SaaS pricing is two decisions, not one: the model, meaning the unit you charge on, then the level, meaning the amount. Pick a metric that tracks the value the customer receives, set a price from that value and the market, test it on real deals, then revise it once a year instead of leaving it frozen for three.
Setting the price of a SaaS (software sold by subscription) is the decision that weighs most on growth, and it is often the one that gets the least method. Many software companies set a price by glancing at a competitor, rounding it off, and never touching it again for three years. We find this pattern in almost every diagnosis: a price that reflects neither the value delivered nor how the product is actually used. It is not a billing detail, it is a lever on margin and positioning, and it sits at the heart of our work on pricing and packaging.
SaaS pricing is two decisions you should keep apart. The first is the model: what you charge on. The second is the level: how much. This article covers both. Splitting the offer into tiers is a separate subject, which we cover in our article on designing your pricing tiers.
The model before the amount
The first mistake is to argue about the amount before settling the model. As long as you do not know the unit you charge on, no price makes sense. The model decides how your revenue tracks the customer’s success, and therefore whether it will grow with them or against them.
Finding the value metric
The value metric is the unit that goes up when the customer gets more out of your product. It is the core of good SaaS pricing. A support tool can charge per agent or per ticket handled, a data platform per volume analysed, a sales tool per active rep.
A good metric has three qualities. It is simple to understand, or every negotiation turns into a technical debate. It is hard to game, or the customer optimises their usage to pay less and your revenue stalls. And above all it tracks the value received: the more the customer succeeds, the more they pay, but they pay because they succeed. When the metric is right, a price increase is not negotiated, it is observed.
The main models and what they demand
The per-user (per-seat) model is the most common because it is legible and predictable. Its weakness is well known: it pushes the customer to limit the number of accounts, which slows internal adoption, the engine of retention.
The usage-based model charges real consumption. It tracks value closely, but it makes revenue less predictable and demands flawless billing. We detail the mechanics in our article on usage-based pricing.
The tiered model sells packages that bundle features and volumes. It simplifies buying and structures the move upmarket, provided each tier maps to a real segment.
The hybrid model combines a fixed base and a variable share. It is often the fairest for a mature SaaS, because it stabilises revenue while letting the bill grow with usage. It is also the hardest to configure and to explain.
No model is good in the abstract. The right model is the one whose billing unit matches how your customers draw value from your product.
Setting the price level
Once the model is chosen, the amount remains. This is where most companies improvise. Three reference points get you out of guesswork.
Start from value, not cost
The cost of producing software tells you almost nothing about what it is worth to the buyer. A module that costs you little can save a customer tens of thousands a year. Pricing on cost caps your revenue at the level of your spending, the opposite of a software model. The starting point is the value created for the customer: time saved, revenue unlocked, risk avoided. This logic is the subject of our article on value before price.
Look at the market without copying it
Competitors’ prices tell you what your buyers have already agreed to pay. That is an anchor, not a target. It reflects someone else’s cost base, target and positioning, rarely yours. If you create more value for a specific segment, a price aligned on the cheapest option leaves money on the table and weakens your message: a price that is too low reads as a confession of weakness as much as a good deal.
Measure willingness to pay
The only data that settles it comes from buyers themselves. In interviews, two questions frame a range: above what price would this product feel too expensive to consider, and below what price would you suspect it cannot keep its promises. Cross those answers across fifteen to twenty accounts in your target and you get a defensible band, far more solid than a number rounded off in a meeting.
Test the price on real deals
A price is not validated in a spreadsheet, it is validated in front of buyers. Put your new level on the next deals and watch what happens: win rate, decision speed, the size of the discounts asked for. If nobody questions the price, it is probably too low. If everybody walks, it is too high or poorly explained. The working zone is in between, where the price makes people think without making them leave.
Test one variable at a time. If you change the model, the level and the pitch at once, you will never know what produced the effect. Move in steps, measure, then generalise what holds.
One price per segment, not one price for everyone
A single price assumes every customer draws the same value from your product. That is almost always false. An SMB and a large account do not share the same usage, budget or tolerance for risk. The same product can be worth three times more to one than to the other, and an average price annoys both: too expensive for the small one, underpriced for the large one.
The answer is not to multiply price lists, but to identify two or three segments that buy for different reasons, and let the billing model do the sorting. A well-chosen value metric segments on its own: the small account pays little because it consumes little, the large account pays more because it gets more. That is the cleanest way to capture value without keeping a price list per customer type.
Revise pricing without breaking the base
A price frozen for three years is not proof of stability, it is a missed opportunity repeated thirty-six times. The good practice is an annual revision, fed by the year’s data: win rate by segment, average discounts, the gap between real usage and the tier bought.
A revision is prepared and sequenced. It applies first to new customers, where no commitment exists, before it reaches the base, which needs notice and a clear rationale. Raising prices is a discipline of its own, which we detail in our article on raising prices. The principle to keep in mind: delivered value must visibly have grown before the bill goes up.
The mistakes we see most
Three traps recur in our diagnoses. The first is the single price set at launch and never revisited, drifting a little further from real value each year. The second is a billing metric disconnected from usage, turning every renewal into a negotiation, because the customer pays for something they do not consume. The third is the reflex discount, which destroys in one signature the margin that pricing took months to build.
SaaS pricing is not a number, it is a living decision. The model decides how your revenue tracks your customers, the level decides your margin, and the revision decides whether both stay fair over time. The companies that treat price as a standing workstream, not a box filled once, are the ones whose margin improves without growth slowing down.
Frequently asked questions
- What is the difference between the pricing model and the price level?
- The model describes what you charge on: a per-user subscription, usage-based billing, a set of tiers. The level describes how much you charge for each. They are two separate decisions. A good price level sitting on the wrong model always creates friction over time, because the customer pays for something that does not track the value they get. Decide the model first, the amount second.
- How do you choose a SaaS value metric?
- Look for the unit that goes up when the customer succeeds with your product. For a support tool it might be tickets handled; for a data platform, volume analysed; for a CRM, active reps. A good metric is easy to understand, hard to game, and it grows with real usage rather than with a box ticked at signature. When the metric is right, a price increase is observed, not negotiated.
- Should you match competitors' prices?
- Look at the market, do not copy it. A competitor's price tells you what buyers have already agreed to pay, which is useful. But it reflects their cost base, their target and their positioning, not yours. If your product creates more value for a specific segment, pricing at the cheapest point on the market leaves money on the table and weakens your message, because a price that is too low reads as an admission of weakness.
- How often should you revise SaaS pricing?
- At least once a year, using the past year's sales data: win rate by segment, discounts granted, real usage against the tier bought. A controlled annual revision beats a price frozen for three years and then raised abruptly. It is prepared, communicated, and it applies to new customers first before it reaches the existing base.