Cost of Revenue for SaaS: What the Metric Tells You

Cost of revenue for a SaaS company is the total cost of delivering your product to paying customers, including hosting, support, onboarding, and any third-party software embedded in the service. It sits above the gross profit line and tells you, with more honesty than most other metrics, whether your business model actually works at scale.

Most SaaS founders and marketing leaders treat cost of revenue as a finance problem. It isn’t. It’s a go-to-market problem, and getting it wrong will quietly undermine every growth initiative you run.

What Is Cost of Revenue for a SaaS Company?

Cost of revenue, sometimes called cost of goods sold or COGS in a SaaS context, is the direct cost of serving customers who are already paying you. It is not the cost of acquiring them. That sits in sales and marketing. Cost of revenue is what it costs to keep the lights on for each customer once they are through the door.

In a traditional product business, COGS is relatively straightforward: raw materials, manufacturing, logistics. In SaaS, the lines are blurrier. The product is software, but delivering it reliably requires infrastructure, people, and a stack of third-party services that most companies undercount.

Typical components of SaaS cost of revenue include:

• Cloud hosting and infrastructure costs (AWS, GCP, Azure)
• Third-party API costs and software embedded in the product
• Customer support salaries and tooling
• Customer success managers, where their role is primarily onboarding and retention rather than expansion
• Professional services and implementation costs
• Payment processing fees
• Security, compliance, and data costs tied to service delivery

What does not belong in cost of revenue: sales commissions, marketing spend, R&D, and general administrative costs. Those sit lower in the P&L. The distinction matters because misclassifying costs distorts gross margin, which is the number investors and acquirers use to assess the quality of your business.

Why Gross Margin Is the Real Signal

Gross margin is what you get after subtracting cost of revenue from total revenue. For SaaS companies, it is one of the most scrutinised figures in any due diligence process, and for good reason. It tells you how much of each dollar of revenue survives to fund growth, R&D, and operations.

The commonly cited target for SaaS gross margin is somewhere between 70% and 80%. Infrastructure-heavy or services-heavy SaaS businesses often land lower, sometimes in the 55-65% range. Pure software businesses with minimal support overhead can push above 80%. None of these numbers are inherently right or wrong without context, but they do set expectations for how the business will perform as it scales.

I spent time working with a mid-market SaaS business that was reporting strong ARR growth and had investors excited. When we looked at the unit economics properly, the gross margin was sitting at 52% because a large chunk of customer success cost had been misclassified as a sales expense. The business was growing, but the engine underneath was running hot. Every new customer added cost faster than it added margin. That is not a sustainable growth model, regardless of how clean the top-line numbers look.

If you are thinking seriously about go-to-market efficiency, gross margin is not just a finance metric. It determines how much room you have to invest in acquisition, retention, and product. Understanding how to structure and interpret it is part of building a go-to-market strategy that holds up under pressure.

The Classification Problem Nobody Talks About

One of the more persistent problems I see in SaaS finance is the inconsistent treatment of customer success costs. Some companies put customer success entirely in cost of revenue. Others split it between cost of revenue and sales, depending on whether the CSM is doing onboarding versus expansion. Others put it all in operating expenses and keep it off the gross margin calculation entirely.

There is no single correct answer, but there is a wrong answer: doing it inconsistently without documenting the rationale. Inconsistent classification makes it impossible to benchmark against peers, creates confusion in board reporting, and can mislead investors who are comparing your margins to industry norms.

The practical test I use: if a cost is incurred to keep a customer using the product at the level they have already paid for, it belongs in cost of revenue. If the cost is incurred to expand the account or renew the contract, it is closer to a sales cost. Onboarding a new customer sits firmly in cost of revenue. Pitching an upsell to an existing customer does not.

This matters beyond the P&L. When you are modelling customer lifetime value against customer acquisition cost, you need the cost side to be honest. Understating cost of revenue flatters LTV:CAC ratios and can lead to growth decisions that look rational on a spreadsheet but destroy cash in practice.

How Cost of Revenue Connects to Pricing Strategy

Most SaaS pricing conversations start with the market. What are competitors charging? What will the customer pay? Those are valid inputs, but they are incomplete without an understanding of what it actually costs to serve different customer types.

I have seen this play out repeatedly across agency clients and businesses I have worked inside. A company prices a lower-tier plan to drive volume, then discovers that the support burden from that segment is disproportionately high. The plan generates revenue but destroys margin. The business is effectively subsidising its smallest customers with the margin from its largest ones, which is a fine strategy if it is intentional and if the smaller customers actually convert upward. It is a problem if nobody has run the numbers.

Cost of revenue analysis by customer segment, or ideally by pricing tier, tells you which parts of your business are actually profitable. A SMB customer paying £200 a month who generates three support tickets a week looks very different from an enterprise customer paying £2,000 a month who has a dedicated CSM but rarely contacts support. The revenue ratio is 10:1. The cost ratio might be 20:1 in the other direction.

This is where cost of revenue stops being a finance exercise and becomes a go-to-market decision. If certain customer segments are structurally unprofitable at your current pricing, you have three options: raise prices, reduce the cost of serving them, or stop acquiring them. None of those decisions can be made well without accurate cost data.

BCG’s work on commercial transformation makes a similar point about the relationship between cost structure and growth strategy. Companies that understand their unit economics at the customer level tend to make sharper decisions about where to invest in growth. You can read more about that framing in their guide to commercial transformation.

Infrastructure Costs and the Scaling Assumption

One of the promises of SaaS is that infrastructure costs scale more slowly than revenue. As you add customers, your per-unit hosting cost should fall because you are spreading fixed infrastructure costs across a larger base. This is the operating leverage argument, and it is real, but it does not happen automatically.

Early-stage SaaS companies often over-provision infrastructure because they are building for anticipated growth rather than current load. That is sometimes sensible engineering, but it inflates cost of revenue in the short term and can create the illusion that margins will improve dramatically with scale. Sometimes they do. Sometimes the architecture needs to be rebuilt before the unit economics improve, and that is an expensive surprise.

The other trap is third-party API costs. Many SaaS products are built on top of other services, and those costs scale directly with usage. If your product uses an AI inference API, a mapping service, or a payments processor, those costs move with volume in ways that hosting costs often do not. When modelling cost of revenue at scale, those variable components need to be tracked separately from fixed infrastructure costs.

For teams interested in how growth mechanics interact with cost structure, this overview of growth hacking principles from Crazy Egg is a useful starting point, even if the framing is more acquisition-focused than unit economics. The underlying point about sustainable growth mechanics applies.

What a Healthy Cost of Revenue Structure Looks Like

There is no universal benchmark that applies to every SaaS business, but there are patterns worth knowing.

Pure-play software businesses with minimal services components and low support intensity tend to run cost of revenue at 15-25% of revenue, producing gross margins of 75-85%. These are the businesses that attract premium multiples because the operating leverage is clear.

SaaS businesses with significant professional services, heavy onboarding requirements, or embedded third-party costs tend to run cost of revenue at 30-45% of revenue. Gross margins in the 55-70% range. Still a good business, but the path to profitability is narrower and the growth investment headroom is tighter.

Below 55% gross margin, you are in territory where the SaaS label starts to obscure what is really a services business with a software component. That is not necessarily a problem, but it changes how the business should be valued, managed, and grown.

The most useful internal exercise is not comparing your gross margin to a benchmark. It is tracking how gross margin changes as you add customers. If gross margin is improving as revenue grows, the model is working. If it is flat or declining, something in the cost structure is not scaling the way you assumed it would.

Cost of Revenue as a Marketing Input, Not Just a Finance Output

This is where most marketing teams disengage from the conversation, and it is a mistake. Cost of revenue data should directly inform marketing decisions in three ways.

First, it should shape ICP targeting. If you know that certain customer segments are structurally more expensive to serve, your acquisition strategy should weight toward the segments that generate better margin, not just better revenue. Acquiring high-volume, low-margin customers at scale is how companies end up with impressive ARR and poor cash generation.

Second, it should inform channel investment decisions. When I ran performance marketing for large accounts, one of the recurring problems was that CAC calculations treated all revenue as equal. A customer acquired for £500 who generated £1,200 ARR looked identical to a customer acquired for £500 who generated £1,200 ARR but required £600 in annual support costs. They are not the same customer. The second one barely breaks even in year one. Cost of revenue data lets you build more honest LTV models, which in turn lets you make better channel investment decisions.

Third, it should influence how you position the product. If your cost of revenue is high because onboarding is complex, that is a product and positioning problem as much as an operations problem. Marketing that sets accurate expectations about implementation complexity reduces support burden and churn. Marketing that oversells ease of use drives up support costs and damages retention.

I have judged enough Effie Award entries to know that the campaigns that win on effectiveness are almost always the ones where the marketing team understood the commercial model underneath the brand. The creative work matters, but it is the commercial grounding that makes campaigns actually move business outcomes. Cost of revenue is part of that commercial grounding.

If you want to explore how unit economics thinking connects to broader growth strategy, the articles in The Marketing Juice’s go-to-market and growth strategy hub cover this territory in more depth, including how to align acquisition and retention investment with the actual economics of your business model.

Common Mistakes and How to Avoid Them

The most common mistake is treating cost of revenue as a static number rather than a dynamic one. It changes as you scale, as you change your support model, as you negotiate better infrastructure contracts, and as your customer mix shifts. Reviewing it quarterly and understanding the drivers of change is more valuable than hitting a particular percentage target.

The second most common mistake is optimising cost of revenue in isolation from customer outcomes. Cutting support headcount reduces cost of revenue. It also increases churn if customers cannot get help when they need it. The metric you are managing is gross margin over the customer lifetime, not gross margin in a single quarter. A company that genuinely delivers for its customers at every touchpoint will retain them longer, reducing the effective cost of revenue per dollar of lifetime revenue. Marketing and support theatre cannot substitute for that.

The third mistake is failing to model how cost of revenue changes under different growth scenarios. If you plan to double ARR in 18 months, what happens to your infrastructure costs? Your support team size? Your third-party API spend? Growth plans that do not include a cost of revenue model are incomplete, and they tend to produce nasty surprises when the growth actually arrives.

For teams building out their growth tooling and wanting to understand how different platforms support unit economics analysis, Semrush’s breakdown of growth hacking tools is a reasonable starting point for the acquisition side of the equation, even if it does not address cost of revenue directly.

BCG’s research on understanding the financial needs of evolving customer populations is also worth reading for how customer segment economics should shape go-to-market decisions, even if the context is financial services rather than SaaS.

Building a Cost of Revenue Model That Is Actually Useful

If you are starting from scratch, here is a practical approach. Pull every cost line that relates to delivering your product to existing customers. Separate them into fixed costs (infrastructure you would pay regardless of customer count) and variable costs (costs that scale with usage or customer volume). Then calculate what each costs per customer per month at your current scale.

Next, model what those costs look like at 2x, 5x, and 10x your current customer count. Where do fixed costs become negligible per customer? Where do variable costs become a problem? That exercise will tell you more about the sustainability of your business model than most investor decks.

Then segment it by customer type. Enterprise versus SMB. High-usage versus low-usage. Long-tenure versus recently acquired. The differences are usually significant enough to change how you think about acquisition priorities and pricing architecture.

Finally, connect it to your marketing and sales data. What is the fully loaded cost of acquiring a customer in each segment, and what is the gross margin that customer generates over their lifetime? That is the LTV:CAC ratio that actually means something, because it accounts for the cost of serving the customer, not just the cost of acquiring them.

Forrester’s work on scaling and agility in business operations touches on related themes around how companies structure themselves for sustainable growth. Their agile scaling framework is worth reading for the operational discipline it requires to grow without letting costs spiral.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions