SaaS pricing strategy: how to price your SaaS product

Most SaaS pricing is set early, by guesswork, and then defended for years. A founder picks a number that feels reasonable, a competitor’s page nudges it up or down, and a few discounts later nobody remembers why the price is what it is.

A pricing strategy replaces the guesswork with a method. Most guides on this subject list three approaches (cost-plus, competitor-based and value-based), tell you to choose value-based and stop. They do not show you how to find out what customers will pay, how to turn a value estimate into a price, how to test a price, or how to raise it on existing customers without a revolt. This guide does all four, with worked examples and the arithmetic shown.

Every rule or statistic here names where it comes from. Where the usual advice online has no traceable source, we say so. If you also need to choose how customers are charged, see our guide to SaaS pricing models.

What is a SaaS pricing strategy?

A SaaS pricing strategy is the approach you use to decide how much to charge and for what. It sets the price points, the packages and the rules for changing them. It is different from a pricing model, which is the structure of the charge: per user, tiered, usage-based and so on.

Two decisions sit inside it. Pricing is how much you charge. Packaging is what is included at each price. They are separate levers and they interact: you can change the effective price of a product by changing what it includes without changing the number on the page.

Three ways to set a price, and why only one is a strategy

Cost-plus pricing starts from what the product costs to build and serve, then adds a margin. It is easy to calculate and it protects you from selling at a loss. Its weakness is that customers do not care what your costs are. They care what the product is worth to them, so cost-plus either leaves money on the table or prices you out. Be wary of formulas that fold customer acquisition cost into the unit price: they make price depend on how well your marketing is working, which has nothing to do with value.

Competitor-based pricing sets your price relative to alternatives in the market. It is a useful check, because buyers will compare. As a method, it assumes that your competitors have priced well, that their customers are your customers and that their product is equal to yours. Often none of those is true.

Value-based pricing starts from what the product is worth to the customer and prices against that. It is harder, because you have to understand your buyers, but it is the only approach that lets price grow with the value you deliver. Use value as the anchor, and use cost and competitors as sanity checks either side of it: cost sets the floor, and the alternatives your buyer really has set a ceiling.

The six steps

1. Decide what you are really selling

2. Choose a value metric

3. Find out what customers will pay

4. Turn value into a price

5. Package it

6. Test it and review it

Step 1: Decide what you are really selling

Value is set by the buyer’s problem. The more it hurts and the more urgent it is, the more they will pay to make it go away. A tool that removes a daily, costly frustration can be priced very differently from a tool that is merely nice to have, even when the two are equally hard to build.

Write down the problem in the buyer’s words, who feels it most, what it costs them today, and what they do about it now. That last point matters most. Your price is always compared with the alternative, and the alternative is often a spreadsheet, a person, an in-house build or doing nothing.

Step 2: Choose a value metric

The value metric is the unit your price is attached to: seats, usage, records, locations, outcomes. Stripe’s pricing guidance gives four tests of a strong one. It should flex with the value the customer gets, be understandable without explanation, resist being gamed, and match how customers budget.

Stripe names the tests but offers no way to score them, so here is a scorecard. Rate each candidate metric from 1 (weak) to 5 (strong) against each test and two more that matter in practice. The scores below are an illustration of the method, not data.

Score your own candidates with customers in the room. A metric that scores high on value and low on predictability (usage, in this illustration) can often be fixed by adding structure: a base fee, an allowance or a cap.

Step 3: Find out what customers will pay

The most reliable way to find out is to ask, in a structured way. Three methods cover most cases. None is perfect, and each has limits you should know before you rely on it.

Van Westendorp’s price sensitivity meter

Introduced in 1976, it asks four questions of people who fit your target customer:

1. At what price would this be so cheap that you would doubt its quality?

2. At what price would it be a bargain?

3. At what price would it start to feel expensive, though you would still consider it?

4. At what price would it be too expensive to consider?

You plot the share of respondents saying each price fits each question. Where the curves cross gives a range of acceptable prices, an indifference point and an optimal point. It gives you a range, not demand. It cannot tell you how many people would buy at each price, and it assumes that price signals value, which is contestable.

A worked example with invented data. Suppose 100 respondents answer. The table shows the share saying each price is too cheap, a bargain, expensive and too expensive:

Finding where the lines cross:

• Optimal price point: where “too cheap” equals “too expensive”. Both are 10% at $30, so $30.

• Indifference point: where “bargain” equals “expensive”. Between $30 (35% against 30%) and $40 (15% against 58%), which interpolates to about $31.

• Point of marginal cheapness: where “too cheap” crosses “expensive”. Between $20 and $30, about $25.

• Point of marginal expensiveness: where “bargain” crosses “too expensive”. Between $30 and $40, about $36.

The acceptable range in this example is therefore about $25 to $36, with $30 at its centre. That is a range to test, not a verdict.

Gabor-Granger

This method shows each respondent one price at a time, in ascending order, and asks whether they would buy. SurveyMonkey’s description sets out the design: five evenly spaced prices, one identical description of the product for everyone and at least 100 completed responses for each segment you plan to report separately. You multiply the share who would buy at each price by that price to get a revenue index, and the peak is the optimum.

It has real limits. It can inflate results if the product is oversold in the description, it works poorly for new kinds of product, and it cannot compare you with competitors.

A worked example with invented data. Suppose the share who would buy at five prices is 70%, 62%, 50%, 36% and 22%:

The index peaks at $25, with $30 close behind. Note how close the top two are: a difference of 50 on a base of 1,500 is well inside the noise of a survey.

Conjoint analysis

Conjoint breaks a product into attributes and levels, such as features, limits and price, and infers how much respondents value each level from the choices they make. Price can be one attribute. It is powerful for packaging decisions. It is also complex to design, prone to measurement error, and respondents tend to simplify when there are many features. Bain’s published account of a conjoint study for a UK accounting software provider described modelling that pointed toward a tiered offer and projected roughly 3.5 times the users and almost double the revenue. Those were projections made in 2014, not measured outcomes, so read them as an illustration of the method.

Which method to use

All survey methods share one weakness: people are not paying real money. Treat results as evidence about a range, then confirm with real purchase behaviour.

Step 4: Turn value into a price

Once you have a view of value, price is a choice about how much of it to capture. Here is the full arithmetic for one fictional product. Every input is an assumption, labelled as such.

The product: a tool that helps support agents resolve tickets faster.

The assumptions:

• Each agent saves 4 hours a week.

• Agents work 46 weeks a year.

• An agent’s fully loaded cost is $40 an hour.

• It costs $18 per agent per month to serve the customer (hosting, support, payment fees).

Value per agent. 4 hours x 46 weeks x $40 is $7,360 a year, or about $613 a month.

Choose a share. Suppose you decide to capture 15% of that value. That is about $92 per agent per month. You could choose a different share. The point is to make the choice deliberately and to check it against the alternatives: if hiring an extra agent would cost far more, 15% looks modest, but if a free tool does most of the job, it does not.

Check the margin. $92 price against $18 cost is a gross margin of about 80%.

Scale it across customer sizes:

A flat $92 per agent is the simplest result. In practice you might lower the per-agent price for larger customers to reflect cost to serve and competition, and you would pair it with packaging. That leads to the next step.

Step 5: Package it

Decide what goes in each plan and how many plans there are. Fewer is usually better, because every plan is another thing to explain, support and bill. Segment customers first: different customers get different value and cost you different amounts to serve. Identify who gets the most value, who would pay the most and who costs the most to support, and make the packages reflect those groups.

For how to design package tiers, and a full account of the three different things "tiered" can mean, read Tiered pricing explained.

Step 6: Test it and review it

What to test, and what not to

• Test on new visitors and new sign-ups. Do not test prices on existing customers. Showing the same customer different prices erodes trust.

• Do not show different prices to the same person. Assign visitors to a version consistently.

• Run each test for at least one full buying cycle, so that you capture the real decision rather than first reactions.

• Decide your success metric first. Conversion alone can mislead: a lower price may convert more and earn less. Look at revenue per visitor, early churn and support volume.

• Check fairness and legal risk in the markets you sell to before you vary prices.

We did not find a verified rule for the minimum sample size of a live price test, because it depends on your traffic and the size of difference you want to detect. The 100 completed responses per segment from the survey guidance applies to surveys, not to live tests.

A quarterly pricing review checklist

Treat the price as something you review, not something you set. Each quarter:

• Look at win rate and the reasons for lost deals.

• Look at discount rates: are discounts becoming the real price?

• Look at churn and downgrades by plan.

• Look at which plans and limits customers hit.

• Re-run the value calculation if the product has changed.

• Check cost to serve per segment.

• Review competitor changes, as a sanity check, not as an instruction.

• Confirm that your billing can handle every plan, change and market you sell in.

How to raise prices without losing customers

Raising prices on existing customers is the part most guides skip. A playbook:

1. Segment customers by plan, tenure, size and how much value they get. A price rise lands differently on a customer who uses the product heavily than on one who barely logs in.

2. Decide whether to grandfather. Keeping existing customers on old prices for a fixed period is gentler. Indefinite grandfathering means supporting two price lists for good.

3. Choose the timing. Moving customers at renewal is cleanest: it is a natural decision point and avoids changing a price mid-term.

4. Give notice. How much notice you owe depends on your contracts and local consumer law, so check both. We did not find an authoritative source for a universal rule, so we do not state one. Marketplaces that host software set their own minimum notice periods for price changes.

5. Communicate the reason and the value, and show each customer exactly what changes for them.

6. Prepare a rollback plan, and watch churn, downgrades and support tickets for the first cycle.

What your billing system does when you change a price

A price decision is only as good as your ability to carry it out. Here is what happens in a billing system, verified against Stripe’s documentation, the only billing platform whose documentation we opened. Other systems behave differently, so check yours.

• Moving a customer to a new price mid-cycle creates proration items: a credit for unused time and a charge for the new plan. Stripe’s example: upgrading from $10 to $20 halfway through a month bills an extra $5, made up of a $5 credit and a $10 charge. Proration is calculated to the second by default.

• Usage-based charges are not prorated, which affects how a hybrid plan behaves on a mid-cycle change.

• You can switch prorations off, but then the customer pays the full new price at the next invoice.

• Downgrade credits can behave unexpectedly. In Stripe’s example, one billing mode credits time the customer never paid for, while another credits at the last price billed, so the result can differ by several dollars.

• Changing a price without specifying the existing item can add a second item, so the customer is billed for both. This is a classic migration mistake.

• Switching from monthly to yearly moves the billing date to the switch date. If the immediate payment fails, the change can still go through and leave the subscription past due.

• Scheduled changes at the end of a period are the documented way to migrate at renewal, and every phase must carry all discounts and settings.

• Previews help. Invoice previews let you check what a change will charge before you make it.

What we could not verify, and so do not claim: how multi-currency pricing and tax behave across billing systems.

OpenByte handles every billing model, currency and local tax rule out of the box. You set up a new plan or a new market once, and it works everywhere. It works out part-month charges, pauses, credits and refunds automatically, and checks every charge against the customer’s plan, history and payments before it goes out. See the Billing page or read about usage-based billing if your pricing follows consumption.

Why pricing strategies fail: mechanisms and warning signs

• A metric that punishes the behaviour you want. Per-seat pricing can discourage inviting colleagues. Warning sign: customers hold seat counts flat while usage climbs.

• A gap between tiers so large that nobody upgrades. Warning sign: customers stay on the lower tier and hit its limits.

• A free tier that carries real cost. Warning sign: support tickets and infrastructure cost from free users rise faster than conversions.

• Discounts that become the price. Warning sign: most deals close below list.

• Underpricing out of fear. A price that is too low signals low value and leaves no room to invest. Warning sign: you win nearly every deal without a fight.

• Setting the price once. Warning sign: the price list is older than the product’s last three releases.

• A price you cannot bill accurately. Warning sign: duplicate charges, wrong proration and a steady stream of “why was I charged?” emails.

SaaS pricing best practices

• Price on value, not on cost. Cost sets the floor, not the price.

• Choose the metric with customers, and score it against the tests above.

• Keep the number of plans small.

• Make the price easy to understand. If a buyer cannot work out what they will pay, they will assume the worst.

• Discount deliberately. A discount you give casually becomes the price.

• Review at least quarterly, as the checklist above sets out.

• Plan for change. Decide how existing customers move to new prices before you announce them.

• Check that billing can cope. A price you cannot bill accurately is not a usable price.

Statistics we did not use

Pricing advice is full of statistics, and many have no traceable source. We left out several that appear widely:

• Claims that companies that regularly refine pricing grow a fixed percentage faster. We found the claim repeated, attributed to another article, and traced to no primary source.

• Benchmarks for the share of free users who convert, and typical churn rates by segment, quoted without a source.

• The assertion that customers “usually pick the middle tier.”

• Adoption percentages for pricing models quoted from reports we could not locate.

If a statistic comes from a company that sells the thing it measures, read the methodology before you repeat it. If there is no methodology, do not repeat it.

Frequently asked questions

What is the best SaaS pricing strategy?

Value-based pricing, anchored on what the product is worth to the customer, with cost as the floor and the buyer’s real alternatives as a check. Cost-plus and competitor-based pricing are checks, not strategies.

How do I price a SaaS product with no customers?

Start with the value calculation in step 4, then use a survey method such as Van Westendorp to find a range, and plan to adjust with real purchase data as soon as you have it.

How do I find out what customers will pay?

Use a structured method: Van Westendorp for a range, Gabor-Granger to find the revenue-maximising point on a price ladder, conjoint for packaging trade-offs, and live tests once you have traffic. Treat survey results as evidence, not proof.

How many customers do I need for a pricing survey?

SurveyMonkey’s guidance for Gabor-Granger is at least 100 completed responses for each segment you will report separately. Live price tests depend on your traffic and the size of effect you need to detect.

How do I raise prices on existing customers?

Segment customers, decide whether and for how long to grandfather, move customers at renewal where you can, give notice that meets your contracts and local law, explain the value, and watch churn for the first cycle.

How often should I review my pricing?

At least quarterly, with the checklist above. Review immediately when the product changes materially or discounts start to become the norm.

Should I offer annual pricing?

Annual billing changes cash flow and when customers decide whether to stay, so it is a billing-model decision as well as a pricing one. See SaaS revenue models and billing models.

What happens in billing when a customer changes plan mid-cycle?

For charges billed in advance, the system credits the unused part of the old plan and charges for the new one. Usage-based charges are not prorated in Stripe’s documentation. Test the cases you care about in your own system.

Key takeaways

• Cost-plus and competitor pricing are starting points. Value-based pricing is the goal, but you only get there by finding out what customers are actually willing to pay.

• Van Westendorp and Gabor-Granger surveys give you a range and a demand curve. Treat the output as evidence to test, not a final answer.

• Your value metric is the heart of the strategy. It should rise as the customer gets more value and be easy for them to understand.

• Test a price before you roll it out, and raise prices on existing customers with notice, a reason and a fair path for those who push back.

• The figures in our survey examples are invented, so use your own customer data.

Nobody gets pricing perfect on the first attempt, and the companies that do well are mostly the ones that keep looking at it. Start small: one question to ten customers this week will teach you more than another month of debate.