Gross revenue retention

Gross revenue retention (GRR) is the percentage of recurring revenue a business keeps from its existing customers over a period, before any revenue from upgrades or expansion is counted. If a company starts the year with 1,000,000 pounds in recurring revenue from its existing base and loses 120,000 pounds to cancellations and downgrades, GRR is 88 percent. Crucially, GRR can never exceed 100 percent. It only ever measures what was lost, never what was gained.

That ceiling is what makes GRR so revealing. Unlike net revenue retention, which adds expansion revenue back in and can rise above 100 percent, GRR strips expansion out entirely. A business with strong upsell can post net revenue retention well above 100 percent while quietly losing a fifth of its base to churn each year. GRR exposes that leak because no amount of expansion can hide behind it. It answers a single, blunt question: of the revenue we had, how much did we keep.

GRR is built from the same movement components as MRR, but it counts only the downward ones. It depends on churn rate and contraction, and it sits at the centre of any retention analysis. Investors lean on it heavily because it measures the durability of revenue. High GRR means revenue is sticky and predictable. Low GRR means the business has to keep refilling a leaking bucket just to stand still.

GRR takes the recurring revenue you started the period with, subtracts everything lost to churn and contraction, and expresses the remainder as a percentage of the starting figure. Expansion is deliberately left out. The inputs below are what the calculation needs, and getting their boundaries right is what separates a meaningful GRR from a misleading one.

1. 1

   Starting MRR or ARR

   The recurring revenue from your existing customer base at the start of the period. This must exclude any new customers acquired during the period, because GRR measures retention of the base you already had, not the growth on top of it.

2. 2

   Churned revenue

   Recurring revenue lost from customers who cancelled entirely during the period. This is the most damaging loss because it is usually permanent and removes the chance of future expansion from that account.

3. 3

   Contraction revenue

   Recurring revenue lost from customers who stayed but downgraded their plan, reduced seats, or cut usage. Contraction is a softer loss than churn but an important early warning that customers are deriving less value.

4. 4

   Exclude all expansion

   Upgrades, seat additions, and cross-sell revenue are left out entirely. Including them would turn the calculation into net revenue retention and hide the very losses GRR exists to surface.

A worked example makes the boundaries clear. Start with 1,000,000 pounds of recurring revenue from existing customers. During the year, 80,000 pounds churns and 40,000 pounds contracts, while 150,000 pounds of expansion comes in from upsells. GRR ignores the expansion completely: (1,000,000 minus 80,000 minus 40,000) divided by 1,000,000, which is 88 percent. Net revenue retention on the same base would be 103 percent once the expansion is added back. The 15 point gap between the two numbers is the whole story, and it is invisible if you look at net revenue retention alone.

A GRR of 88 percent tells you 12 percent leaked out, but not where the leak is or whose job it is to plug it. A metric tree decomposes the lost revenue into its causes, so a falling GRR points to a specific, ownable problem rather than a general sense that retention is slipping.

The first level splits the loss into its two components, churn and contraction, because they behave differently and call for different responses. Churn then decomposes into voluntary churn, where customers actively choose to leave, and involuntary churn, where revenue is lost to failed payments and expired cards despite the customer wanting to stay. Voluntary churn decomposes further into reasons: competitive loss, lack of value realised, and budget cuts. Contraction decomposes into seat reductions and plan downgrades, each tracing back to how customers are using the product.

This structure turns retention from a number into a diagnosis. If GRR is falling, the tree shows whether the cause is customers leaving for a competitor, customers quietly downgrading because they are not using what they bought, or simply cards expiring unnoticed. Each branch has a different owner. Customer success owns value realisation, product owns adoption, and billing owns the involuntary losses that are often the easiest of all to recover.

GRR benchmarks vary by customer segment more than by company stage, because the size and commitment of the customer drives how much revenue leaks. Enterprise revenue is stickier than small business revenue, which churns faster and contracts more readily. The ranges below reflect annual gross revenue retention by the segment a business primarily serves.

Read GRR alongside net revenue retention to understand the shape of the business. A company with 85 percent GRR and 120 percent net revenue retention is leaking meaningfully but masking it with strong expansion, which works until expansion slows. A company with 95 percent GRR and 105 percent net revenue retention has a far more durable base, even though its headline net retention is lower. The gap between the two numbers is the single most useful retention signal there is.

Improving GRR means reducing the revenue that leaks out, which is a different discipline from growing it through expansion. Because GRR only counts losses, every lever here is about preventing churn and contraction rather than adding new revenue. The branches of the tree point directly to where the effort should go.

The metric tree approach starts by finding which loss branch is largest. If involuntary churn dominates, billing fixes will move GRR fastest. If voluntary churn from poor value realisation is the issue, the work sits with onboarding and customer success. If contraction is rising, the question is whether customers are buying more capacity than they use.

KPI Tree connects each branch of the retention tree to the team that owns it. Billing owns involuntary churn, customer success owns the early-warning signals behind voluntary churn, and product owns the adoption that drives value realisation. When GRR moves, the platform pushes a signal to the accountable owner of the branch responsible, rather than leaving the drop to surface in a quarterly board pack. The verified impact loop then checks whether the retention action actually slowed the leak, so effort is judged on whether GRR recovered, not on whether the work was done.

1. 1

   Letting expansion creep into the calculation

   The most common GRR error is including upsell or expansion revenue, which lifts the number above what was truly retained. The moment expansion is counted, GRR becomes net revenue retention and the leak it exists to expose disappears.

2. 2

   Including new customers in the base

   GRR measures retention of the customers you started the period with. Adding new customers acquired during the period to the base distorts the denominator and overstates retention. Measure the starting cohort only.

3. 3

   Reporting only net revenue retention

   A healthy net retention number can hide serious churn behind strong expansion. Without GRR alongside it, a leaking base looks fine until expansion slows. Always report the two together and watch the gap.

4. 4

   Lumping contraction in with churn

   Downgrades and cancellations are different problems with different fixes. Folding contraction into churn hides the signal that customers are reducing rather than leaving, which calls for a different response.

5. 5

   Ignoring involuntary churn

   Treating all churn as customers who chose to leave misses the portion lost to failed payments and expired cards. This is often the easiest revenue to win back, and missing it leaves the cheapest GRR gains on the table.