How to set a SaaS pricing strategy that scales

What a SaaS pricing strategy actually is

A SaaS pricing strategy is three decisions made in order:

1. The pricing metric — the unit you charge on.
2. The tiers — how you package that metric into a small set of options.
3. The unit economics — the model that proves the price pays for the business.

The order is not cosmetic. The metric constrains the tiers, and the tiers feed the economics. Teams usually start at step two, arguing about tier names and price points, while the decision that actually determines whether pricing scales — the metric — goes unmade. Get the metric wrong and no amount of tier-tuning rescues it.

Why the pricing metric is the decision that matters most

A pricing metric is the thing you charge per unit of — per seat, per contact, per API call, per gigabyte stored. It is the most consequential pricing decision because it determines the relationship between your revenue and your customer's success.

The right metric has one property: it grows as the customer gets more value. When the customer succeeds, they consume more of the metric, and your revenue rises with them — no renegotiation, no upsell motion, no friction. The account expands on its own.

The wrong metric breaks that link in one of two ways:

• It caps your revenue — you charge a flat fee, the customer's usage explodes, and you capture none of the new value you are delivering.
• It punishes growth — the customer pays more precisely when the product hurts, so success feels like a penalty and they cap their own usage to control the bill.

A messaging platform that charges per message punishes its best customers for sending more; one that charges per active recipient grows with the audience the customer is proud of. Same product, opposite trajectories, because the metric points at value in one case and at pain in the other.

How to find the value metric

The reliable way to find the metric is to ask what unit of your product maps to the customer's own definition of success. Not what is easy to measure — what the customer would point to and say "the more of this, the better my life."

Test any candidate metric against three questions:

Question	Why it matters
Does revenue rise as the customer wins?	This is the whole point — value alignment
Is it easy to understand and predict?	A metric buyers cannot forecast, they distrust
Does it punish the customer for growing?	If yes, they will cap usage and resent you

A metric that passes all three scales quietly. One that fails the first is cost-plus in disguise; one that fails the third turns your product's success into the customer's problem.

This is the judgment AI cannot make for you. Locating value requires knowing what your customers actually care about, and that comes from interviews, not from a benchmarking table.

The four shapes a price can take

The metric answers what you charge for; the pricing model is the shape that metric takes. Four dominate SaaS, and each fits a different kind of value:

• Per-seat — you charge per user. Simple to forecast and the default for collaboration tools; Figma, Notion, and Linear all price this way. It shines when value rises with headcount, and strains when a handful of power users create most of the value.
• Per-usage (consumption) — you charge for what the customer consumes: hosts monitored, events logged, credits burned. Datadog and Snowflake price this way. Revenue scales with the customer's own growth automatically, with no upgrade decision required — but the customer trades predictability for it, and a surprise bill erodes trust fast.
• Per-account / flat — a fixed price per company or per tier. Predictable for the buyer, but it caps your capture and needs a strong expansion path to keep growing.
• Per-outcome — you charge for the result delivered: a booked meeting, a resolved ticket. The tightest possible value alignment, and the hardest to attribute cleanly.

These are not interchangeable. Charging Datadog's customers per seat would be strictly worse — engineers don't buy "seats," they consume infrastructure. The model has to match where the value actually accrues.

Why value-based beats cost-plus and competitor pricing

There are three ways to set a price, and two of them ignore the only party who matters.

• Cost-plus pricing starts from what the product costs you to build and adds a margin. It ignores the customer entirely — your costs are your problem, not their reason to pay.
• Competitor pricing copies a rival's number. It outsources your strategy to a company whose costs, positioning, and value you do not share.
• Value-based pricing starts from the outcome the customer receives and captures a fair share of it.

Only the third scales, because only the third is anchored to something that grows — the customer's value. A product that saves a customer real money each month has pricing headroom that has nothing to do with what it cost to build. The metric is how you connect your price to that value; value-based pricing is the philosophy that says you should.

Your price is a positioning signal, not just a number

Before a buyer evaluates a single feature, your price has already told them what kind of company you are. A number far below the category says "for individuals and small teams"; a number far above says "for enterprises with a procurement process." Pricing is a positioning decision as much as a revenue one — the same product at $9 a month and at $900 a month is sold to two different buyers, through two different motions, against two different competitors.

The practical rule: price for the customer you want, not the one who wanders in. A price set low to avoid scaring anyone off attracts the smallest, least committed, highest-support accounts — the anti-ICP of your unit economics — and quietly repels the buyers who read a low number as a low-stakes tool. Let the price match the value your best customers get, and it becomes a filter that does some of your qualification for you.

Research willingness to pay before you name a number

A price set from inside the building is a guess. The number should come from the same place your metric did — the buyer. You are not asking "what would you pay?", which people answer badly; you are triangulating from what they already spend on the problem, what the alternatives cost them, and how much the outcome is worth to them.

Two habits sharpen it. First, segment willingness to pay by ICP — a self-serve solo user and a mid-market team feeling the same pain will pay very differently, and a single price leaves money on one side and deals on the other. Second, read the range, not a point: buyers who all name roughly the same number tell you the price is obvious; a wide spread tells you the value is unclear, which is a messaging problem to fix before a pricing one. Price is a signal as much as an exchange — set it too low and buyers question the value; too high without proof and they walk before they hear the pitch.

How to package the tiers

Once the metric is set, package it into three tiers — the good-better-best shape. Three is the reliable number: one tier gives no sense of choice, five or more create paralysis and bury your best option.

The three tiers do three jobs:

• The entry tier removes the reason to say no. It is the low-commitment way in, priced so the decision is easy.
• The middle tier is the one you want most buyers to choose. Design everything so it is the obvious pick — put the feature buyers most want to cross a tier for right here, not in the top tier.
• The high anchor makes the middle look reasonable. Its main job is comparison; a buyer who sees the anchor reads the middle as sensible rather than expensive.

The mistake is putting the most-wanted feature in the top tier to drive upgrades. It backfires — buyers who need it balk at the jump, and buyers who do not need it feel the middle is thin. The feature that pulls people across a boundary belongs in the tier you want them to land in.

Freemium, free trial, or reverse trial — the way in

Tiers decide what you charge; the packaging above them decides how a buyer gets far enough to want to pay. Four shapes dominate, and the choice is strategic, not cosmetic:

• Free trial — the full product, time-boxed. Best when value is fast and obvious, and when the setup itself is the commitment; Datadog's two weeks give an engineer time to wire in real data.
• Freemium — a free tier that lasts forever, with limits. Best when the free users are themselves the growth engine — Figma made viewers free, so every paid designer became an acquisition channel.
• Reverse trial — start everyone on premium, then fall back to free if they don't convert. Notion's default: no countdown anxiety, users just discover one day that the best features cost extra.
• Hybrid — a lasting free tier plus trials of the paid tiers, common in multi-product suites like HubSpot.

The one non-negotiable: whichever you pick has to get the user to the product's aha moment while the door is open. A free trial that expires before the customer reaches first value is a demo with a deadline.

Pricing and your acquisition motion have to fit

The packaging and the pricing model are not independent choices — get the pairing wrong and each undermines the other. Per-usage pricing and a free trial fit naturally: the trial buys time to set up, and consumption billing captures the upside automatically once the customer is live (Datadog). Per-seat pricing pairs with freemium or a reverse trial, where the job is to hook one user and let seats multiply as they pull in their team (Notion, Figma). Consumption pricing suits a sales-led proof-of-concept, where the customer commits to a credit balance rather than a seat count (Snowflake).

The failure mode is bolting a pricing model onto an acquisition motion it fights. A per-seat price on a product whose value is consumption-driven caps your revenue; a self-serve free trial on a product that genuinely needs a sales-led evaluation just adds a step nobody uses. Choose the metric, the model, and the way in as one decision — because your buyer experiences them as one.

Why unit economics are the proof

Pricing is not done when the tiers look right. It is done when the unit economics show the price pays for the business. Three numbers carry the proof:

• CAC (customer acquisition cost) — what you spend to win one customer.
• LTV (lifetime value) — the total profit that customer returns over their life.
• Payback period — how long it takes the customer's revenue to repay their CAC.

LTV must exceed CAC by a healthy multiple, or you are buying revenue at a loss. The common rule of thumb is an LTV of at least ~3× CAC, with CAC payback under roughly a year — conventions, not laws, but a useful floor to price against. Payback has to be fast enough that growth does not drain your cash, because a long payback means every new customer is a hole you fund up front and fill slowly. A pricing structure that looks attractive but produces a payback period your cash cannot survive is not a strategy — it is a countdown.

Model these per tier. If the middle tier — the one you want most buyers in — does not clear payback fast enough, the pricing is not finished, no matter how clean the tiers look.

Expansion is where the durable revenue lives

New-customer revenue is expensive; revenue from customers you already have is not. A pricing structure is only finished when it has a path for accounts to grow — which is the whole reason the value metric matters. When your price rises automatically as the customer succeeds, expansion happens without a new sale, and net revenue retention can climb above one hundred percent: the existing base grows faster than any of it churns.

This is the quiet advantage of a metric tied to value and a model — per-usage or per-seat — that scales with the account. Land on the tier that removes the reason to say no, then let the customer's own growth do the upselling. A business that has to replace churned revenue with new logos every month is running up a down escalator; one whose accounts expand on their own is compounding. The pricing decisions above — the metric, the model, the tiers — are what determine which of those two businesses you are building.

How to stage price changes

Pricing is not set once. As you learn what customers value, you will raise prices — and how you raise them decides whether it costs you trust.

Model the downside before you launch: what happens to the unit economics under a churn spike, or if a competitor starts a discount war. Then plan the increase you know is coming. Grandfather existing customers where you can, give notice, and tie the new price to visibly more value. A price increase that arrives with new capability reads as fair; one that arrives as a bill reads as a betrayal. The metric you chose at the start is what makes future increases feel earned — when price rises with value the customer already receives, the increase is a conversation, not a shock.

Discount deliberately, or not at all

A discount is a permanent statement dressed as a temporary favor. The first time you cut price to close a deal, you teach that buyer — and, through their network, the next one — that your list price is negotiable, and you compress the margin on every renewal that follows. Discount when it buys something you want back — a longer commitment, an annual prepay, a reference logo — and refuse when it only buys the close. Hold the line publicly, too: a price that visibly caves under pressure signals that the value was never worth the number. And if reps are discounting constantly to win, that is rarely a pricing-flexibility problem — it is a value or a targeting problem wearing a discount as a bandage.