Gross Revenue Retention

In a company where revenue rises every month, leaks are hard to see. As long as new customers arrive slightly faster than existing ones drain away, the total keeps climbing. One number holds both what sales brought in and what the company managed to keep, and from that number alone you cannot tell which half is working.

Pagedeck was in exactly that spot. Grace put an up-and-to-the-right chart in front of the board, and one investor asked what the customers under contract a year ago were paying now. Nobody in the room could answer. Pulling that list a few days later showed that accounts paying $75,000 a month back then were paying $61,500 now. While the total climbed, the layer underneath it was shrinking.

Gross revenue retention (GRR) exists to answer that question. New customers are excluded from the calculation entirely, and only what existing customers still pay is counted. The rule goes one step further and excludes increases too. With nothing but losses left in the number, the size of the leak can't be covered up by anything else.

Put what last year's customers were paying in the denominator and that figure minus what cancellations and downgrades took away in the numerator. It runs in four steps.

1. Pick a reference point and freeze the list of customers who existed then, usually the same month 12 months earlier. Anyone who signed after that drops out of the calculation completely.
2. Add up the starting amount for that list. Say it comes to $75,000 in monthly recurring revenue (MRR), the revenue that arrives again every month.
3. Subtract what was lost, which comes in two forms. Churned MRR, revenue that vanished because the customer left, accounts for $3,750, while downgrades to cheaper plans and contraction in seat counts together account for $2,250.
4. Divide what remains by the starting amount. $69,000 over $75,000 is 92 percent.

If customers on that same list added seats or moved to higher plans and paid $11,250 more, none of it enters the calculation. Because there is nothing to add, the number cannot pass 100 percent, and a result above 100 means the math is wrong. That ceiling is the whole point. However many new customers arrive and however much the remaining ones spend, the leak stays visible.

The window is usually 12 months. Measured monthly, the number swings on whether one large customer's renewal falls in this month or the next, which makes the trend hard to read.

The bar depends on who you sell to. Companies selling to small teams do fine in the 85 to 90 percent range, while companies selling to enterprises look for 95 percent and up. Small teams leave for reasons that have nothing to do with the product, like going out of business or losing the person who championed it, so the baseline leak is larger. Enterprise accounts can be worth $75,000 a year on their own, so losing one costs several percentage points at once.

Below 80 percent reads as a warning. It means losing a fifth of the base every year, and five years at that pace leaves a third of what the original customers were paying. Filling that hole with new customers means sales has to run every year just to end up where it started.

Breaking the number apart matters more than the number itself. The same 92 percent means different work depending on whether the missing eight points came from cancellations or from downgrades. A cancellation means the customer left the product and has to be sold again; a downgrade means they stayed and use less, which calls for widening what they use.

Setting it beside logo churn, the same loss counted in customer accounts, adds another layer. Revenue at 92 percent against accounts at 80 percent means a lot of small customers are leaving. Revenue at 85 percent against accounts at 95 percent means one large customer walked. The first sends you to the onboarding process, the second to that one account's renewal conversation.

"GRR is 88 percent. Our NRR of 105 was hiding it, but we're leaking underneath."

Total revenue is growing while existing customers shed 12 percent of their spend a year.

"The drop this quarter wasn't cancellations, it was seat reductions."

The customers are still here, so the fix is widening usage rather than selling them again.

"Can you pull GRR for enterprise accounts on their own?"

A request to split the number because the blended average is hiding differences by customer size.

Pagedeck started measuring this monthly after the board meeting. The first three months came in at 92, 90 and 88 percent, and two thirds of the missing revenue was seat reduction rather than cancellation. Contracts were holding while the number of people actually using the product fell.

Sofia called the twelve accounts that had shrunk, and eight of them said the same thing: they had bought seats for a whole department at signing, and only two or three teams ever used the product. Starting the next quarter, a check on actual headcount went into month two of every contract, and six months later the number was back at 91 percent.

The most common mistake is reading it as a number that can exceed 100 percent, since net revenue retention sits right next to it and produces values like 110 percent. But this calculation has no additive term, so 100 is the ceiling. Mixing in expansion revenue, the extra revenue that comes from existing customers spending more, inflates the figure, and an investor recomputing it on the standard definition after a board meeting will not get the same answer. Companies genuinely do count the same metric differently, so it pays to write down in one line what was excluded whenever you hand the number over.

Reading a high number as proof of a good product is nearly as common. Customers stayed, so they must be happy. But contract structure alone raises this number. With a three-year commitment and an early termination fee attached, a customer who wants to leave stays inside the figure for the length of the term. A high value can be lock-in rather than satisfaction, and in the year those commitments expire together, several accounts leave at once and 95 percent drops into the seventies in a single step. Measuring customers with similar contract structures as their own group surfaces that risk in advance.