Why the numbers agree

Every headline SaaS metric is a different view of one book of business, so they have to reconcile or the dashboard is lying. Rather than generating each series on its own, the generator walks a real book day by day — each subscription with a plan, an MRR, a signup date and maybe a churn date — and derives everything from it. Two identities are asserted before the data is written:

• ending MRR = starting MRR + new + expansion − contraction − churn
• active customers = yesterday's + new − churned

So the waterfall below balances exactly, and the cohort table shows the same churn the daily series reports.

Revenue metrics

• MRR — monthly recurring revenue at the end of the range. ARR is MRR × 12.
• MRR movement — New (first payment), Expansion (seats or upgrades on existing accounts), Contraction (downgrades), Churn (cancellations). Net New MRR is new + expansion − contraction − churn.
• ARPA — MRR ÷ active customers.
• NRR (net revenue retention) — what the starting book is worth now, including expansion: (start + expansion − contraction − churn) ÷ start. GRR (gross) excludes expansion, so it can never exceed 100%.
• Quick ratio — MRR gained ÷ MRR lost. Above 4 means growth comfortably outruns leakage.

Customer & funnel metrics

• Trial → Paid — trials that converted ÷ trials started. Trials run 14 days, so a conversion is credited on the day it converts.
• Logo churn — accounts lost ÷ accounts at the start. Revenue churn — MRR lost ÷ MRR at the start.

Watch those two apart. They answer different questions, and when one large account leaves they diverge hard — which is exactly what happens here on 10 June.

Unit economics

• CAC — sales & marketing spend ÷ new customers.
• LTV — ARPA × gross margin × expected lifetime, assuming an 80% gross margin and a lifetime of 1 ÷ monthly churn capped at 36 months. The cap matters: Enterprise churns at 0.7%/month, which implies a twelve-year customer and prints an LTV:CAC of 47x. Nobody underwrites that, so it is capped — the usual convention.
• LTV : CAC — 3x or better is generally considered healthy. CAC payback is the months of gross profit needed to repay CAC.

Rates are per month

The default range is 90 days. A churn figure that silently meant "per quarter" would be the most misleading number on the page, so every rate here is normalised to a month. NRR and GRR are the exception — they are quoted over the whole selected range, as is conventional.

Stocks vs. flows

MRR and active customers are stocks — a level at a point in time. They are read at the end of each period, never summed across days (adding "active customers" over 90 days would report 160,000 customers for a company that has 1,800). New MRR, churn, trials and spend are flows and do accumulate.

Cohort retention

Rows are signup months, columns are months since signup. Toggle between logo and MRR retention — the gap between them is the story. Logo retention can only fall; MRR retention can pass 100% when expansion inside the surviving accounts outruns what churned. The triangle covers the whole book back to August 2025 and deliberately ignores the date and plan filters, since slicing a cohort curve to 90 days would leave a single meaningless column.

What the data is built to show

• 10 June — two Enterprise accounts churn. Logo churn barely moves (8 accounts that day out of ~1,750), revenue churn spikes to many times its usual level. The argument for tracking both, in one chart.
• 6 July — self-serve onboarding revamp. Trial → Paid steps up from ~18.6% to ~24.3% with no change in trial volume.
• Revenue concentration. Enterprise is ~5% of customers and ~46% of MRR. The plan-mix table is where that shows up — and why the Enterprise churn event hurt.
• NRR sits just under 100%. That is the honest consequence of a Starter-heavy book: expansion almost, but not quite, covers churn, so growth has to come from new sales. The plan table shows which plans are above water.