SaaS Metrics

In a business that sells once and moves on, this month's sales figure tells you most of what you need. Sell 1,000 pairs of shoes for 100,000 dollars and next month starts at zero again. Revenue, cost of goods sold, and operating profit were built for that rhythm, and the columns in an accounting ledger are laid out to match.

In a software as a service (SaaS) business sold by subscription, the same sentence means something else. Selling 100,000 dollars this month means 100,000 dollars arrives next month too, and the month after. The ledger can't show that difference. Take 120,000 dollars up front for an annual contract in January and only 10,000 of it counts as January revenue; the remaining 110,000 is recorded as a debt for eleven months of service not yet delivered. Run the reverse case — customers paying monthly, growing steadily — and the ledger still shows only that month's slice.

So meetings kept talking past each other. When the founder asked how much the company earned, finance read the 30,000 dollars of recognised revenue, sales read the 180,000 dollars of contracts signed that month, and engineering counted how many paying accounts were active. None of the three numbers was wrong, and none of them answered the same question. Trying to forecast next month in that room, there was no line to build the forecast on. Someone would eventually multiply a plausible-looking figure by a plausible-looking growth rate, and nobody could say where either came from.

Investor meetings hit the same wall. What an analyst wants to see is how much money is coming in every month right now and how fast that figure is climbing, and financial statements have no such row. Every company arrived with a number calculated its own way, which made it impossible to set two companies in the same market side by side. One counted one-time implementation fees as part of the recurring figure. Another booked a full annual contract into the month it was signed instead of dividing it across twelve. Since the incentives all pointed toward the larger-looking method, analysts stopped accepting the number as presented and started asking how each one had been counted.

SaaS metrics are the set of counting rules the industry settled on to clear that up. Count only money contracted to repeat, split the change since last month into where it grew and where it leaked, and calculate what an average customer pays and how much of that survives — all the same way at every company. They don't replace the accounting ledger. They sit beside it and get read together with it.

Thirteen terms sit under SaaS metrics in five groups, divided by what they measure: the size of the money coming in now (revenue), how that size changed over a month (revenue movement), how much one customer accounts for (unit revenue), how much of it survives (profitability), and how long the current pace can be held (health).

Revenue measures size. Two terms, differing only in the period.

1. Monthly recurring revenue (MRR) — the sum of what is contracted to repeat this month. A hundred customers averaging 500 dollars a month gives 50,000. Annual contracts are divided by twelve and only one month's share goes in. One-time setup or consulting fees are excluded, since they won't arrive again next month, and leaving them in is the single most common way this figure gets quietly inflated.
2. Annual recurring revenue (ARR) — the same money on a yearly basis. Multiply that 50,000 by twelve and you get 600,000. Consumer businesses billed monthly tend to speak in MRR; enterprise businesses built on annual contracts speak in ARR. They state the same fact in two units, so mixing them in one meeting makes the company look twelve times bigger or smaller than it is.

Revenue movement breaks the change in that size into four pieces. The gap between last month's recurring revenue and this month's gets written down separately for what came in and what went out.

1. New MRR — added by customers paying for the first time this month.
2. Expansion MRR — added by existing customers buying more seats or moving to a higher plan.
3. Churned MRR — lost to cancellations and downgrades.
4. Net new MRR — the three combined. New of 8,000, expansion of 3,000, and churn of 5,000 nets out to 6,000.

Splitting it four ways matters because identical totals hide different companies. A net 6,000 built from 11,000 new against 5,000 churned and a net 6,000 built from 7,000 new against 1,000 churned look the same on one line, but the first is pouring water into a leaking bucket and the second has an intact bucket. Cut the marketing budget and only the first one collapses. The four pieces also point at different owners: new business belongs to sales, expansion to customer success, and churn to whichever team let the reason for leaving take root months earlier.

Unit revenue looks at what one customer pays on average. The only difference is the denominator.

1. Average revenue per user (ARPU) — divided by the number of people. Used where individuals pay for themselves.
2. Average revenue per account (ARPA) — divided by the number of customer companies. Used where many people work under one contract. Monthly recurring revenue of 50,000 across 100 accounts gives 500 per account, and if those accounts hold 2,000 users, 25 per user.

Profitability asks how much of the money survives. Two terms, differing in what gets subtracted. Gross margin is revenue minus the costs of actually delivering the service — servers, support salaries, payment processing fees — expressed as a percentage. Software is built once and sold many times, so 70 to 80 percent is common, and a figure in the forties usually means a business that runs on people rather than on product. Contribution margin is what one unit leaves after variable costs, which then goes toward fixed costs like office rent. It's an amount rather than a ratio, so it tells you directly what one more customer adds. At 300 dollars per account per month against 90,000 in fixed costs, the company breaks even at 300 accounts.

Health asks how long the pace can be held. All three compare two numbers, and all three put growth next to what that growth cost. The rule of 40 adds the annual growth rate to the operating margin and checks whether the sum clears 40. Growing 60 percent at a margin of minus 25 gives 35, which falls short; growing 20 percent at a margin of 25 gives 45, which clears. Burn multiple divides the cash burned over a period by the ARR added in it — under 1.5 reads as efficient, over 3 means revenue is being bought with money. CAC payback period divides the cost of acquiring one customer by the profit that customer leaves each month, counting the months to break even. Spend 1,200 dollars to win a customer who leaves 100 a month and the answer is twelve. Past twelve months, the odds that the customer leaves before paying for itself climb quickly.

The five groups read as one line. Revenue gives the current size, movement gives the direction, unit revenue says whether that movement came from large customers or small ones, profitability says whether the money is money that stays, and health says what holding this pace costs. Skip the first four and you can't explain why the health numbers came out the way they did. Read only the first four and you don't know how much longer the pace can last.

"Net new has been positive three months running, but it's coming from expansion, not new business."

Existing customers are buying more while nobody new arrives, so the figure looks healthy and sales still needs attention.

"Recurring revenue is up this month but revenue per account is down."

Only small customers are coming in. The total rose because the count rose, and each account is worth less, which can mean support load growing faster than income.

"On the rule of 40 we're sitting at 35. Either growth goes up or costs come down."

The sum of growth and margin misses the bar, so one of the two has to move before the next round.

Pagedeck reads these numbers together on the third Wednesday of each month. Grace runs the meeting and Mia reads out recurring revenue and its four pieces in order. Last month net new was positive while churn came in at twice the prior month. Sofia pulled the list of departing accounts and found that six of the eight had signed during a half-price promotion the previous quarter. They all left in the month the full rate was first charged. Which made the churn spike a pricing question rather than a product one, and moved the conversation from what was wrong with the software to what the promotion had been selecting for.

Ravi reads the CAC payback line in the same meeting. Customers acquired through advertising take fourteen months to pay back; customers who arrive through referrals take four, and the gap had held its shape for months. Grace used those two lines to move part of next quarter's advertising budget into referral rewards. The numbers didn't make the decision. They moved the argument about what to cut off instinct and onto a shared table. Mia left the meeting with one line for next month: track separately what share of promotional accounts survive their first full-price invoice.

The most common mistake is reading recurring revenue as money in the bank. The word revenue invites it. But the figure describes what the contracts currently in force are worth per month, and it moves independently of when payments land. Collect 120,000 dollars at once for an annual contract in January and recurring revenue is still 10,000, while a company with 50,000 in recurring revenue whose customers pay in arrears may have an empty account this week. Treat the two as the same thing when planning cash and a company that looks healthy on paper comes up short on payday. Recurring revenue and actual cash flow always get read separately.

The other one is the belief that tracking more numbers means managing better. Tooling made it easy to put all thirteen on one screen, which encourages it. In practice only two or three of them change a decision at any given stage. A company with twenty customers measuring burn multiple weekly changes nothing by doing so, and the same hour spent calling the two customers who churned would be worth more. Put all thirteen on the wall and the meeting spends forty minutes reading numbers and ends without deciding anything. Decide what this quarter's open questions are first, then keep the numbers that separate the answers — reverse that order and metrics add work instead of removing it. Pagedeck started with twelve lines on the dashboard and now keeps four. The rest are still calculated, and get pulled up once a quarter or whenever a specific question makes one of them the deciding number again.