Start from value and cost of the problem, not from your costs
The wrong question is "what does it cost me to run this." Cloud and your time are nearly fixed at small scale, so cost-plus pricing for software almost always lands too low. The right question is "what is this problem costing the buyer right now, and what fraction of that can I credibly capture."
Quantify the status quo in the buyer's own terms. If your tool saves a freelancer five hours a month and they bill 60 dollars an hour, the problem is worth roughly 300 dollars a month to them. You are not going to charge 300, but that number tells you a 9 dollar price is leaving almost everything on the table. Anchor to value, then discount to a price that closes.
- Time saved times their hourly rate, per month.
- Money made or recovered (extra sales, fewer refunds, recovered churn).
- Pain avoided (a fine, a missed deadline, an angry client, a 2am page).
Pick a pricing model that matches how value grows
Before you pick a number, pick a structure. The model should scale with the value the customer gets, so that bigger accounts naturally pay more without you renegotiating. Tie the meter to the thing that grows as they succeed.
Keep it to one main axis. Solo founders bury themselves trying to combine seats, usage, and feature tiers all at once. Buyers cannot estimate their bill, and you cannot build the metering. Choose the single dimension that best tracks value and lead with it.
- Per seat: good when value scales with team size (collaboration tools).
- Usage based: good when value scales with volume (API calls, emails sent, rows processed).
- Flat tiers: simplest to sell and to build; great for a first version.
- Per outcome: charge per result delivered. Powerful but hard to meter early, so usually a later move.
Set the first number with anchors, not vibes
Build three tiers, not one. Humans choose by comparison, so a lone price has nothing to push against. A common shape is a cheap entry tier, a middle tier you actually want most people on, and a higher tier that makes the middle look reasonable. Design the middle to be the obvious choice and let the top tier do the anchoring.
For the actual numbers, triangulate. Look at what the nearest competitors charge and where the gaps are. Sanity check against the value math from section one so you are capturing a small slice, not a rounding error. Then bias upward, because new founders almost universally pick a number that is too low and then struggle to raise it later.
- Three tiers, with the middle one designed to win.
- Round, confident numbers (29, 49, 99) read as more credible than 27.50.
- Offer annual at roughly two months free to pull cash forward and cut churn.
- When unsure between two prices, ship the higher one. Discounting down is easy; raising is hard.
Test the price on real buyers before you trust it
Stated willingness to pay is unreliable. People are polite in surveys and ruthless with their cards. So test against behavior. A landing page with live tiers and a checkout that captures intent tells you far more than any "would you pay for this" question. A smoke test where the buy button leads to a short waitlist or a real Stripe page is the cleanest early signal you can get.
The Van Westendorp set of four questions is a useful directional tool when you cannot yet sell: at what price is this too expensive to consider, too cheap to trust, getting expensive, and a bargain. It will not hand you the exact number, but it brackets a sane range and exposes when buyers expect to pay far more or far less than you assumed.
- Run a fake door or smoke test with prices visible and measure click-to-checkout.
- Watch which tier people pick, not just whether they buy.
- Talk to people who bounced at checkout. The objection is the data.
Avoid the cheap trap
Underpricing feels safe and is usually the most expensive mistake you can make. A low price attracts the most demanding, least loyal customers, the ones who churn fast and file the most support tickets. It also starves you of the margin you need to do support, marketing, and the next feature. Cheap does not mean easy to sell; for B2B especially, a price that is too low signals the tool is not serious.
Raising prices is allowed and normal. Grandfather existing customers at their current rate, announce the new pricing for new signups, and watch conversion. If conversion barely moves when you raise, you were too cheap. The goal early is not to maximize signups; it is to find a price that filters for customers who get real value and stick around.
Treat your price as a living number
Pricing is never finished. Revisit it every quarter against your real numbers: what customers actually pay on average, how much it costs to acquire them, and how long they stay. When those line up, your pricing is healthy. When acquisition cost creeps above the value a customer brings over their lifetime, the price (or the model) is wrong, not your ad spend.
Change one variable at a time so you can read the result. Adjust a tier, test a new anchor, or move the metering, then give it a few weeks before the next change. Pricing improves the same way the product does: small, deliberate experiments, each one measured.