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.
There are also various purchase types:
Prepaid credits create timing gaps. A large credit purchase shows as a revenue spike, but consumption might stretch across months. The revenue gets recorded immediately while the actual usage (the signal that indicates ongoing engagement) happens gradually.
Promotional purchases compress timelines. A "buy 12 months, get 3 free" offer turns 15 separate transactions into one. Your billing data shows a spike followed by apparent inactivity, even though the customer is actively using the product. Usage data reveals what's actually happening.
Enterprise contracts can layer both patterns together. A customer commits to annual spend with subscription-style predictability, but consumes based on project cycles that look like pay-as-you-go usage. You need both signals: the contract value for forecasting, the consumption pattern for understanding adoption.
Separating revenue streams lets you see not just how much you earned, but how customers are actually engaging and what that means for retention and growth.
Don't mistake noise for signal
A customer who purchased heavily last month and nothing this month might be exactly as engaged as before, just working through prepaid credits. The inverse is also true: an annual subscriber who depletes their entire balance in a single day looks healthy in the billing record right up until they cancel. Their subscription renewal was on schedule; their consumption pattern signalled the problem weeks earlier.
The month-to-month variations in hybrid models are usually timing artifacts, not changes in behavior. The structural causes are predictable:
- Promotions compress future purchases into single events.
- New customers move through onboarding at different speeds.
- Project-based usage creates natural spikes and valleys.
Short-term numbers tell you almost nothing.
Cohort analysis shows how behavior develops as customers mature with your product. Rolling windows smooth out promotional artifacts and seasonal patterns. You start seeing directional trends instead of reacting to individual monthly fluctuations.
Growth happens in dimensions, so track separately
Revenue growth can come from entirely different places, and each source tells you something distinct about the business:
Take a common upgrade path. A customer starts on pay-as-you-go, tops up three or four times as their usage grows, then converts to a subscription once the flat monthly rate makes economic sense. After converting, they supplement the subscription with additional top-ups when they exceed their monthly allotment. In a billing report, they appear as a PAYG purchaser, then a subscriber, then a hybrid — four distinct revenue events that standard MRR treats as unrelated. Consumption tracking sees a single expansion arc.
These patterns describe different dimensions:
- Volume: How much customers consume or purchase
- Predictability: How much revenue is recurring versus variable
- Margin: Which revenue streams are most profitable
- Durability: Whether customers are embedding the product deeper into workflows
A business growing primarily through usage expansion operates differently than one growing through subscription upgrades, even when total revenue looks the same. Understanding where growth originates and how these dimensions interact reveals more than aggregate numbers alone.
Your data structure determines what you can actually measure
If you can get the definitions right upstream, the dashboard will mostly take care of itself.
Most teams, however, don't make those definitions explicit until after the reporting infrastructure is already built. By then, billing systems, payment processors, usage logs, and campaign tools are operating independently with revenue in one system, usage in another, customer attributes in a third.
The decisions that matter most happen upstream of any dashboard. Do you track revenue by transaction or by customer? By product or by feature? By subscription versus usage? These choices determine what questions you can answer later, and they're harder to change once your reporting infrastructure is built on top of them.
How you structure account data alongside transaction records is one of those decisions. Companies that prioritize data coherence before analytics make metrics easier to interpret downstream.
That applies to enrichment data as well: when billing records are joined to structured account data with a consistent schema, patterns like consumption by segment or revenue concentration by company size become visible instead of requiring manual lookup.
Revenue, churn, and growth all shift meaning depending on which customers you're counting, which signals you're reading, and whether your data connects them cleanly enough to tell the difference.
Coherence has to be built upstream, before the reporting layer exists.
