SaaS metrics glossary

The subscription revenue a business can expect to bill again next month, normalised to a monthly amount. Annual contracts are divided by twelve rather than counted in the month they are signed. One-off implementation fees, professional services and usage overages that will not recur are excluded — the point of the measure is that it is repeatable, and a number inflated with one-time revenue stops being a base you can project from.

See also ARR, ARPA and Net new MRR.

ARR = MRR × 12

The annualised run rate of recurring revenue. It is a snapshot of the current month projected forward, not a statement about revenue actually earned over the past year, and confusing the two is a common error in board material. A company at $400,000 MRR has $4.8M of ARR even if it only earned $3.1M over the preceding twelve months.

See also MRR and Run rate.

ARPA = MRR ÷ active accounts

Total recurring revenue divided by the number of paying accounts. It appears inside lifetime value and inside CAC payback, so a business that reports ARPA across wildly unequal customers is feeding a distorted average into both. Where a long tail of small accounts sits alongside a handful of large ones, segment ARPA and model the segments separately.

See also LTV and CAC payback period.

Net new MRR = new + expansion − contraction − churned

The month-over-month change in recurring revenue broken into its four components. Reporting only the net figure hides the mechanics: a flat month can be a quiet one or it can be aggressive acquisition exactly cancelling severe churn, and those are different companies with different problems.

See also MRR, Revenue churn and Expansion revenue.

In this simulator, the share by which recurring revenue grows each month from sources other than the re-invested acquisition budget: inbound demand, word of mouth, expansion inside existing accounts and any sales activity already inside the fixed cost base. It is applied to the current MRR, so it compounds. Revenue bought with re-invested profit is modelled separately, through CAC and ARPA, so that the cost of that growth is visible rather than assumed.

See also Re-investment rate and CAC.

Any figure from one period multiplied out to a year. Useful shorthand, dangerous as a forecast: a run rate contains no information about growth, seasonality or the durability of the period it came from. Quoting a run rate from the strongest month of the year is a well-worn way to make a business look larger than it is.

See also ARR.

The value of contracts signed in a period. Bookings lead revenue, sometimes by a long way: a three-year deal signed in March is one booking and thirty-six months of recognised revenue. Bookings, billings and recognised revenue are three different numbers, and a plan that mixes them will not reconcile to the bank account.

See also ARR and Deferred revenue.

Money already received for a subscription period that has not yet elapsed. It sits on the balance sheet as a liability because the obligation to deliver is outstanding. Annual prepayment converts future revenue into present cash, which is why it matters enormously to runway and not at all to recognised revenue.

See also Cash runway and Bookings.

Revenue churn = MRR lost ÷ MRR at start of period

The proportion of recurring revenue that disappears in a period through cancellation or downgrade. This is the churn measure used throughout the simulator, because dollars pay payroll and logos do not. It compounds: a cohort surviving at rate (1 − churn) each month retains (1 − churn) to the power n after n months, which is why a point of monthly churn changes the size of company you can build.

See also Logo churn, Gross revenue retention and LTV.

Accounts lost as a proportion of accounts held. It is the right measure for support load, onboarding capacity and product-fit questions, and the wrong one for financial modelling. Logo and revenue churn diverge whenever customers are unequal in size, which is always: losing 5% of logos can cost 1% of revenue, or 9%, depending entirely on which logos left.

See also Revenue churn.

Annual GRR = (1 − monthly revenue churn) ^ 12

The share of last period's recurring revenue still present this period, counting cancellations and downgrades but not upsells. Because expansion is excluded it can never exceed 100%, which is precisely what makes it useful: it is the one retention number that cannot be flattered by selling more to the customers who stayed.

See also Net revenue retention and Revenue churn.

NRR = (starting MRR + expansion − contraction − churn) ÷ starting MRR

Gross retention with expansion revenue added back in. Above 100% the installed base grows without a single new customer, which turns growth from additive into compounding and removes the revenue ceiling that churn otherwise imposes. Read it beside gross retention rather than instead of it: a business losing 6% of revenue monthly while expanding 7% reports a healthy-looking 101% while its customer base quietly empties.

See also Gross revenue retention and Expansion revenue.

Revenue growth inside accounts you already have — more seats, higher usage tiers, added modules. It is the cheapest revenue in the business because it carries no acquisition cost, and it is the mechanism behind net revenue retention above 100%. In this simulator expansion is folded into the organic growth rate rather than modelled as a separate stream.

See also Net revenue retention and Organic growth rate.

Customers grouped by when they arrived, so that retention and expansion can be followed for each group separately. Cohort analysis separates a genuine improvement in retention from a change in the mix of customers being acquired, which blended churn cannot do. If recent cohorts retain better than older ones, the product is improving; if the blended number improves while cohorts do not, you have simply grown the denominator.

See also Revenue churn.

Cancellations caused by expired cards, insufficient funds and failed renewals rather than by dissatisfaction. It is a meaningful share of self-serve churn and the cheapest kind to recover: dunning sequences, card-updater services and pre-expiry notices convert directly into retained revenue without touching the product.

See also Revenue churn.

Ceiling = new MRR per month ÷ monthly churn rate

If a company adds roughly the same amount of new recurring revenue each month, its MRR does not grow forever — it converges on the point where the amount churning out equals the amount coming in. A team adding $15,000 of new MRR monthly at 5% churn asymptotes near $300,000 MRR regardless of how long it runs. Raising acquisition raises the ceiling; only retention removes it.

See also Revenue churn and Net revenue retention.

CAC = total acquisition spend ÷ new accounts won

Everything spent to acquire customers — paid media, sales salaries and commission, marketing headcount, content production, events and the tooling behind them — divided by the accounts won in the same period. Excluding sales payroll is the single most common way teams flatter their own unit economics. Blended CAC includes organically acquired accounts in the denominator; paid CAC counts only attributable ones. This simulator uses blended CAC, because that is what actually consumes cash.

See also LTV, CAC payback period and LTV:CAC ratio.

LTV = (ARPA × gross margin) ÷ monthly revenue churn

The gross profit a single account is expected to produce over its life. The gross margin term is not optional: revenue you never keep is not value, and two companies with identical ARPA and churn can differ in LTV by 40% on cost of revenue alone. Because churn sits in the denominator, LTV is inversely proportional to it — doubling churn halves lifetime value exactly.

See also CAC, Gross margin and Revenue churn.

How many times over a customer repays the cost of winning them. Below 1:1 every sale destroys value. Between 1:1 and 3:1 the business is usually paying too much for growth or losing customers too quickly to earn it back. Above 5:1 is not automatically good news — it frequently means acquisition is under-funded and reachable market is being left alone. Always read it beside payback period, because a 4:1 ratio with a thirty-month payback still starves the balance sheet.

See also LTV, CAC and CAC payback period.

Payback months = CAC ÷ (ARPA × gross margin)

The cash-flow view of the same relationship LTV:CAC describes in aggregate. The ratio tells you whether a customer is eventually worth acquiring; payback tells you how long you finance them before finding out. Best-in-class business software lands under twelve months. Beyond eighteen, growth is effectively debt-funded, and the runway projection will show it.

See also LTV:CAC ratio, CAC and Cash runway.

A way of comparing acquisition channels that return the same lifetime value on different timescales. A channel with a low CAC and a twenty-four-month payback consumes more working capital than one with a higher CAC that repays in eight, and at constrained runway the second is worth more despite the worse headline number.

See also CAC payback period and Cash runway.

Magic number = (net new ARR in quarter) ÷ (prior-quarter S&M spend)

A sales-efficiency measure expressed per dollar rather than per customer. Above 0.75 is generally read as a signal to invest harder; below 0.5 suggests the motion is not yet repeatable and more spend will not fix it. It answers roughly the same question as CAC payback but from the top down, which makes it easier to compute from a public income statement and blunter as a management tool.

See also CAC payback period and Burn multiple.

Gross margin = (revenue − cost of revenue) ÷ revenue

The share of revenue left after the costs of actually delivering the product: delivery hosting, support, payment processing, third-party data and API fees, and the customer-success effort that is delivery rather than sales. Research, development, sales and marketing sit below the line and do not belong in it. Gross margin is embedded in lifetime value and in every month of the projection, so misclassifying a cost here moves every downstream number.

See also Cost of revenue, LTV and Opex.

The costs that scale with serving customers rather than with building or selling the product. In software this is dominated by production hosting, support staffing and payment fees. The test is counterfactual: if this customer did not exist, would the cost disappear? Engineering salaries generally fail that test and belong in operating expenditure.

See also Gross margin and Opex.

The running cost of the organisation below the gross-profit line. In this simulator it comprises engineering payroll and contractors indexed by salary inflation, fixed cloud spend indexed by price inflation, variable cloud spend proportional to revenue, and the licence cost or amortised build cost and permanent maintenance of the chosen software path.

See also Gross margin and Fully-loaded cost.

The true annual cost of an employee, typically 1.25 to 1.4 times base salary once employer taxes, benefits, insurance, equipment, software seats and allocated overhead are counted. Financial models built on base salary alone understate payroll by a quarter or more, which over a five-year horizon is not a rounding error.

See also Opex and Salary inflation.

The rate at which the cost of the same headcount rises through raises, promotions and market re-levelling. It compounds against a cost base that does not shrink, so a five-year projection that holds salaries flat understates payroll materially by year three. Modelling it explicitly is what makes a long-horizon cost comparison honest.

See also Fully-loaded cost and Opex.

The portion of the hosting bill that tracks usage, and therefore roughly tracks revenue. Expressed here as cost per $1,000 of MRR, it is the part of infrastructure that makes growth cost money. Modelling only the fixed baseline understates the cost of scaling and makes high-growth scenarios look better than they are.

See also Fixed infrastructure cost and Gross margin.

The floor under the cloud bill — staging and development environments, monitoring, base compute, data stores that must exist regardless of load. It is largely insensitive to revenue and highly sensitive to architectural choices, and it is the part of the bill that hurts most before scale arrives.

See also Variable infrastructure cost.

The unit most business software is priced in. Seat counts track the whole organisation rather than one team: a platform bought for six engineers is invariably paid for by adjacent staff too, which this simulator models at roughly 1.8 seats per engineer. Licence spend therefore climbs with every hire, quietly, in a line nobody reviews monthly.

See also TCO and Fully-loaded cost.

How long the business survives without new capital. The back-of-envelope version divides cash by last month's burn and assumes burn stays flat, which it never does. This simulator instead projects sixty months with growth, churn, salary inflation, infrastructure scaling and re-investment all compounding against each other, and reports the last month in which the balance is still positive. The two numbers frequently differ by a year or more.

See also Net burn, Minimum cash and Break-even.

Net burn = cash out − cash in

The monthly decline in the cash balance. Gross burn is everything the company spends; net burn is what is left after collections, and it is the figure runway is computed from. A business can raise gross burn substantially while lowering net burn, provided the spending buys revenue quickly enough.

See also Cash runway and Burn multiple.

The trough of the projected cash curve. It is a more useful planning number than runway for a company that eventually turns profitable, because it states how much capital is actually required to get to the other side. A plan that survives on paper but dips to $40,000 in month nineteen has no tolerance for a slow quarter.

See also Cash runway and Break-even.

The month in which gross profit finally covers the operating cost base — the month the business stops losing money. It says nothing about whether the losses accumulated on the way there have been recovered, which is a separate and usually much later milestone.

See also Cumulative payback and Minimum cash.

The point at which everything earned since month zero offsets everything lost before break-even. A company can be profitable every month for two years and still be short of cumulative payback, which is why "profitable" and "has repaid its history" are different claims. Investors care about the first; a bootstrapped founder living off the balance sheet cares about the second.

See also Break-even and Cash runway.

Burn multiple = net burn ÷ net new ARR

A capital-efficiency measure that cannot be improved by spending more, which is what makes it a useful cross-check on growth rate. Under 1x is exceptional, 1x to 1.5x is good, 2x to 3x indicates growth is being bought rather than earned, and above 3x is unsustainable outside a deliberate land grab. A rising burn multiple alongside a rising growth rate is the classic signature of a business paying for revenue it cannot retain.

See also Net burn, Magic number and Rule of 40.

The lever that converts profit into growth. In this simulator a share of each month's positive operating profit is spent on acquisition and converts to new recurring revenue the following month at the stated CAC and ARPA; the remainder moves the cash balance. Raising it accelerates revenue and delays cash accumulation, which is why the revenue curve and the cash curve can move in opposite directions at once.

See also CAC, Cash runway and Burn multiple.

Rule of 40 score = revenue growth % + profit margin %

A shorthand for the trade-off between growing and earning: a company growing 60% while losing 20% of revenue scores the same as one growing 15% at a 25% margin. It is a statement about acceptable trade-offs rather than a target to optimise, and it is sensitive to which profit measure is used — free cash flow, operating margin and EBITDA give materially different answers for the same company.

See also Burn multiple and Gross margin.

The gap between spending on a customer and collecting from them. Annual upfront billing produces negative working capital, effectively financing growth from customers rather than investors, while monthly billing with net-60 enterprise terms does the opposite. It is invisible in a profit-and-loss statement and decisive in a cash projection.

See also Deferred revenue and Cash runway.

The ownership cost of raising capital. It is deliberately excluded from this simulator, which models the operating business rather than the cap table, but it is the reason capital efficiency has a value beyond survival: two companies reaching the same revenue can leave their founders with very different outcomes depending on how much equity was sold along the way.

See also Burn multiple and Cash runway.

The complete cost of a decision across its life: for a licensed product, the per-seat fee multiplied by a seat count that grows with headcount and drifts with price inflation; for an internal system, the one-time build plus the permanent maintenance it consumes forever. Comparing a build estimate against an annual licence fee, rather than both against the same horizon, is the most common way build-versus-buy analyses go wrong.

See also Maintenance FTE, Seat and Crossover point.

The full-time-equivalent engineering capacity an internal system consumes after it ships: dependency upgrades, incidents, security patching, cloud migrations and the feature requests that never stop. It is small, invisible and permanent, which is precisely why it is left out of business cases. At a $145,000 loaded salary, 1.2 FTE is roughly $174,000 a year, every year, rising with salary inflation.

See also TCO and Fully-loaded cost.

Crossover seats ≈ (maintenance FTE × loaded salary) ÷ (12 × price per seat)

Where the ongoing cost of maintaining an internal system equals the ongoing licence bill it replaces. Below that line no build estimate, however disciplined, changes the answer, because the run-rate comparison alone rules it out. Above it, the build cost becomes an amortisation question rather than a decision.

See also TCO, Maintenance FTE and Seat.

Recognising a large upfront spend gradually rather than all at once, so that a build appears as a monthly cost comparable to a licence fee. It makes the comparison fair and it flatters nothing: the cash still leaves in the build months, which is why a project can be amortisation-neutral and still be the reason the company ran out of money.

See also TCO and Cash runway.

The revenue-generating work an engineering team does not do while building an internal system. It rarely appears in a business case because it cannot be invoiced, and it is frequently larger than the build cost itself. A team of four spending nine months rebuilding a solved problem has spent nine months not building the thing customers would have paid for.

See also TCO and Maintenance FTE.

The cost of leaving a vendor once workflows, integrations and data models have been shaped around it. It is the honest argument for building, and it is a cost rather than a certainty: the right response is usually to price it — what would migration actually cost, and how likely is it — rather than to treat it as decisive on its own.

See also TCO and Crossover point.