How to Price a SaaS Product Before Launch

SaaS pricing before launch: why price before you ship

Most founders look up how to price a SaaS product once the product is built, and treat price as the last field on a pricing page. SaaS pricing before launch is closer to the first decision. Jason Cohen, who founded Smart Bear and WP Engine, made the case in Pricing determines your business model (2014): at $10 a month you need over 8,000 customers to reach $1 million a year, and those customers still expect support, while at $100 a month 200 to 300 customers can make a very profitable small company. The price decides who you sell to, how you reach them and whether you can afford to.

Pricing is also far cheaper to get right before anyone pays. Madhavan Ramanujam of Simon-Kucher told First Round Review that about 80% of companies in a 2014 Simon-Kucher study left the price discussion to the last minute, and he argues for asking whether people would pay before you build, while they are not yet in a negotiating mood. Before launch, changing your price means editing a page. After launch it means emails, exceptions and churn.

And before launch you have something most pricing guides assume you lack: a list of people who asked for the product. If you have started a pre-launch waitlist, every person on it is a potential respondent who matches your market better than any panel you could buy. The rest of this guide uses that list.

How to price a SaaS product in 7 steps

The short version of how to price a SaaS product with real input, in roughly the order you would do it over four weeks:

1. Work out the three inputs: value to the customer, the alternatives they pay for now, and your costs at a target margin.
2. Choose a value metric: the unit your price scales with.
3. Decide how many tiers to launch with: usually one or two.
4. Run 5 to 10 willingness-to-pay interviews that anchor on what people spend today.
5. Run the Van Westendorp price sensitivity meter on 30 to 50 people from your list.
6. Check one or two candidate prices with the Gabor-Granger method.
7. Set the billing terms: monthly or annual, free plan, free trial or neither.

Then test the price with real money before you announce it.

Step 1: Work out the three inputs

Every pricing method you will read about is a way of estimating one of three numbers. "How much should I charge for my SaaS?" has no single answer, but these three numbers bound it. Get rough versions of all three before you ask anyone anything.

Value to the customer. What changes for the customer, in money or time, when the product works? Count only what they would agree with: hours saved at what that person costs, revenue they recover, a tool or contractor they stop paying for. This is the ceiling. Nobody pays more than the value they believe they get, and most will want to keep the larger share of it.

The alternatives they pay for now. Very few products replace nothing. Your customers solve the problem today with a competitor, a spreadsheet, a part-time assistant, an agency or their own evenings. What they pay for that is the anchor your price will be judged against, whether or not you mention it. Jason Lemkin's advice in SaaStr's take on early pricing is to price close to comparable products before product-market fit, because buyers already know roughly what software like yours costs.

Your costs and target margin. Add up what one more customer costs you each month: hosting, third-party APIs, email sending, payment fees, and the support time they will need. Then divide by one minus your target gross margin. Software has traditionally run at high gross margins: Martin Casado and Matt Bornstein of a16z put the benchmark for comparable SaaS at 60% to 80% or more, and found AI companies often at 50% to 60%, because model calls cost money every time someone uses the product. If you are building with AI, measure your cost per customer from real usage before you set a floor. This number is the floor: below it, every new customer makes the business worse.

A worked example

Take a made-up product: a tool that chases unpaid invoices for agencies of 5 to 20 people. All numbers below are illustrative.

So the price lives somewhere between $35 and $180, and the alternatives say most buyers will judge it against $20 to $100. A first hypothesis of $49 a month sits above the floor, inside the range of what they pay now, and at about a quarter of the value. It is a hypothesis, not an answer. The next steps test it.

A tip for the cost line: if you do not know your monthly costs yet, a startup runway calculator forces you to list them, and the same list gives you the cost per customer.

Step 2: Choose a value metric

A value metric is the unit your price scales with. It matters as much as the number, because it decides who pays more, when bills go up and how predictable they are. Patrick Campbell, the founder of ProfitWell, wrote in a guest post for Lenny's Newsletter that companies pricing on a value metric grew at double the rate with half the churn of those charging a flat fee or by feature tier. That is ProfitWell's own research, so read it as a vendor claim, but the logic holds: when price grows with the value a customer gets, small customers can start cheaply and big ones pay more without a renegotiation.

DHH, co-founder of 37signals, explained why Basecamp avoided per-seat pricing in Why we never sold Basecamp by the seat (2017): per-seat pricing "by definition makes your biggest customers your best customers", and with that money comes influence over the roadmap. Basecamp's pricing page still lists flat plans with no per-user fees (checked 31 October 2026). Slack went the other way and charges per member, but its Fair Billing Policy treats anyone who has not used Slack in over 28 days as inactive and credits the unused part of the bill, which removes the main objection to per-seat pricing: paying for people who never log in. Zapier charges by tasks, counting only work it successfully completes.

Three tests for your value metric:

• Does it grow with value? A customer using more of it should be getting more out of the product.
• Can the customer predict it? If they cannot guess next month's bill, they will hesitate to start.
• Can you measure it cleanly? If you argue with customers about what counted, it is the wrong unit.

Campbell's suggestion in the same post: write down 5 to 10 candidate metrics, then ask customers and prospects which one or two feel fair. You can put that question in your interviews and your survey (it is question 4 in the survey below).

Step 3: Decide how many tiers to launch with

Tiers exist to serve different segments: groups that need different things and will pay different amounts. Before launch you usually know one segment well, the one you have been talking to. Inventing three tiers for segments you have not met means guessing at the fences between them, and every guess is a pricing page decision a customer can be confused by.

Two plans is fine at launch when the evidence supports it, and the Superhuman story below shows what that evidence looks like: different buyer types giving clearly different answers to the same price questions. Without that evidence, launch with one plan and a clear way to contact you for anything bigger. Adding a plan later is easy. Removing one means moving customers.

Step 4: Run 5 to 10 willingness-to-pay interviews

Before you send a survey, talk to 5 to 10 people from your list who match the customer you are building for. The interviews do two jobs: they tell you how people describe the value in their own words, which you need to write the survey, and they surface the alternatives and budgets you guessed at in step 1.

Do not ask "how much would you pay for this?", because hypothetical answers run high: a meta-analysis of 28 studies by Murphy, Allen, Stevens and Weatherhead (2005) found stated values a median 1.35 times what people actually paid. Ask about the past and the present instead, and save price ranges for last.

Willingness to pay questions that work

Anchor on what they do and spend today:

• "How do you handle [the problem] today? Walk me through the last time."
• "What do you pay for that now, in money? Roughly how many hours a month does it take?"
• "Who decided to pay for it, and how long did that take?"
• "What else did you try before this, and why did you stop?"
• "If [the problem] disappeared tomorrow, what would you do with the time or money?"
• "What would a tool need to do for you to switch from what you use now?"

Then, near the end, Ramanujam's three range questions from the First Round interview: "What would be an acceptable price for this? What would be expensive? What would be prohibitively expensive?" Ranges are easier to answer honestly than a single number, and the gaps between them show you where demand drops off.

Two follow-ups make every answer more useful. When someone says a price is fine, ask "why would it be worth that to you?", and listen for the value they repeat back. When someone says it is too much, ask "compared to what?", and you learn the anchor they are using. Write down the exact words people use for the value, because they become the wording of your survey and later your pricing page.

Step 5: Run the Van Westendorp price sensitivity meter on your list

The Van Westendorp price sensitivity meter is a four-question survey introduced by the Dutch economist Peter van Westendorp at the 1976 ESOMAR Congress (Wikipedia summary). Instead of asking for one price, it asks for four, each a different threshold, and the overlap of everyone's answers shows the range of prices your market finds reasonable. It works well before launch because it needs nothing but a description of the product and a list of people who understand the problem.

The four questions, and what each tells you

Ask all four as open number fields, in a fixed order, and do not suggest a range: Penn State's course notes on the method warn that offering ranges imposes your view on the answers. Describe the product in two or three sentences before the questions, using the words your interviewees used, and state the billing period ("per month, for your whole team") so everyone prices the same thing.

How many answers you need

Be honest with yourself about what a pre-launch survey can tell you. Thirty or more answers from your list give you a rough, directional range, not a reliable reading. For that, researchers recommend much larger samples, often in the hundreds: Mike Pritchard of 5 Circles Research, in the comments on his guide to the method, treats 10 to 20 answers as directional, 50 as a minimum and 100 as better, and uses at least 200 answers for business studies and 400 for consumer studies, which he puts at a margin of error of about 7 and 5 points. For a pre-launch list, aim for 50 answers from people who match your customer, accept 30 as the floor, and treat the result as a range to test with real payments. If you want to compare segments (founders against employees, small teams against large), you need that number in each one.

That also tells you when to run it. If you have 300 people on your list and a typical survey response rate, you will get your 30 to 50. If you have 40, run the interviews and wait; getting your first 100 waitlist signups comes first.

How to read the Van Westendorp chart

For each price, work out four shares of your respondents: the share who said that price or more would be too cheap, the share who said that price or more would be a bargain, the share who said that price or less would be getting expensive, and the share who said that price or less would be too expensive. The first two lines fall as the price rises and the last two climb. Where they cross gives four points:

• Lower edge of the acceptable range: where "too cheap" crosses "expensive". Below this, more people distrust the price than hesitate at it.
• Optimal price point: where "too cheap" crosses "too expensive". The fewest people reject the price for either reason.
• Indifference price point: where "cheap" crosses "expensive". As many people call it a bargain as call it expensive.
• Upper edge of the acceptable range: where "cheap" crosses "too expensive". Above this, more people rule it out than see value.

Different tools invert different curves, so their exact points can shift a little. Do not let that worry you: the output is a range, and the decision is yours.

A worked example with illustrative numbers

Back to the invoice tool. Suppose 40 people from the waitlist who run or work at agencies answer the four questions. These numbers are illustrative, made up to show how the chart reads:

• Acceptable range: $30 to $60 a month.
• Optimal price point: $35. Indifference price point: about $48.
• At $50, 7 of the 40 called the price too expensive. At $60, 10 did.

How to read it against step 1: the lower part of the range ($30 to $35) falls below the $35 cost floor, so it is out regardless of what people would accept. The $49 hypothesis sits at the indifference point and well inside the acceptable range. Going to $59 would put the price near the upper edge, where a quarter of respondents rule it out. A reasonable decision: launch at $49, and watch whether the people who do not buy are the ones who answered low.

Two limits to keep in mind. The method shows which prices people find acceptable, not how many will buy; Sawtooth Software, which makes pricing research tools, prefers choice-based conjoint for measuring price sensitivity in depth. And a new category is hard to price for people who have never paid for anything like it, so lean harder on the alternatives from step 1 when your product is truly new.

Split the answers by segment

The most useful thing a Van Westendorp survey can do before launch is show you that different people answer differently. Gaurav Vohra, part of Superhuman's founding team, told Schematic that Superhuman's 2017 Van Westendorp data split clearly by buyer: individual contributors around $15 to $20 a month, founders and CEOs around $30, and investors around $45 and up. The team anchored on founders and CEOs, set the price at $29, and an intern rounded it to $30. Accepting that some individual contributors would not buy at that price was part of the decision.

That is why the survey below asks about role and company size before any price question. A single blended range would have hidden Superhuman's answer.

Step 6: Check a price with the Gabor-Granger method

Van Westendorp gives you a range. The Gabor-Granger method, named after the economists André Gabor and Clive Granger, tests specific prices: respondents see one price and say whether they would buy at it, then see a higher price if they said yes or a lower one if they said no. Sawtooth's explainer describes the output: a demand curve (the share willing to buy at each price) and a revenue curve (that share times the price), whose peak suggests the price to pick.

You can run a light version as one question in your survey: show one of three or four prices, chosen at random per respondent, and ask how likely they are to buy at it. With the invoice example (illustrative numbers again):

Revenue peaks at $49 and falls with every step above it, which agrees with the Van Westendorp read. Sawtooth lists the method's limits plainly: it tests price alone, and it does not show respondents the alternatives they would compare you against. Stated purchase intent also runs high, so use the shape of the curve, not the percentages, and never forecast revenue from "would buy" answers. For a revenue forecast, plug a cautious paying share into a pre-launch MRR calculator and try two or three prices side by side.

Step 7: Set the billing terms

Price is more than the monthly number. Three decisions shape what people actually pay.

Monthly vs annual billing

Offer both once you have a monthly price you trust. Annual billing gives you cash up front and fewer cancellation decisions each year, and the discount is the price of that. The common choice is two months free: in Recurly's study of plan structures, covering more than 1,000 SaaS businesses billing through Recurly from January 2017 to July 2018, the most popular annual discount was 16.7%, which is exactly two months free, and the median across all price segments was the same. Recurly is a billing vendor and the data is from its own customers, so read it as a vendor benchmark. Campbell's post adds two findings from ProfitWell's data: whole-dollar offers such as "2 months free" work better than percentages, and the size of a discount correlates almost perfectly with churn, so he advises against discounts above about 20%.

Before launch, keep it simple: one monthly price, one annual price at 10 times the monthly, and no other discounts.

Free plan, free trial or neither

For a sense of scale, the benchmark Kyle Poyar and Lenny Rachitsky published in What is a good free-to-paid conversion rate (2023), from a survey of 1,000+ mostly B2B products, put good conversion at 3% to 5% for freemium self-serve and 8% to 12% for free trials. A free plan needs many more users to produce the same number of customers, which is a cost before launch, when your list is small. If price is what you need to learn, paid from day one, or a trial that ends in a payment, gives you the answer fastest. For B2B buyers in particular, many founders skip both and charge from the first customer, with a demo or a hands-on setup instead of a trial.

How to show the price on your pricing page

Once you have a price, how you present it shapes how it reads. Three things matter most at launch:

• An anchor price. Buyers judge a price against whatever number sits next to it. If you have a second, larger plan or a "talk to us" option, show it beside the main plan, and put the comparison from step 1 in a line of copy ("less than 4 hours of an account manager's time a month"), so $49 is read against $180 of value rather than against zero.
• A recommended plan. With two plans, mark the one most customers should pick, with a label such as "Most teams choose this" and a filled button. It answers "which one is for me?" before the buyer has to ask.
• An annual toggle. Show monthly and annual prices with one switch, and state the annual saving in whole months ("2 months free") rather than a percentage, which Campbell's data favored. Keep the monthly price visible so the annual one has something to be compared with.

Then test the price with real money

Everything above is people telling you what they would do; the test that matters is a payment. Before you announce, ask a small group from your list to pay, for example a first cohort that prepays or a paid pilot, and if people who called $49 reasonable will not pay it when the link is in front of them, believe the payment over the survey. To reward those early buyers, the early access program guide covers founding member pricing, and the guide on how to convert a waitlist into paying customers covers the launch offer.

Raising prices later, and who to grandfather

Your first price will be wrong in one direction or the other, usually too low. Raising it later is normal. Three choices shape how it goes:

• Who gets the new price, and when. New customers can see the new price immediately. For existing customers, Patrick Campbell recommends what he calls a legacy discount: they keep their current price for a set period, then move (Verblio's summary of his podcast interview, 2021). Keeping old prices forever is the other option, and it gets harder to manage with every change.
• How much notice. Superhuman, raising its price from $30 to $40 in 2024, gave three months' notice by email (an August email for a November change), with options: move to $40, move to a separate $30 plan, or switch to annual at a limited-time rate. Gaurav Vohra told Schematic that churn ran 1 to 2 points above normal for 60 to 90 days, and that the change was the largest year-over-year revenue impact that year.
• What you say. Campbell's advice for the email: remind customers of the value they got, point to features they asked for that now exist, explain that the price funds more of that work, and end with an offer to talk if the change genuinely hurts their business.

Whatever you decide, write it down before your first customer pays, so "early customers keep their price for 12 months" is a policy and not a promise you have to reconstruct later.

Common SaaS pricing mistakes before launch

1. Pricing too low. The most common one. A low price needs more customers, more support per dollar and more acquisition spend to reach the same revenue, as Cohen's price zones show, and it tells buyers something about quality (that is what the "too cheap" question measures). Multiply your price by the customers you can realistically reach, with a TAM SAM SOM calculator if it helps, and check the business still works.
2. Pricing by cost alone. Cost is the floor, not the price. A product that costs $7 a month to serve and saves someone $180 is not worth $9.
3. Copying competitors without checking. Comparable prices are a fair anchor, as Lemkin says. But a competitor's price reflects its costs, segment and mistakes. Check that your buyers compare you with that competitor before you match it.
4. Too many tiers. Three or four plans for segments you have not met means guessing at the fences, and buyers stall when they cannot tell which plan is theirs.
5. Asking friends "how much would you pay?" Friends are polite and rarely the buyer, and the question produces a number nobody had to commit to. Ask people on your list about current spend.
6. Treating a survey as the answer. Survey results are a range to test. The payment is the answer.
7. Discounting to close the first deals. Deep early discounts set the wrong anchor, and Campbell links larger discounts to higher churn. Offer a first-cohort price with written terms instead.

If people join your list and then do not pay, the problem may be earlier than price; the guide to why your waitlist is not converting walks through the other causes.

A pricing survey you can copy

Use this as one short survey sent to your list, after the interviews. Replace everything in square brackets. It takes about four minutes to answer. Questions 1 and 2 let you split results by segment; questions 6 to 9 are the Van Westendorp questions; question 10 is the light Gabor-Granger check (show each person one of the prices at random).

Help us set a fair price for [product]

[Product] [one sentence on what it does, in your interviewees' words].
For example: [one concrete outcome, e.g. "it chases unpaid invoices
so your team does not have to"]. All prices below are per month,
for [your whole team / one person].

1. Which best describes you?
   [Founder or owner] [Manager] [Individual contributor] [Other]

2. How many people work at your company?
   [Just me] [2 to 10] [11 to 50] [51 to 200] [More than 200]

3. How do you handle [the problem] today?
   [Tool or competitor] [Spreadsheet] [Someone does it by hand]
   [We do not handle it] [Other]

4. Roughly what do you pay for that each month, including people's time?
   [Number]

5. Which way of paying would feel fairest to you?
   [A flat monthly price] [Per person on the team]
   [By how much I use it, e.g. per [unit]] [Not sure]

6. At what monthly price would [product] be so cheap that you would
   doubt its quality?   [Number]

7. At what monthly price would [product] be a bargain, a great buy
   for the money?   [Number]

8. At what monthly price would [product] start to feel expensive,
   so that you would have to think before buying?   [Number]

9. At what monthly price would [product] be so expensive that you
   would not consider it?   [Number]

10. If [product] cost [$X] a month, how likely would you be to buy it
    when it launches?
    [Definitely] [Probably] [Not sure] [Probably not] [Definitely not]

11. Anything that would make you more or less likely to pay for it?
    [Long text]

12. Can we contact you about a short call or early access?
    [Yes] [No]

When you read the answers, drop anyone whose four Van Westendorp answers are out of order (a "too cheap" price higher than their "too expensive" one) before you build the chart; the Penn State notes recommend cleaning invalid answers first. Then build the chart once for everyone and once for each segment from questions 1 and 2.

The email that invites your list to answer

Keep it short, from a person, and honest about why you are asking. Send it to people who match the customer first, and expect most answers in the first 48 hours. For more on what gets opened, see these waitlist email templates.

Subject: Can you help me price [product]? (4 minutes)

Hi [first name],

You joined the waitlist for [product] [time ago, e.g. "in September"].
Thank you. We are getting close to launch, and before we set a price
I would rather ask the people who will use it than guess.

Would you answer 12 quick questions? It takes about 4 minutes:
[survey link]

There are no right answers. Honest numbers help most, including
"I would not pay for this," which tells us something important.

As a thank you, everyone who answers by [date] will be in the
first group we invite. [Only promise this if you will keep it.]

Thanks,
[Your name]
[Your role] at [product]

P.S. If you would rather talk for 15 minutes, just reply to this
email and I will send a time.

Collecting pricing answers with LaunchList

LaunchList is pre-launch waitlist software for founders. It will not set your price, analyze survey results or draw a Van Westendorp chart, and it does not take payments, so the pre-sale step happens in your own billing tool. What it does is help you collect pricing answers from the right people and reach them again.

• Ask at signup. The waitlist form builder adds questions to your signup form on every plan: dropdowns, single or multiple choice, number fields and long text. One question such as "What do you pay for [tool] today?" or "How many people are on your team?" at signup means every new person arrives with the context you need to segment later. Keep it to one or two questions, because every extra field costs signups.
• Run the survey as a second form. A project can have several forms, including a survey form, and they all feed the same audience: someone who signs up and later answers your pricing survey is one contact with two submissions. You read a survey's answers under Forms, and you can export your audience as a CSV to build the chart in a spreadsheet.
• Split the answers. Tags work on every plan, so you can tag interviewees or anyone who answered a certain role. On the Launch and Grow plans you can build your own segments, for example agencies of 11 to 50 people, and see them in waitlist management.
• Send the survey to your list. From the Collect plan you can email your waitlist with the invitation above, to everyone or to a tag or segment. Pricing has what each plan includes.
• Model revenue. The free pre-launch MRR calculator linked in step 6 turns a list size, a paying share and a price into launch-day and first-year revenue, which is the quickest way to compare $39 against $49.

If you are still deciding whether the idea is worth pricing at all, start with an idea validation landing page and the idea validation tools guide, and come back to pricing once people are signing up. If you are building a subscription product, SaaS waitlist software explains how founders use LaunchList before a SaaS launch. For the rest of launch preparation, the SaaS launch checklist and the SaaS pre-launch marketing playbook cover what comes next.