How to Price a New SaaS Product: Three Decisions, Four Questions, and the Only Pricing Test That Survives Launch

A founder spends three months building a small SaaS dashboard for indie e-commerce stores. The product is solid. The first five users, all from a Reddit thread, sign up for the $9/month plan. The founder is elated. A year later, the company has 400 paying customers, $43,000 in annual recurring revenue, a 6% monthly churn rate, and a quiet, persistent feeling that the price is wrong. Raising it loses customers. Keeping it starves the product. The original $9 was a guess, and the guess is now baked into the renewal curve.

The product is fine. The market is fine. The price is the problem, and the price was set in the first week, when there was no evidence to set it with.

This is the most common shape of an early-stage SaaS pricing failure, and it is not solved by reading more pricing articles. It is solved by treating the price as a testable hypothesis from the first day — one with a model, a value metric, a price point, and a willingness-to-pay experiment that survives contact with a real credit card. That is the work this article walks through, in the order a founder actually owns it.

This article is the "how do I set the price?" companion to How to Test Whether People Will Pay Before You Build. That article owns the six tactical experiments and the seven-day payment plan; this one owns the three pricing decisions a founder makes before any of those experiments run. If you are new to the discipline, read the Startup Validation hub first. If you already have a SaaS-shaped idea and need to pick a price, start here.

Key takeaways

• Pricing is three decisions, not one. Model (how you charge), value metric (what you charge for), and price point (how much). Most new SaaS founders pick a price by skipping all three and copying a competitor.
• The first price is a hypothesis, not a verdict. Stripe Atlas's published pricing guide explicitly recommends treating the launch price as a starting point that will be revised quarterly, not a number to defend. The damage from a wrong first price compounds; the cost of revising it is low.
• The Van Westendorp Price Sensitivity Meter produces the best behavioural willingness-to-pay data founders can run themselves. Four questions, ten minutes per respondent, a panel of thirty to fifty target buyers. Patrick Campbell's published research at Price Intelligently is the most-cited field methodology.
• Charge at the real price from day one. Discounts to make the first sale easier contaminate the renewal curve. A founder who can only get paid customers at a steep discount has learned that the price is wrong, not that the audience is hard to reach.
• A defensible price is one you can explain. If a buyer asks "why does it cost this much?" and the founder cannot answer without "because that's what I picked", the price will not survive a renewal conversation.
• Pricing changes at six-month intervals produce roughly two times the ARPU growth of annual updates according to OpenView's published 2023 SaaS Benchmarks report. The expected cadence is faster than most founders realise.

Why this matters

Most generic pricing advice for new SaaS products has the same shape: "research your competitors, talk to customers, pick a number." Each of those steps is true and individually useless. Competitors are usually priced as badly as the founder would be, customers cannot tell you what they would pay before they have used the product, and a number picked without a methodology has no defence when the renewal curve starts to bend.

The compounding cost of a wrong first price is what most founders underestimate. A $9 product that should have been $29 will sign up twice as many customers in month one, generate $18 in revenue per signup instead of $29, and quietly set the buyer's mental anchor at a third of where the founder needs it to be. Six months later, raising to $29 means a 220% increase that the renewal curve was never asked to support. Most founders do not raise. They add a higher tier above the $9 plan and watch almost no one upgrade, because the $9 plan covers the buyer's actual usage.

This is the failure mode Stripe Atlas warns against in its published pricing guide: founders anchor on the launch price, then treat later revisions as "risky" rather than as the planned second experiment. The OpenView 2023 SaaS Benchmarks Report finds the same pattern at scale — companies that update pricing every quarter grow ARPU 2–4x faster than those that update annually, and the average SaaS company spends fewer than ten hours per year on pricing. The problem is not the absence of data. The problem is treating pricing as a one-time event instead of an ongoing experiment.

Three decisions, four questions, and an eight-week playbook are the smallest version of that experiment that produces a defensible number. None of it requires a pricing consultant or a market-research panel. It does require treating the launch price as a hypothesis with a known expiry date.

The three pricing decisions a new SaaS founder actually owns

When a founder says "I need to price my SaaS product", they are usually asking one question. They are actually answering three.

Decision one: the pricing model

The model is the structural shape of how the customer pays. The four that cover most new SaaS products:

• Flat-rate subscription. One price for the whole product, billed monthly or annually. Common in tools with a single use case and a tightly defined buyer (Basecamp, Notion before tiers, Mailchimp's earliest plans).
• Tiered subscription. Two to four plans, usually named by user count or feature set. The most common shape for B2B SaaS, because it lets the same product serve a freelancer at one price and a team at another without breaking the mental model.
• Per-seat subscription. A specific form of tiered subscription where the price scales linearly with the number of users. Linear and predictable; expensive when the buyer's per-user value drops off.
• Usage-based. Price scales with a measured unit of consumption — API calls, documents processed, transactions, GB stored. Common in infrastructure (AWS, Twilio) and increasingly in AI products where the unit of value is the query or the token. The hardest model to predict revenue with, and the easiest to underestimate when customers have variable consumption.

Stripe Atlas's published pricing guide and the OpenView Product Benchmarks reports both note that the model is usually inherited from the category, not chosen. A scheduling tool is flat-rate or tiered. An analytics tool is tiered or usage-based. A communication tool is per-seat. The model is rarely the lever that moves a price significantly; it is the structure the other two decisions sit inside.

Decision two: the value metric

The value metric is what the customer is paying for. It is the most under-considered of the three decisions, and it is the one that determines whether the price scales with the customer's success.

A founder who charges per "user" for an analytics tool has chosen a value metric that disconnects from the buyer's actual gain — the buyer gets more value as they process more events, not as they add more users. A founder who charges per "project" for a freelance tool has chosen a metric that caps the buyer's spend at the number of projects, not at the value of the work the tool replaces. The right value metric is the one that goes up when the customer's life gets better, and stays flat when it does not.

Patrick Campbell's published work at Price Intelligently (now Paddle's pricing arm) repeatedly emphasises the value metric as the highest-leverage decision a SaaS founder makes. A well-chosen value metric is the difference between a product that grows linearly with the customer and a product that hits a ceiling at month four and cannot raise the price without churn.

Decision three: the price point

The price point is the number. This is what most founders think they are choosing when they say "I need to price my SaaS product." In practice, it is the decision they should make last — after the model and the metric are set, and after a willingness-to-pay experiment has produced data they are willing to defend.

Three numbers actually matter:

• The entry tier. The price a buyer pays the first time. Low enough that the friction of trying the product is lower than the friction of saying no.
• The annual discount. The percentage gap between monthly and annual billing. Standard SaaS practice is 15–20%. The discount exists because annual billing extends the buyer's commitment and lowers the founder's churn; the size of the discount should reflect how much that extension is worth to the business, not be copied from a competitor.
• The renewal price. The price the same buyer pays in month thirteen. The least visible of the three numbers, and the one that determines whether the business survives year two.

A price point is defensible if the founder can answer "why does it cost this much?" in two sentences, without mentioning a competitor. "Because it's twice as fast" works. "Because everyone in the category is around this number" does not. The answer has to be about the buyer's life, not the founder's spreadsheet.

The four questions that produce real willingness-to-pay data

The single most-cited methodology for willingness-to-pay research in the SaaS pricing literature is the Van Westendorp Price Sensitivity Meter. It was published by Peter Van Westendorp in a 1976 ESOMAR paper and has been the standard pricing-research instrument in consumer goods and SaaS for five decades. The instrument asks four questions of each respondent:

1. At what price would you consider the product too expensive to buy?
2. At what price would you consider the product too cheap to trust the quality?
3. At what price is the product expensive but still worth considering?
4. At what price is the product a great deal for what it does?

The intersection of the four curves identifies an "acceptable price range". The intersection of "too cheap" and "too expensive" identifies the "indifference price point" — the price at which the same number of respondents think it is too cheap as think it is too expensive. That indifference price point is the most robust single number the methodology produces.

Patrick Campbell's published Price Intelligently methodology uses Van Westendorp as the willingness-to-pay layer of a four-step framework (personas → feature preference via MaxDiff → Van Westendorp → tier alignment). The methodology has been used by companies from Atlassian to Wistia, and the documented outcomes include Wistia doubling new sales after a per-video value-metric shift grounded in the same survey data.

Three rules matter when running this for a new SaaS product:

• Survey the right three populations. Current prospects who know the product, target customers who have never heard of it, and current customers if any exist. Stating intent and revealing it through payment are different signals; the panel needs both.
• Keep the survey under four minutes. Non-compensated surveys with completion times over four minutes lose respondents to fatigue and produce unusable data on the tail.
• Run it on at least thirty respondents per population. Below thirty, the curves are too noisy to draw a defensible indifference price. Above one hundred, the marginal signal drops sharply.

This is not a substitute for a real payment test. Van Westendorp produces a defensible willingness-to-pay range from verbal reports; the next step is to confirm the range holds when the same buyers face a real checkout. Patrick Campbell's own caveat, repeated in his public interviews, is that "the biggest mistake in SaaS pricing is mistaking what customers say for what they will actually pay." The Van Westendorp survey narrows the range. A paid pilot narrows it further.

The eight-week playbook that turns the answers into a price

Week one is the value-metric decision. Talk to five target buyers. Ask them to walk through the last time they had the problem your product solves. Ask what they would pay to make that problem go away, in their own words. The point is not the number they say — it is the unit they reach for. If they describe cost in "hours saved", the value metric should track hours or their dollar equivalent. If they describe cost in "projects per month", the value metric should track projects. The conversation tells the founder which unit the buyer is already pricing the problem in.

Week two is the Van Westendorp survey. Build the four-question survey. Send it to thirty prospects who know the category, thirty who do not, and as many current waitlist signups as exist. Plot the four curves. Identify the indifference price point. Expect the indifference price to be 30–50% above whatever number the founder originally considered — Van Westendorp research consistently finds that buyers' verbal willingness-to-pay exceeds their actual willingness-to pay, and the survey gives the founder a structured read on the gap.

Weeks three and four are the value-metric stress test. Take the indifference price from week two and build three pricing pages — flat-rate, tiered, and usage-based — each at the indifference price point. Show them to ten target buyers individually. Ask which they would choose and why. The pattern of answers tells the founder which model fits the buyer's mental model. A buyer who reaches for the tiered page is signalling that they expect the product to grow with them. A buyer who reaches for the flat-rate page is signalling that they want a fixed commitment. A buyer who reaches for the usage-based page is signalling that they cannot yet predict the volume.

Weeks five through seven are the paid pilot. Sell a thirty-day paid version of the product to five to ten target buyers at the indifference price. Discount nothing. Do not grandfather. The point of the pilot is to find out whether the buyers who said they would pay actually pay, and whether they renew for month two. If fewer than 60% of pilot buyers convert and pay month one, the indifference price is above what the buyer will actually pay. If fewer than 60% of month-one buyers renew for month two, the value-metric is wrong, not the price. Renewal is the test the value metric has to pass; conversion is the test the price point has to pass.

Week eight is the read-out and the launch price. Write down the price point that converted at 60% or better, the value metric that retained at 60% or better, and the model the buyers chose most often. That is the launch price. Set a calendar reminder to revisit it in six months. The OpenView 2023 SaaS Benchmarks Report finds that companies updating pricing every six months see roughly twice the ARPU growth of those updating annually; the launch price is a starting hypothesis with a known expiry date, not a permanent commitment.

Common mistakes

Six named failure modes. Most new SaaS founders hit at least three of them.

Picking the price before picking the value metric. A flat-rate monthly price for a product whose value scales with usage is structurally wrong. The first decision is the metric. The second is the model. The third is the number. Reversing the order produces a price the founder cannot raise without churn.

Copying a competitor's price. The competitor's price reflects their value metric, their cost structure, and the four or five rounds of pricing iteration they have already done. None of which match the founder's situation. Stripe Atlas's published guide explicitly warns against using competitor pricing as the primary input, for the same reason: a competitor's price is evidence about their hypothesis, not about the buyer's.

Discounting to make the first sale easier. A steep launch discount trains buyers to expect a discount on renewal. If the founder cannot get paid customers at the indifferent price, the price is wrong. Discounting hides the wrongness behind a temporarily higher conversion rate and contaminates the renewal curve for the life of the business.

Annual-only pricing for a new product. Annual billing is a commitment the buyer has no basis to make on day one. Forcing annual billing at launch converts fewer buyers and produces a cohort with lower intent. Monthly billing with a modest annual discount (15–20%) preserves the conversion signal and lets the founder test the renewal curve in month twelve instead of month one.

Setting the price, then never revisiting it. The OpenView benchmarks consistently find that the median SaaS company spends fewer than ten hours per year on pricing. Pricing is the highest-leverage decision in the business; ten hours a year is a rounding error. Quarterly reviews of the value metric, the model, and the price point are the cost of a defensible price.

Treating a low conversion rate as a marketing problem. If paid traffic converts at 2% on a $29 plan and 1.5% on a $49 plan, the founder has learned something about the price elasticity of their specific audience. They have not learned that they need better marketing. Pricing research is the diagnosis; pricing decisions are the treatment.

Worked example

A founder has built a dashboard for Shopify stores that surfaces the ten products each merchant could cross-sell to a different audience. Two thousand merchants on the waitlist. The product is real. The founder has no idea what to charge.

Week one. Five calls with merchants. The buyers price the problem in "additional monthly revenue from cross-sell" — they want to know what the tool is worth against the revenue it will produce. The value metric is monthly revenue influenced, not users or projects.

Week two. Van Westendorp survey to 100 merchants. Indifference price point lands at $79/month, well above the founder's initial guess of $29. Survey also surfaces a long right tail — twenty merchants say they would pay $300/month if the revenue-attribution is accurate. The cohort splits.

Weeks three and four. Three pricing pages. Flat-rate at $79. Tiered at $39 / $99 / $299. Usage-based at 1% of attributed revenue. Seven of ten merchants pick the tiered page. The founder now has a value metric, a model, and a price range.

Weeks five through seven. Paid pilot to ten merchants at $79. Eight convert. Six of the eight renew for month two. The founder learns the price is acceptable and the value metric holds.

Week eight. Launch. $39 / $99 / $299 tiered, monthly and annual billing, with the $99 plan as the "most popular" anchor. Calendar reminder set for six months. First quarterly review shows that the $299 tier converts at 12% — the survey's right tail was real. The price the founder originally considered ($29) would have left $400,000 of annual recurring revenue on the table at the same conversion rate.

The whole sequence takes eight weeks, costs less than $1,000 in survey tooling, and produces a price the founder can defend without mentioning a competitor. None of it requires a pricing consultant.

How Yibud treats this

The monetization dimension in Yibud's scoring engine is built from six rules across the four pricing decisions above: a subscription model adds to the monetization score, an ad-supported model subtracts, a marketplace model is neutral because the chicken-and-egg risk is structural, and a "not sure" monetization model subtracts more than any specific wrong answer. The rule engine does not pick the price — the price is a founder decision with no defensible default. What it does is flag the absence of a pricing decision as a signal of weak business-model thinking, which is the most common early-stage cause of monetization failures.

The "monetization × distribution" joint signal is also surfaced. A subscription model on paid ads is structurally weaker than a subscription model on SEO or community, because the cost of acquiring the buyer exceeds the lifetime value the price supports. A founder who submits "subscription" with "cold outreach" or "paid ads" will see the joint signal in the report. The pricing decision is upstream of the distribution decision; the engine reflects that.

If you have an idea and want to know whether the pricing shape you have in mind survives the engine's read on it, Yibud's startup analysis takes about five minutes and returns the seven-dimension score, including the monetization dimension, with the rule that fired and the reason it fired.

FAQ

Should I price my SaaS before or after I build the product?

After. The Van Westendorp methodology produces usable data on the willingness-to-pay of a target buyer before the product exists, but only if the buyer can describe the value they expect in concrete terms. For most new SaaS products, this means a clickable prototype, a written description with three sample use cases, or an early-access waitlist with the value proposition spelled out. Pricing without any of those produces numbers from a hypothetical product the buyer will revise once they see the real one.

What if my target market is too small for thirty respondents?

For a niche B2B product with a total addressable market of two hundred accounts, the population of qualified respondents may be smaller than the methodology recommends. In that case, run the survey with everyone in the TAM, then expand the willingness-to-pay inference from comparable-but-larger markets. The data will be noisier and the indifference price point will have wider error bars. The alternative — picking the price without any data — has wider error bars still.

How do I price a freemium SaaS?

Freemium is a separate model with its own pricing logic. The free tier exists to lower the friction of trying the product; the paid tier has to convert at a rate that covers acquisition cost. Patrick Campbell's published view, summarised in his Acquired podcast appearance, is that freemium is "not a pricing model, it's an acquisition strategy". The conversion rate from free to paid is the metric that determines whether freemium works for a specific product. The paid tier inside freemium should still follow the model, metric, and price-point sequence above.

Is usage-based pricing right for my product?

Usage-based pricing works when the value to the buyer scales with consumption. If the buyer's value scales with users (a collaboration tool), per-seat is the natural fit. If it scales with transactions (a payment tool), usage-based is the natural fit. If it scales with neither — a fixed-scope tool like a small CRM — flat-rate or tiered is the natural fit. The OpenView 2023 Product Benchmarks Report finds that value-metric companies grow about twice as fast as feature-differentiated companies, but only when the value metric actually tracks the buyer's gain. A bad value metric is worse than no value metric.

What if my pilot customers pay month one but do not renew month two?

The value metric is wrong, not the price. A founder whose pilot customers convert at 80% and renew at 30% has built a product that satisfies the buyer's first-month curiosity and does not integrate into the buyer's workflow. The fix is to redesign the value metric — not to lower the price. Lowering the price on a product the buyer does not use produces a longer non-renewal curve, not a renewed one.

How often should I change my SaaS pricing?

The OpenView 2023 SaaS Benchmarks Report finds that companies updating pricing every quarter grow ARPU 2–4x faster than those updating annually. Quarterly is the upper bound for most early-stage products; every six months is the floor. Whatever the cadence, it should be on the calendar before launch. Pricing without a scheduled revision date is pricing with an indefinite expiry.

What is the most common pricing mistake first-time SaaS founders make?

Anchoring on the first price and treating later revisions as risky. The first price is a hypothesis. The cost of revising it quarterly is one to two days of work per quarter. The cost of not revising it is the slow erosion of the renewal curve until the founder cannot raise the price without losing customers they cannot afford to lose.

Summary

A defensible SaaS price is the result of three decisions, four questions, and one paid pilot. The three decisions are the model (how the customer pays), the value metric (what the customer pays for), and the price point (how much). The four questions are the Van Westendorp Price Sensitivity Meter, run on at least thirty target buyers per population. The paid pilot is a thirty-day test at the indifference price, with no discounts and no grandfathering, that the founder uses to confirm whether the buyers who said they would pay actually do.

The first price is a hypothesis with a known expiry date, not a verdict. Quarterly revisions produce two to four times the ARPU growth of annual revisions, and the average SaaS company spends fewer than ten hours per year on pricing — which is the cost of a defensible price, not the cost of an undefended one. Founders who treat the launch price as a test will revise it within six months. Founders who treat it as a commitment will defend it until the renewal curve bends.

Related reading

• How to Test Whether People Will Pay Before You Build — the six tactical experiments and the seven-day payment plan. The "what happens when money actually moves" companion to this article.
• How to Validate a SaaS Idea Before You Build It — the SaaS charter pillar. Reads the recurring-revenue assumptions behind the price.
• Willingness to Pay Validation — the framework article. What willingness to pay is and why interest is not payment.
• Startup Validation Checklist — the validation pipeline the pricing decision sits inside.
• The Pricing Validation cluster — the canonical starting point for the pricing sub-cluster (jump to "Testing Monetization").

Sources

• Stripe Atlas, "Pricing strategy: How to choose the right pricing strategy" — https://stripe.com/resources/more/pricing-strategy-guide
• Stripe Atlas, "SaaS pricing strategy: How to price your SaaS product" — https://stripe.com/en-nl/resources/more/saas-pricing-strategy
• OpenView Partners, "2023 SaaS Benchmarks Report" (with Paddle) — publicly distributed PDF covering pricing update cadence, ARPU growth by revision interval, and pricing-time-per-year benchmarks across 3,500+ SaaS respondents
• OpenView Partners, "2023 Product Benchmarks" — https://openviewpartners.com/2023-product-benchmarks
• Peter Van Westendorp, "NSS — Price Sensitivity Meter" (1976 ESOMAR paper) — the original Price Sensitivity Meter methodology
• Patrick Campbell, "Face it, you guessed. That's not okay." — https://www.profitwell.com/subscription-pricing-strategy — the published Price Intelligently methodology, including the four-step framework and the Van Westendorp integration
• Patrick Campbell, "Pricing: everything you wanted to know but were afraid to ask" — the Acquired podcast episode covering the Wistia and Atlassian case studies
• Lenny's Newsletter — Lenny Rachitsky's published essays on SaaS pricing and value-metric selection

Next action

Pick the value metric first. Five calls with target buyers, asking them to walk through the last time they had the problem your product solves, and to describe what they would pay in their own words. The unit they reach for is the value metric your product should be priced against. Everything else follows from that decision.

To see how the pricing shape you have in mind affects the rest of the validation, run Yibud's startup analysis and read the monetization dimension alongside the distribution dimension — the engine flags pricing × channel combinations that are structurally weak, regardless of the absolute price.