What does a seat cost us?
In Maya’s second month, Brindlepath’s chief financial officer, Priya Nair, asked her a question in the hallway that she could not answer: “What does it cost us to serve one navigator seat for a year?”
Maya knew the price: Brindlepath charged per navigator seat, plus a platform fee per health system. She knew roughly what engineering cost. She did not know how much of the hosting bill, the support team, the implementation engineers and the integration maintenance each customer consumed. Nobody did. When Priya and Maya worked it out together over the following two weeks, they found that Brindlepath’s three smallest customers cost more to serve than they paid, mostly because each had needed a custom integration that the company now maintained for free.
That discovery changed the roadmap more than any customer interview. It also changed the way the CEO, the board and the sales team listened to Maya. A product leader who can explain the economics of a decision is a different kind of colleague from one who can only explain its features.
This chapter is the financial literacy that every product leader needs. It is not accounting. It is the handful of numbers that tell you whether your product is a good business.
Reading a SaaS P&L
A profit and loss statement, or P&L, shows revenue and costs over a period. For a software company, the structure is usually the same.
Revenue. For subscription businesses, the most watched figure is annual recurring revenue (ARR): the yearly value of active subscriptions. One-time revenue, such as implementation fees, is reported separately, because it does not recur.
Cost of revenue, also called cost of goods sold (COGS). The costs of delivering the service: hosting and infrastructure, third-party software embedded in the product, customer support, implementation and integration work, and, for AI products, model and inference costs.
Gross profit and gross margin. Revenue minus cost of revenue. Gross margin, gross profit as a percentage of revenue, is the single most important number for judging whether a software product is a good business. Mature software companies typically target gross margins well above 70 percent; services-heavy or infrastructure-heavy products run lower.
Operating expenses. Sales and marketing, research and development (where most product and engineering salaries sit), and general and administrative costs.
Operating income. Gross profit minus operating expenses. Many growing software companies run at an operating loss by choice, investing in growth. That is a strategy, not a failure, as long as the unit economics underneath are sound.
Figure 6.1 shows Brindlepath’s P&L in simplified form. The product leader’s influence is everywhere in it. Features that reduce support tickets raise gross margin. Integrations that need custom work for each customer lower it. A product that sells itself reduces sales and marketing cost. A product that needs a six-month implementation increases cost of revenue and delays revenue recognition.
Unit economics: is each customer worth it?
The P&L tells you how the business did. Unit economics tell you whether each new customer makes the business better or worse. Five measures matter most.
Customer acquisition cost (CAC). Total sales and marketing spend in a period, divided by the number of new customers won. In enterprise healthcare, where sales cycles run nine to eighteen months and involve many stakeholders, CAC is high.
Annual gross profit per customer. Average annual revenue per customer multiplied by gross margin. This is what a customer contributes toward paying back acquisition cost and funding everything else.
CAC payback period. CAC divided by monthly gross profit per customer: how many months it takes to earn back the cost of winning the customer. Shorter is better; investors often look for well under two years.
Retention. Gross revenue retention is the share of last year’s recurring revenue you kept from existing customers, ignoring upsells. Net revenue retention (NRR) adds expansion: upsells, extra seats, new modules. NRR above 100 percent means existing customers grow faster than others leave, which is the engine of the best software businesses.
Customer lifetime value (LTV). Gross profit per customer per year multiplied by the expected years a customer stays. The ratio of LTV to CAC is a common shorthand for the health of a growth model; a ratio around three or higher is a frequently quoted rule of thumb.
A worked example
Here is Brindlepath’s typical mid-sized customer after Maya and Priya finished their analysis. The numbers are illustrative.
- 30 navigator seats at $1,800 per seat per year, plus a $36,000 platform fee: $90,000 annual recurring revenue.
- Cost to serve, including hosting, support, integration maintenance and model costs: $22,500 a year, so gross margin is 75 percent and gross profit is $67,500 a year, or $5,625 a month.
- Acquisition cost, including a long sales cycle and a pilot: $120,000.
- CAC payback is therefore $120,000 ÷ $5,625, or about 21 months.
- With gross revenue retention of 90 percent, the average customer stays about ten years, giving a lifetime gross profit of roughly $675,000 and an LTV-to-CAC ratio of about 5.6.
Figure 6.2 shows why product decisions are financial decisions. A self-serve pilot that shortens the sales cycle makes the dip shallower. Fewer custom integrations raise gross margin and steepen the climb. A second module that customers buy in year two bends the curve upward. A clunky onboarding that causes churn in year three cuts it off before it pays back.
The three small customers Brindlepath discovered were losing money for one reason: their custom integrations had pushed cost to serve so high that gross margin was close to zero. Their curves never crossed the line. The fix was not to fire them; it was to move their integrations onto Brindlepath’s standard connector, which became the first item on the Now column of the roadmap.
Pricing: the most underused lever
Pricing is one of the most powerful levers a product leader has, and one of the least studied. In a widely cited analysis of 2,463 companies, Michael Marn and Robert Rosiello found that a 1 percent improvement in price, with volume unchanged, increased operating profit by 11.1 percent on average, far more than the same improvement in variable cost, volume or fixed cost.1 Later McKinsey analyses of larger samples put the effect somewhat lower, at around 8 to 9 percent, but the conclusion holds: small changes in price have outsized effects on profit.
Price to value, not to cost
There are three broad ways to set a price. Cost-plus pricing adds a margin to what the product costs to deliver. It is simple and almost always leaves money on the table, because customers do not care about your costs. Competitor-based pricing anchors to what alternatives charge. It is necessary context, but it makes you a follower. Value-based pricing starts with the value the customer receives and captures a fair share of it.
For Brindlepath, value was measurable. A mid-sized health system with meaningful readmission penalties, a few hundred high-risk discharges a month and a goal of preventing even a modest number of readmissions could point to savings and avoided penalties many times the price of the software. That calculation, built with the customer’s own numbers, became the heart of every proposal. In the United States, the Medicare Hospital Readmissions Reduction Program can reduce a hospital’s base Medicare payments by up to 3 percent for excess readmissions,2 which gave finance teams a concrete number to put in the model.
The pricing metric
The pricing metric is the unit you charge for: per user, per seat, per transaction, per patient, per facility, per outcome. A good metric has three properties: it grows as the customer gets more value, it is easy to understand and predict, and it is hard to game.
Brindlepath’s per-navigator seat price was easy to understand but had a flaw: as the product made navigators more efficient, customers needed fewer of them, so success shrank revenue. The company added a per-health-system platform fee scaled by annual discharges, which grew with the size of the problem the product solved. Finding the right metric is often worth more than finding the right price.
Packaging
Most software is sold in tiers, often three, sometimes called good, better, best. Packaging decides which features go in which tier. Two principles help.
- Put the features that define your best-fit customer in the middle tier, and price it to be the obvious choice. The top tier exists partly to make the middle look reasonable.
- Use features to fence tiers that larger customers genuinely need, such as advanced security, audit reporting, single sign-on and integrations, rather than artificially limiting the core experience.
Pricing deserves the same discovery as features: interviews about willingness to pay, tests of different packages with new prospects, and careful analysis of win and loss rates by price point.
Making the business case
When you propose a significant investment, a new product, a major feature or an integration, bring a business case. It does not need to be long. It needs five things:
- The problem and the customers who have it, from your discovery work.
- The revenue effect: new customers won, expansion, churn prevented, price increase supported. Show your assumptions.
- The cost to build: engineering, design and product time, plus any third-party costs.
- The cost to run: hosting, support, maintenance, model costs, partner fees. This is the part most business cases forget.
- The risks and how you will know: what has to be true, and the measure you will check after launch.
Show a range, not a single number. A case that says “between $400,000 and $1.2 million of new ARR in the first year, most likely around $700,000, if pilots convert at our historical rate” is more credible, and more useful, than one that says “$1 million.”
Brindlepath in practice: changing the pricing metric
As Chapter 6 described, Brindlepath’s per-seat price had a flaw: the more efficient the product made navigators, the fewer seats customers needed. At one customer, a successful rollout had let the health system cover more patients with the same team, and its finance department asked, reasonably, whether it could reduce seats at renewal.
Maya and Priya spent a quarter on the problem. They interviewed finance leaders at eight customers about how they budgeted for the product and what they thought they were paying for. The answers were consistent: customers thought of Brindlepath as paying for covering their high-risk patients, not for navigator logins.
The new model kept a smaller per-seat price, to cover the cost of each user, and moved most of the value into a platform fee scaled to the number of high-risk discharges the product covered each year. For a typical customer, the total price stayed about the same in the first year. As customers expanded coverage to more hospitals and more patients, revenue grew with the value delivered, regardless of how many navigators they employed.
The team tested the model with new prospects before offering it to existing customers, and gave existing customers the choice to switch at renewal. Most did, because it made their budget easier to explain: the cost tracked the problem they were solving. Net revenue retention rose over the following year, and the conversation about reducing seats disappeared.
The lesson is not that platform fees are better than seats. It is that the pricing metric should grow with the value customers receive, and that finding it took the same discovery work as any feature.
Common mistakes
Pricing on cost. Customers do not care what your product costs to build. Cost-plus pricing systematically leaves money on the table for products that create real value.
Ignoring the cost to serve. Features that require custom work, heavy support or expensive AI calls can make customers unprofitable while revenue looks healthy.
Discounting to close. Deep discounts win this quarter’s deal and set expectations for every renewal. Discount scope or terms before price, and attach discounts to commitments.
Choosing a pricing metric that punishes success. If customers pay less as your product makes them more efficient, your growth works against you.
Business cases that forget running costs. A case that counts only the cost to build will overstate the return of almost everything.
Single-number forecasts. One revenue number invites false precision. A range with explicit assumptions is more honest, and more useful for decisions.
Try this
Pick one typical customer and calculate, with help from finance if you need it, their annual revenue, their annual cost to serve (hosting, support, implementation, integration maintenance and AI costs), their gross margin, the cost of winning them and their payback period. Then repeat the exercise for your smallest customer and your most demanding one. Compare the three. Most product leaders who do this find at least one surprise, and at least one roadmap item that would change the numbers more than any new feature.
How AI changes this
AI changes your cost structure. Traditional software has very low marginal costs: serving one more user costs almost nothing. AI features add a real variable cost for every request, paid to a model provider or for your own infrastructure. A feature that summarizes every patient record every morning can quietly add meaningful cost per seat. Track AI costs per customer, per feature and per request from day one, and put them in your gross margin. Products that lean heavily on AI often run at lower gross margins than traditional software for exactly this reason, so plan for it rather than discovering it.
AI changes pricing. When software does work that a person used to do, pricing per seat can undercut the value you create, as Brindlepath found. Many AI products are experimenting with usage-based pricing (per request or per task), outcome-based pricing (per resolved case or prevented readmission) or hybrid models. Each shifts risk between you and the customer. Outcome-based pricing is compelling to buyers and hard to administer; agree in advance exactly how outcomes will be measured.
AI makes analysis cheap. An assistant can build a pricing model, a cohort analysis or a business-case spreadsheet in minutes, and run the sensitivity analysis you would never have had time for. Use it. Then check the formulas, because a confident spreadsheet with a broken formula is worse than none.