How much of your growth did you have to buy?

Opening ARR

Contracted recurring revenue, annualised, at the start of the period. One-off implementation fees, training and professional services are not ARR and should not be here.

New ARR

From customers who were not there at the start. Strictly new logos, not upgrades.

Expansion ARR

Upgrades, seats added and price rises on customers you already had. This is the line that decides whether the business compounds.

Contraction ARR

Downgrades and seats removed by customers who stayed. Easy to overlook, because nobody reports a partial loss.

Churned ARR

The full ARR of customers who left. Count the ARR they were paying when they left, not what they paid at the start.

Opening plus new plus expansion, less contraction, less churn, equals closing ARR. If it does not, one of the five is wrong, and it is almost always because an upgrade was counted as new business.

Illustrative figures for a fictional company, over twelve months. Every number is invented.

A 32% growth year, which most founders would be pleased with. Underneath it, gross retention of 82.8% means nearly a fifth of the opening base disappeared, and net retention just under 100% means expansion did not quite cover it.

The last line is the one worth sitting with. At the same loss rate, this company has to find ₹145 lakh of new and expansion ARR next year purely to stay at ₹846 lakh. The year before, standing still cost ₹110 lakh. The treadmill speeds up as the base grows, and nothing in a growth rate shows it.

Neither number means much alone, and the pair is diagnostic.

• ·High gross and high net: customers stay and spend more. The business compounds, and acquisition spend does most of its work once.
• ·Low gross and high net: a leaky base held up by expansion, usually from a handful of large accounts. This is the pattern most worth catching early, because it looks healthy on the net number and the underlying product problem is getting worse.
• ·High gross and low net: customers stay and never spend more. Often a pricing or packaging problem rather than a product one, and one of the more fixable findings.
• ·Low gross and low net: growth is entirely rented. Every rupee of ARR has to be bought, every year, and the cost of standing still rises with the base.

• ·Counting upgrades as new business. The most common error by a wide margin, and it inflates new ARR while hiding the expansion that is the most valuable thing you have.
• ·Putting implementation and services fees into ARR. They are real revenue and they are not recurring, and a diligence team will strip them out.
• ·Counting a churned customer at their original ARR rather than what they were paying when they left. If they had downgraded first, part of that loss was contraction.
• ·Including customers who have given notice but have not yet left. They are churn, not base.
• ·Measuring over a period shorter than a year and annualising it. Renewals cluster, so a quarter is not a twelfth of a year and NRR calculated from one is misleading.
• ·Mixing segments. Enterprise and self-serve retain nothing like each other, and a blended bridge describes neither. Run it once per segment if you have the volume.

Each of these makes the numbers look better than they are, which is worth noticing. The errors are not random, and diligence teams know exactly where to look.

The bridge is diagnostic rather than prescriptive, and each of the four patterns has a different first move.

1. 01If gross retention is the problem, find out where it is concentrated before doing anything else. Split it by plan, by contract value and by acquisition channel. Churn is almost never spread evenly, and the answer is usually that one segment is doing most of the damage.
2. 02If that segment turns out to be your cheapest plan, the question is whether it should exist at the current price, with the current support entitlement, or at all. Cheap customers who churn fast can consume more in acquisition and support than they ever contribute.
3. 03If expansion is the problem, look at packaging rather than at the account management team. A flat annual licence gives a growing customer nothing to buy. Per seat, per unit of their activity, or tiered pricing all make expansion happen without anyone selling.
4. 04If contraction is rising while churn is flat, customers are staying and using less. That is usually an adoption problem and it is an early warning of churn to come, not an alternative to it.
5. 05If a handful of large accounts are carrying net retention, calculate it again with the top three removed. The figure that remains is the one that describes the business you are building.

One habit worth adopting regardless: record the reason each customer left, in their words rather than the account manager's. Twenty of those is more useful than any amount of further analysis of the same five numbers.

There is no published Indian benchmark for net revenue retention. Reported figures come from American and global surveys of private B2B software companies, and those surveys show medians around or slightly above 100% for private companies in recent years, with gross retention medians in the high eighties.

Use them as orientation rather than as a target. Two things make a borrowed benchmark unreliable: retention varies enormously by contract value, with enterprise holding far better than small business, and the definitions differ between surveys in ways that move the number by several points.

The comparison that actually helps is against yourself. Run the bridge each quarter, keep the definitions fixed, and watch the direction. A company whose gross retention has improved four points over a year has something to say that no benchmark provides.

• ·It works on revenue, not on customer counts. Logo retention is a different figure and usually lower, because small customers churn more often than large ones.
• ·It is one period at a time. The pattern across several periods is what matters, and that needs the bridge kept as a running record.
• ·It does not tell you where the churn is concentrated. That needs a cohort table, and the answer is almost never that churn is spread evenly.
• ·It says nothing about whether the growth was worth buying. Acquisition cost and payback are separate questions, and the CAC payback calculator covers them.
• ·It assumes your ARR definition is right. Garbage in this bridge is garbage that looks authoritative, which is worse than no bridge.