By sector
SaaS diligence
How to read a subscription software business: retention, expansion, payback and the gap between reported ARR and revenue you can rely on.
Realistic time Six to ten hours
Software subscriptions are the best-understood business model in venture, which is both an advantage and a trap. The metrics are standardised enough that founders know which ones to present, and standardised enough that a practised reader can tell within twenty minutes whether the numbers describe a compounding business or a leaky one.
The central question in SaaS is whether the customer base grows on its own. A business where existing customers spend more each year needs to acquire fewer new ones to grow, and that property compounds into the kind of outcome venture requires. A business where existing customers churn must buy all its growth, and the cost of buying growth rises.
The second question is how the revenue is defined. Annual recurring revenue is a convention rather than an accounting standard, and companies include very different things in it. Establishing what is actually inside the ARR number is often the most productive hour of SaaS diligence.
01
What is actually inside the ARR
Annual recurring revenue should mean contracted, recurring subscription revenue from customers who are live and paying. In practice it frequently includes annualised monthly revenue from customers on rolling contracts, pilot revenue that has not converted, one-off implementation fees, and signed contracts that have not started.
None of those inclusions is dishonest and all of them change what the number means. A company reporting $1.2m of ARR where $300k is implementation services and $200k is unconverted pilots has roughly $700k of the thing that investors are pricing.
Ask for the definition, then ask for the bridge from the previous period: opening ARR, plus new, plus expansion, minus contraction, minus churn, equals closing ARR. That single table answers more questions than any other document in a SaaS pack.
Check
- Written definition of ARR, and what is included and excluded.
- The ARR bridge: new, expansion, contraction and churn, by period.
- Proportion of revenue on annual versus monthly contracts.
- Services and implementation revenue, separated out.
- Revenue from pilots or trials that has not converted to a contract.
02
Retention and expansion
Gross retention measures how much revenue you keep from existing customers before any expansion. Net retention adds upsell and expansion back in, and can exceed 100% where customers grow their spend. The gap between the two tells you whether expansion is masking a churn problem.
Logo retention — the proportion of customers retained regardless of value — matters most in businesses selling to small companies, where a few large accounts can hide the loss of many small ones. Ask for both revenue and logo retention, because a company reporting only one is usually reporting the better one.
Cohort charts are the honest form. Aggregate retention figures blend cohorts of very different quality, and a company whose recent cohorts are worse than its early ones can show a flattering blended number for a surprisingly long time.
Check
- Gross revenue retention and net revenue retention, separately.
- Logo retention alongside revenue retention.
- Cohort charts by month or quarter, not blended aggregates.
- Are recent cohorts retaining better or worse than early ones?
- What is the stated reason for the last five churned accounts?
03
Acquisition cost and payback
Customer acquisition cost is straightforward to calculate and easy to understate. The honest version includes all sales and marketing costs — salaries, commission, tooling, agencies — divided by new customers acquired in the same period, with a sensible lag between spend and signature.
Lifetime value is much less reliable, because it requires assuming a customer lifetime that has not been observed. At seed and Series A, prefer payback period: how many months of gross profit from a customer are needed to repay the cost of acquiring them. It uses only observed data and it is much harder to flatter.
Then look at the trend. A payback period that has lengthened over four consecutive quarters means the cheap acquisition has been used up, and the company is now buying growth at rising prices.
Check
- Fully loaded customer acquisition cost, including salaries and commission.
- Payback period in months, and its trend over the last four to six quarters.
- Gross margin, and what is in cost of revenue.
- Proportion of new revenue from inbound versus outbound versus partnerships.
- Sales cycle length and whether it is lengthening.
04
Product and technical questions
You do not need to audit the code, but you should understand what would break the business technically. The usual candidates are a dependency on a single platform or API, infrastructure costs that scale faster than revenue, and a product whose value depends on integrations the company does not control.
AI-heavy products deserve a specific question about inference cost. A product whose gross margin depends on model pricing set by a third party has a cost structure it does not control, and margins in that position have moved substantially in both directions.
Check
- What single external dependency would most damage the product if it changed?
- How do infrastructure and inference costs scale with usage?
- Are there integrations the business depends on but does not control?
- What is the engineering team's view of the largest technical debt?
- Any security incidents, and how were they handled?
Stop and think
Red flags
- ARR that includes one-off services revenue or unconverted pilots without disclosure.
- Net retention presented without gross retention.
- Blended retention figures with no cohort detail available.
- A payback period that has lengthened for four consecutive quarters.
- Gross margin below the sixties in a business described as pure software.
- Reluctance to provide the ARR bridge.
- Growth driven by a single channel whose cost per acquisition is rising sharply.
The other side of the table
What founders are taught about this
Ilya Volodarsky - Analytics for Startups
What early companies actually measure and how the measurements are constructed — background for reading a cohort chart critically rather than accepting the version you are shown.
Published by Y Combinator. Channel verified from ycombinator.com, and YouTube's own oEmbed response names that channel as this video's author. Pressing play loads content from YouTube.
Take these into the room
Questions to ask the founders
- How exactly do you define ARR, and what is in it?
- Show me the ARR bridge for the last four quarters.
- What is gross retention, before expansion?
- Are your newer cohorts retaining better or worse than your first ones?
- What is your payback period, and how has it moved?
- Which single dependency would hurt most if it changed its terms?
- Why did your last five churned customers leave?
From the decoder