Frameworks, core principles and top case studies for SaaS pricing, learnt and refined over 28+ years of SaaS-monetization experience.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
.png)
With your metric chosen, you can set the actual price. Rate setting is step four, and it comes near the end on purpose.
A lot has to be settled first: who you sell to, what you are trying to achieve, how the packages are built, and what metric you charge on. Once those are set, the price itself is the easier decision. Rate setting is simply where you lock in the number using data.
You cannot set a rate without a clear goal, and this is the step people skip most. Executives ask for "new pricing" or "pricing that fits AI" without saying what success looks like.
Two goals are common, and they often conflict:
Capture market share. Price low and move aggressively to take the market before rivals do. This matters more now, because AI-assisted development lets competitors build and ship faster.
Protect margin. Make sure you never lose money, even in the short term, which matters when AI inference is driving up your cost of goods sold.
You want both, but they pull in opposite directions. Decide which one wins before you set a number, because it changes the answer.
So what data should inform the decision?
Competitive comparison. The most common starting point, and a fair one, but not the end of the process. Use it to form hypotheses: should we go freemium, raise or lower prices, split one package into two?
Internal data. Your telemetry, usage, COGS, and billing data on who buys what and why. The blind spot: internal systems only show people who became customers. They say nothing about the larger pool of prospects who looked and said no, and why. You have to look outside your own base.
External market testing. Quantitative and qualitative research across the whole market, not just your current customers.
You need both halves of the story: the quantitative "how much and how big," and the qualitative "why."
It also helps to remember that there is no single fair price in software. The same unit can be worth wildly different amounts to different buyers. A DocuSign signature was once worth about seven dollars to a bank using it to speed up mortgage applications, and about twenty cents to a gym using it to sign a facility-tour waiver that earned no revenue. You cannot squeeze different price sensitivities into one price that is fair to everyone.
The answer is to build the full range and capture it through differentiation: packaging that separates higher-value use from lower-value use, plus a clear set of discounting rules. That deliberate differentiation is a core part of good pricing, not a failure of it.
Finally, while I would never tell anyone to price purely on cost, cost matters far more in the Agentic AI era than it did in classic value-based SaaS. Expect GenAI and agentic margins to sit below traditional SaaS, and remember that the choice of LLM model can swing your economics hard.
With those principles in place, here is how each of the five companies set its actual prices, and the strategy behind each number. Cursor ($20/mo Pro, $40/user/mo Business): The gap between what Cursor delivers and what it charges is huge, but the crowded field (Windsurf, Copilot, Claude Code, and a dozen more) holds the price down. This is a penetration play: trade short-term revenue for market share and workflow lock-in. Strategic intent: Market share.
Devin ($20/mo entry, ACUs at $2 to $2.25): The 96% price drop from $500 to $20 for the entry tier reflected competitive reality. The ACU rate is tied to compute cost, not output value, the right call at today’s reliability. As success rates improve, the ACU price can shift toward value without changing the structure. Strategic intent: Revenue optimization, with a path to margin as reliability improves.
Harvey AI (about $1,200/user/month): At this rate, Harvey’s annual minimum of roughly $288K is firmly set for Am Law 100 and Fortune 500 buyers. The rate holds up against the value ceiling: at $1,200 a month per lawyer, Harvey needs to save only a few hours a month to justify the cost. What matters most is the competitive ceiling, as CoCounsel, Legora, and others chase the same wallets. Strategic intent: Margin optimization within a premium segment.
11x (about $5,000/month): Anchored to the cost of a human SDR, roughly a $60K-a-year rep. The anchor makes sense. The problem is that Alice replaces about 40% of the job at 100% of the cost. With Agent Frank at $499/mo and AiSDR at $900/mo, the price is getting harder to defend. Strategic intent: Margin, but under pressure.
Sierra (per resolution, about $150K to $350K+ in Year 1): The per-outcome rate will face downward pressure as competitors multiply. Sierra’s moat is not the AI. It is the depth of enterprise implementation, the multi-channel orchestration, and the outcome-measurement infrastructure that makes the model work. Strategic intent: Value capture at the enterprise level. A hybrid model (a modest platform fee for predictability plus outcome fees for alignment) eases the budgeting anxiety CX leaders feel when moving from predictable seats to variable outcomes.
Join companies like Zoom, DocuSign, and Twilio using our systematic pricing approach to increase revenue by 12-40% year-over-year.