Ask most companies what a customer could buy next and you get a list of ways to raise the subscription. A higher tier, an add-on, a second product on the same invoice.
That is one revenue model. There are at least six, and the one you choose decides whether the customer can buy the thing at all.
The Models Available to You
Subscription. A tier increase, an add-on module, an adjacent product billed on the same cycle. The default, and frequently the only one anybody considers.
Capacity. More of a bounded resource: storage, seats, projects, campaigns, environments, volume. The most milestone-legible model there is, because the customer hits the limit and knows it before you tell them.
Credits. Prepaid units drawn down against usage. Common for compute, AI features, anything consumed in bursts the customer cannot forecast.
Services, recurring. Ongoing advisory, managed delivery, a standing engagement billed on a cycle. For customers who have the tool and lack the capacity or the expertise to run it. Carries real delivery cost, so it moves revenue LTV more than it moves margin LTV.
Services, project-based. Implementation, migration, a defined engagement with an end date. Often the highest-value item available and the one nobody prices.
Transactional. The customer buys a discrete quantity when they want it, with no ongoing commitment. The model most often missing entirely.
Three of Those Do Not Recur
Credits, project services, and transactional purchases are real revenue that arrives without a subscription attached.
Which means they do not appear in NRR. They do not appear in ARR. They do not appear in any expansion report built on recurring revenue.
So a company can be expanding genuinely, in dollars that clear the bank, and have the entire category be invisible to everyone reading the numbers. Not undercounted. Absent.
Where This Turns Into Churn
Consider a product whose value can be extracted in a single sitting. The customer signs up, gets what they came for, and has no reason to still be there next month.
They cancel. Not because anything failed. Because the thing they wanted was a purchase and you sold them a subscription.
Now take the customer who genuinely does want it every month, and this month wants more than their plan allows. Their only available move is to upgrade, take the extra, and downgrade again. Some of them do that. Some forget to downgrade, notice the larger invoice, and ask for a refund or file a dispute. Some skip the whole exercise and simply do not buy the thing they wanted.
The product worked exactly as promised. The revenue model was the wrong shape for the demand, and the reporting cannot tell the difference.
Every one of those outcomes lands in the numbers as a problem. The cancellation is churn. The downgrade is contraction. The dispute is a support ticket and a chargeback. And the review that follows will look at satisfaction, at onboarding, at the product, and find nothing, because there is nothing wrong with any of them. Not every cancellation is a verdict on the product, and this kind is a verdict on the order form.
Manufactured Churn Has a Tell
You can usually spot it without a survey.
Look for customers who cancel without complaint, having used the product exactly as intended. Look for repeated upgrade-then-downgrade cycling on the same accounts. Look for customers who return months later, sign up again, extract again, and leave again. That last pattern is not a win-back. It is a customer buying transactionally through a subscription form, because you did not offer them the form they wanted.
None of that is a satisfaction problem. It is a pricing model that does not match the shape of the demand, and it will not respond to anything you do about onboarding.
Adding a Model Is Expansion Work
The fix is not always to make the subscription stickier. Sometimes the honest answer is that the demand is episodic and always was, and the business has been fighting that rather than pricing for it.
Offer the transactional option and two things happen at once. The customers who were subscribing and cancelling now buy repeatedly instead, which raises what they are worth over the relationship even though nothing recurring is attached. And the customers who wanted more this month buy more this month, without touching their plan, without cycling, without a dispute.
The revenue goes up and the churn goes down, and neither of those came from a retention program.
This Is the Same Argument as Unbundling
Strategic unbundling is about when you sell something: hold it back, attach it to the milestone that earns it, price it at the moment the customer can use it.
Revenue model is about how you charge for it.
Both fail in the same way. You take a customer whose need has a particular shape, force it into the shape your billing system prefers, and then read the resulting damage as a fact about the customer. Sold too early, it comes back as contraction. Sold in the wrong form, it comes back as churn.
What to Do With This
Take your expansion inventory and put a revenue model next to every line. Then look at the distribution.
If every item is a subscription increase, you do not have an inventory. You have re-read your price list. Services, capacity, credits, and one-off purchases are all expansion revenue, and not one of them requires the customer to raise their recurring commitment.
If nothing in the list can be bought without increasing that commitment, ask what your customers currently do when they want more but not permanently. They are doing something. It is probably showing up in your churn number.
Next Steps
The metric definitions, including which numbers can and cannot see non-recurring expansion revenue, are on the expansion metrics reference.
The number underneath all of it is what your existing base should be producing and is not. That is latent revenue, and finding roughly what yours is takes about ninety seconds.
Lincoln Murphy formally named and popularized Customer Success starting in 2010 and has spent 15 years connecting it to expansion revenue and commercial outcomes. Read The Premise.