SaaSPricingSubscriptionMVP
How to price a B2B SaaS product
10 min read
A B2B SaaS price is set from the value the product delivers, between two limits: your cost to serve each customer at the bottom, and what the customer already pays to solve the same problem at the top. In between, you choose a unit the price scales with (seats, volume, features), build two or three plans, and test them with your first customers.
The big pricing playbooks are written for companies with thousands of customers and a pricing team. This guide is for the earlier stage: your first price, set with little data, that you can correct without breaking trust. It goes deeper than the pricing step in our guide to launching a B2B SaaS as a non-technical founder.
The three limits on a price
The floor: your cost to serve
Before thinking about value, work out what one active customer costs you each month: their share of hosting and database, payment fees on every charge, transactional emails, support time and, if your product calls an AI model, the cost of those calls. This floor does not set your price. It tells you where you cannot go.
Payment fees are published by providers; our guide to SaaS subscription billing with Stripe breaks them down. Other running costs are listed in how much it costs to build a SaaS platform.
The ceiling: value to the customer
A business customer pays for an outcome: time saved, mistakes avoided, a compliance requirement met, more revenue. Put a figure on that outcome with the customer, in an interview, instead of guessing it.
A made-up example, to show the method: if your tool saves two hours a week for each of three office staff at a small company, the value runs to dozens of hours a month. Your price can capture a fair share of that; the customer keeps the rest, which is why they sign.
The reference point: what they use today
Your prospect always compares with something: a spreadsheet, a generic off-the-shelf tool, an agency, a new hire, or doing nothing. If the alternative is free (a spreadsheet), your case is time and errors. If it is expensive (a full suite), you can compete on simplicity.
SaaS pricing models
Stripe's documentation lists four recurring models: flat rate per tier, per seat, tiered and usage-based. In practice, products mix them.
| Model | How it works | Fits when | Watch out for |
|---|---|---|---|
| Flat rate per tier | Two or three plans at fixed prices | Value depends on features | Plans so close they cannibalise each other |
| Per seat | A price per user account | The product is collaborative, each user adds value | Customers sharing one login |
| Tiered | Unit price drops as volume grows | Big customers should pay more, but not linearly | A grid too complex to explain on a call |
| Usage-based | Customers pay for what they consume | Your costs follow consumption, AI included | Bills buyers cannot predict |
| Hybrid (base + usage) | A plan includes a volume, overage is charged | You want stable revenue and a protected margin | Explaining the overage clearly |
Two entry options sit on top of these models:
- Freemium: users start free and upgrade when they hit a limit. It needs a very low cost to serve and a product people can learn on their own.
- Free trial: everything is open for a set period, then a subscription is required. It suits products that need some setup before they show their value.
Choose your value metric
The value metric is the unit your price grows with: seats, locations, documents, cases, projects, API calls. It shapes your business more than the price point itself. A good metric passes four tests:
- It grows with the value received. A customer who gets more pays more without feeling taxed.
- The customer understands it and can predict the bill.
- It is hard to game. Per-seat pricing invites shared logins; pricing per case processed, much less so.
- Your application can measure it. You cannot bill on a number you do not count reliably.
For vertical B2B software, the most natural metric is often the core object of the trade: the quote for a contractor, the case file for a law firm, the location for a multi-site business.
Usage-based pricing vs subscription
AI made this question urgent. When each user action triggers a call billed by a model provider, an unlimited subscription can lose money on your heaviest users.
- A flat subscription reassures buyers and makes your revenue easy to forecast. It works when cost per customer barely varies.
- Pure usage pricing ties revenue to costs. It worries buyers who need to get an annual budget approved.
- A plan with an included volume and a known overage price gives you both: the buyer can budget, and you stay covered.
- Prepaid credits are another form of usage pricing: the customer buys a balance and draws it down.
Build the price list
- Two or three plans are enough at launch. Each needs a reason to exist: an audience, a metric, a set of features.
- Quote prices excluding tax for business customers, and say so. Sales tax and VAT depend on where the customer is.
- Offer annual billing, with a discount if you like. It helps cash flow and cuts fixed payment fees.
- Keep a "contact us" tier for large accounts that need integrations, dedicated hosting or a specific service level.
- Publish your prices if your sales cycle is short. Small and mid-sized buyers want to know before they book a call.
How the subscription model compares with paying for a build up front, from the customer's side this time, is covered in software subscription vs upfront build.
Test, then adjust
A first price is a hypothesis. You validate it three ways:
- In interviews, before the product exists: show a prototype and state a price. The reaction tells you more than a survey.
- With the first paying customers: watch who signs, who hesitates, who picks which plan.
- In usage data: which customers hit their limits, which features actually get used.
To change a price, follow two rules. Give existing customers notice, under the terms of your agreement; we cover the relevant clauses in the SaaS agreement clauses your customers will ask for. And decide explicitly what happens to early customers: many companies keep them on the old price for a while.
The most common mistakes
- Pricing from your build cost. Customers do not pay for your budget, they pay for their outcome.
- Starting too low "to get traction". A low price attracts price-sensitive buyers, and raising a price is harder than lowering one.
- Too many plans. Beyond three, buyers cannot choose and you cannot tell which plan works.
- Ignoring AI costs. Unlimited use of a service billed per call is a bet on your customers' behaviour.
- Hard-coding prices. Every pricing test then becomes a release.
What we learned from our own products
LPR Studio builds, owns and runs two live products. Izi-Devis, our quoting and invoicing software for tradespeople, has a free plan capped at five quotes with no card required and a Pro subscription with a published price: the entry metric is the core object of the trade, the quote.
Tubissime, our AI-generated personalised song service, is paid per order, and its creative studio runs on credits. Its back office shows generation costs, revenue and margin for each customer, and prices can be changed without redeploying. That measurement is what makes a fair price possible when every order has a variable cost.
Where to start
- Estimate your cost to serve one customer, even roughly.
- Interview five prospects about what the problem costs them today and what they use instead.
- Pick a value metric that passes the four tests.
- Sketch two or three plans, excluding tax, monthly and annual.
- Make prices configurable in the software, so you can change them without a development cycle.
- Review the price list after your first ten customers.