Most metric problems are not measurement problems. The number is right and the decision is still wrong, because the metric was read on its own instead of against the one upstream of it.
A SaaS business is described by ten metrics. They are not a scorecard, they are a chain: demand becomes revenue, revenue survives retention, retention sets what acquisition may cost, and cash decides how long you have to get it right.
This guide is the map. Each section states what a metric decides, what breaks when it is read alone, and links to the full working guide for the calculation itself.
Why metrics work as a stack
Read in isolation, every metric can be defended. Churn of 3% looks tolerable until it is set against a twelve-month payback period, at which point the average customer leaves before paying for their own acquisition. Burn looks aggressive until it is set against net new revenue, and then it is simply the price of a working engine.
The order matters. Fixing acquisition cost before retention buys more customers for a leaking product. Forecasting revenue before standardising pipeline stages forecasts an opinion. Work the chain from the bottom: retention, then unit economics, then acquisition, then reporting.
Revenue: ARR and MRR
Recurring revenue is the base every other metric is expressed against. It is also the number most often calculated wrong, because multi-year contracts, one-time fees and services revenue all inflate it if they are not stripped out first.
Track the four movements separately - new, expansion, contraction, churned - or the total will hide the fact that growth is coming from discounts rather than demand. The ARR and MRR guide has the formulas and the movement table.
Retention: churn
Retention is the first thing to fix because it multiplies everything downstream. A monthly churn rate of 5% compounds to losing roughly 46% of the customer base a year, which forces sales to run at capacity just to stand still.
Measure customer churn and revenue churn separately: losing ten small accounts and one enterprise account are the same logo number and entirely different businesses. The churn rate guide covers both formulas, benchmarks by segment, and the usage signals that precede a cancellation by weeks.
Unit economics: LTV and CAC
Lifetime value sets the ceiling on what acquisition may cost, so the two are one metric read as a ratio. Below 3:1 the model is usually buying revenue it cannot keep; far above it, the company is under-investing in growth.
Payback period is the companion number, because a healthy ratio with a 24-month payback still starves the bank account. Start with the lifetime value guide, then the acquisition cost guide for what belongs in CAC and how to read it by channel.
Cash: burn and runway
Burn rate is the constraint on every plan above it. Gross burn says what the company costs to run; net burn says what it costs after revenue, and runway turns that into the only deadline that matters.
Judge burn against net new revenue rather than in absolute terms - the burn multiple is the honest version of efficiency. The burn rate guide has the calculation and the mistakes that flatter runway.
Demand: forecast and audiences
A forecast is a claim about demand, and it is only as good as the pipeline definitions underneath it. Standardise stage entry and exit criteria first, then choose a method appropriate to the data you actually have.
Where demand comes from is the other half. Behaviour and intent identify buyers that demographics never will, which lowers acquisition cost without touching the ad budget. See the sales forecasting guide and the high-converting audiences guide.
Product: usage and adoption
Usage is the leading indicator for everything financial. Adoption of the core action predicts retention, and a drop in usage precedes churn by weeks, which is time enough to intervene.
Audit features by usage and value rather than by roadmap sentiment. The product usage guide sets out the four-quadrant audit and the metrics worth instrumenting.
Reporting: KPIs and board packs
A metric nobody sees on time is not a metric. Automating collection is what makes the stack usable: the numbers arrive continuously, and the finance team spends its month analysing rather than reconciling.
Reporting then becomes narrative work instead of assembly. The financial KPI guide covers the automation pipeline, and the executive reporting guide covers turning it into something a board can read in one page.
All ten guides
Read in this order if you are building the stack from scratch. Read straight to a section if you already know which number is lying to you.
Automating the stack
Every guide here ends at the same place: the calculation is straightforward, and keeping it current by hand is not. That is the work worth automating, and it is also the work most companies cannot send to a public AI tool, because it means handing over billing, CRM and payroll data.
Moterra deploys Claude inside your own cloud tenancy, so the analysis runs where the data already lives and nothing leaves your environment. Private Claude Cowork is the workspace teams use; Moterra AI Bridge connects it to the billing, CRM and finance systems the metrics come from. The case studies show what the deployments produced.
FAQ
Which metric should we fix first?
Retention, in almost every case. It is the only metric that improves the others without spending anything: lower churn raises lifetime value, which raises the acquisition cost you can afford, which relaxes the constraint on burn.
How many metrics should a board pack carry?
Five to ten, reported against target with a stated reason for each variance. Teams that track fifty metrics are usually reporting activity rather than performance.
How often should these be reviewed?
Revenue, churn and cash weekly; unit economics and product usage monthly; forecasts continuously, since a forecast reviewed once a quarter is a target with a date on it.
Can AI calculate these metrics for us?
Yes, and the constraint is deployment rather than capability. Connected to billing and CRM data, an AI analyst maintains the whole stack and explains the movements; the question worth asking a vendor is where that data goes. Talk to us about running it inside your own tenancy.
