SaaS analytics
SaaS analytics shows whether revenue grows from customers who stay and expand, which onboarding step decides who pays, and how long a new customer takes to pay back. It starts from a monthly MRR movement rebuilt from billing, and closing MRR has to agree with billing before any other figure is read.
MRR movement, example
Net retention 101.0% hides £10,000 lost: gross retention 95.0%
| Movement | Accounts | MRR | |
|---|---|---|---|
| Opening MRR, 1 October | 400 | £200,000 | |
| New | +25 | +£14,000 | |
| Expansion | 31 accounts | +£12,000 | |
| Contraction | 9 accounts | −£3,000 | |
| Churn | −14 | −£7,000 | |
| Churn: 14 accounts whose MRR fell to zero in October | |||
| Closing MRR, 31 October | 411 | £216,000 | |
| Closing: agrees with active subscriptions in billing on 31 October | |||
Retention, from the rows above
- Net revenue retention
- 101.0%
(£200,000 + £12,000 − £3,000 − £7,000) ÷ £200,000- Gross revenue retention
- 95.0%
(£200,000 − £3,000 − £7,000) ÷ £200,000
SaaS metrics: the numbers that run a SaaS business
Net and gross revenue retention always sit side by side, so expansion can't hide the accounts that left.
| KPI | Formula | A good reading | The common mistake |
|---|---|---|---|
| MRR and its movements | Opening MRR + new + expansion − contraction − churn = closing MRR, where MRR is each active subscription's monthly price and annual contracts count at one twelfth |
Closing MRR agrees with billing's active subscriptions every month, with nothing left unexplained | Booking an annual contract's full value in the month it's signed, which shows a spike that month and nothing for the next eleven |
| Net revenue retention | (Opening MRR + expansion − contraction − churn) ÷ opening MRR, for the accounts paying at the start. Monthly here. The annual version compares the same accounts twelve months apart. |
Above 100%, read with gross revenue retention beside it | Adding new accounts' MRR to the top line, which turns a retention rate into a growth rate |
| Gross revenue retention | (Opening MRR − contraction − churn) ÷ opening MRR |
Steady or rising. It can't go above 100%. | Setting one account's upgrade against another's downgrade before counting, which lets expansion back into a figure built to leave it out |
| Logo churn | Accounts that stopped paying in the month ÷ accounts paying at the start of the month |
Read by plan size, beside revenue churn, so the loss of many small accounts and of one large one both show | Counting an account that moved to a free plan as retained. It no longer pays, so it has churned. |
| Activation rate | New accounts that did the agreed activation event within the agreed window ÷ new accounts that signed up in the cohort |
Rising by sign-up cohort, on an event shown to separate accounts that go on to pay | Using a login as the event. Every account logs in, so the rate says nothing about who will pay. |
| CAC payback | Sales and marketing cost to acquire in the period ÷ (new MRR won in the period × gross margin %), in months. This row works on gross margin. Some companies work it on revenue. |
Shorter than the typical contract term, read for self-serve and sales-led deals apart | Counting ad spend alone and leaving out sales salaries, commission and tools, which shortens payback for a sales-led product |
| Expansion share | Expansion MRR ÷ (new MRR + expansion MRR), in the month |
Rising from upgrades customers chose, with any price rise on its own line | Counting a list price rise applied to every customer as expansion, when no customer chose to buy more |
| Trial to paid conversion | Trials from a start-date cohort that became paying within the agreed window ÷ trials started in that cohort |
Read by cohort and split by whether the account activated | Dividing this month's new paying accounts by this month's trial starts, when some of the payers started their trial last month |
There are more examples in the KPI library.
Expansion from the accounts that stayed can cover for every account that left.
Where the data for a SaaS dashboard usually lives
Analytics services for software companies
The MRR ledger is rebuilt from billing, and data monitoring and support checks each day that the billing and product event feeds reached the warehouse, so a partial export shows up before anyone reads the month.
In the example, 14 of the 400 accounts paying on 1 October left, a logo churn of 3.5%. They paid £500 a month on average, the same as the whole base.
AI in SaaS analytics
The AI-led analytics demo runs on an invented software company's onboarding events. Ask it where people drop out and it shows each step, the query it ran and the table it read. On your data the same question needs activation defined first.
Questions about SaaS metrics
What should a SaaS dashboard show first?
The month's MRR movement from opening to closing, with closing checked against billing, then net and gross revenue retention from the same rows. Activation and trial to paid conversion by sign-up cohort come next.
How is SaaS analytics different from subscription and media analytics?
A SaaS account can grow and shrink inside the month through seats, plan changes and contracts, so its revenue retention needs expansion and contraction rows. The subscription and media analytics page covers what moves a consumer subscription.
How do you count annual contracts in MRR?
At one twelfth of the annual value each month, from start date to end date, however the customer paid, and a three-year deal at its yearly value ÷ 12. Setup fees, training and usage overage stay out of MRR.
How do you choose the activation event?
The analysis compares how new accounts go on to pay by the events they did in their first days, and picks one that separates payers and that product can act on. We agree it with product and finance, with its time window, and keep it for at least a quarter.
Should net revenue retention be monthly or annual?
Either, if you say which. The monthly figure here is one month's movement on the accounts paying at its start, and the annual one compares the same accounts twelve months apart. Compounding the monthly figure won't match the annual, because accounts join during the year.
Does your closing MRR agree with billing?
On the call we rebuild one month's movement from your billing exports and check it against your closing MRR.