SaaS metrics work fine for simple subscription businesses, but the moment you add usage-based pricing, prepaid credits, or on-demand purchases, the standard playbook starts to dissolve.
Let's consider two different customer types:
Subscription customers consume value through time, and access continues as long as the subscription stays active.
Usage-based and credit-based customers consume value through activity, running through their balance in a week or stretching it across six months. The timing of consumption tells you something entirely different than the timing of payment.
This dual scenario creates an interpretive challenge that pure subscription businesses don't face. A customer who hasn't purchased anything in three months might look dormant in your billing system while actively consuming prepaid credits at a steady rate. The billing data says they're disengaged, but the usage logs say otherwise.
Which signal tells you what you need to know? It depends on the question you're asking. Metrics that track actual consumption patterns reveal more about customer health than a binary active/inactive flag based on a customer's subscription renewal.
Revenue structure tells different operational stories
If consumption and payment don't always align, how do you track what actually matters?
The standard approach treats revenue as a single number. Add up subscriptions, usage fees, one-time purchases, prepaid credits, and call it MRR or total revenue. But when different customer types behave in fundamentally different ways, collapsing it all into one number makes interpretation impossible.
Subscription customers give you clean signals. They pay a predictable amount each month, and as long as they don't cancel, you can reasonably forecast that revenue continuing. Standard churn metrics work because cancellation is a clear event.
Pay-as-you-go and credit-based customers introduce a tracking problem. There's nothing to cancel, so you watch activity windows instead. What counts as dormancy depends on the plan: as a general rule, a monthly subscriber with no usage in seven days is worth flagging; for an annual subscriber, the threshold is closer to thirty. At sixty days, churn probability is typically high regardless of plan type. At ninety, they have almost certainly left, whether or not they ever formally cancelled.