the napkin. Your test said $150 CAC. You need $450 of lifetime value. At 80% margin and six months of average retention, that’s $450 / (0.8 x 6), or about $94 a month. Almost nobody’s gut says $94. And when the napkin spits out a number like that, the broken input is rarely your ad costs. It’s the price you were planning to charge.

If $94 sounds impossible for your product, the equation has two honest exits. Either move to a buyer for whom $94 is a line item, meaning sell the same capability to businesses instead of consumers, or change the channel, meaning grow somewhere clicks don’t cost $1.50. What you don’t get to do is keep the consumer price and the paid channel and hope.

Run it in reverse and the second exit gets sharp. A $19 product with six-month retention and 80% margin generates $91 of lifetime value, so you can afford about $30 of CAC, which at benchmark conversion backs all the way out to a 30-cent click. Meta has not sold 30-cent clicks to a US software audience in a very long time.

Cheap products can’t buy ads. Your price picks your marketing channel, not the other way around.

That single line explains most indie marketing frustration. The $19/month product isn’t failing at ads; it was never allowed to use them. It has to grow on SEO, communities, and word of mouth, which is fine, as long as you chose that on purpose.

Three calibration notes before you run it. First, retention is the number you’ll guess worst, so use defaults until data arrives: 4-6 months for consumer, 6-12 for prosumer, around 24 for B2B, then re-run the napkin quarterly. Second, the floor is structural: on a $3 price, Stripe’s standard 2.9% plus 30 cents takes about 13% before you see a cent, so sub-$5 prices are broken regardless of what the equation says. Third, “just charge more” is a strong default and still not a law: Anthony Castrio raised his community from $29 to $49, watched signups stall through a year that ended five figures in the red, reverted, and recovered. He published the whole P&L, which is exactly the kind of generosity that makes the rest of us smarter. If your product depends on easy trial, the entry price is part of the funnel, not just the revenue line.

First prices aren’t tattoos either. Grandfather existing customers, raise on new ones, and keep raising until conversion flinches. In fact, for nearly all my products i’ve aggressively tested pricing. no one notices. That’s part of the game anyway.

• Apple pulled calorie app Cal AI from the App Store over a checkout dressed up to look like Apple’s own payment sheet, then reinstated it within a day once fixed. The hard-paywall meta works, and now has a referee.

• AppSumo’s founder says revenue fell by half over two years as the lifetime-deal machine sputters. A lifetime deal is deferred cost with no deferred revenue; even the marketplace built on them is feeling it.

• Screen Studio’s founder publicly said he regrets his $229 one-time price as the app moved to subscriptions. Changing models mid-flight cost him reputation even though he handled existing customers generously. Pick the model that survives year three on day one.

Make your agent run the napkin. Paste this into Claude with your real numbers:

“Here is my product: [one paragraph]. My buyer’s likely search terms and Keyword Planner CPC: [terms, $X]. My funnel shape: [opt-in trial / card-required trial / freemium / hard paywall]. Estimate my per-user monthly cost to serve, including AI tokens at median and heavy usage if my product calls models. Using standard conversion benchmarks, walk the funnel to a CAC estimate, then solve price x margin x months >= 3 x CAC for my minimum viable price with retention defaults for my category. Show every step, state every assumption, then tell me: the price, the model (subscription, one-time, or credits), and whether that price can afford paid acquisition or requires an organic channel.”

In two minutes, you’ll know more about your pricing than most funded startups know about theirs :D

• Anthony Castrio’s full post-mortem of his $49 pricing mistake: rare honest numbers, and the counterweight to every “just raise prices” thread.

• Patrick McKenzie on why your pricing, not your product, doubles revenue: fourteen years old and still the canonical argument for charging more.

• Lenny’s interview with Base44’s Maor Shlomo: how a solo founder took an AI product to $1M ARR in three weeks without losing money on it.

• The full timeline of Cursor’s pricing meltdown: read it before you write the word “unlimited” anywhere near a paywall.

• Tony Dinh on making $22K in his first week: the bring-your-own-keys playbook, from the builder himself.

• What happens to lifetime deals when the company changes hands: VPNSecure’s new owners axed thousands of “lifetime” accounts. Debt comes due.