Recurring Revenue: Why Predictability Is Valued
Recurring revenue is revenue a business has a standing arrangement to receive: a subscription, a licence, a maintenance contract. Revenue that merely happens again carries no such agreement in advance. A shop whose customers return every week has revenue that repeats and no commitment. Two businesses can report the same annual total and differ entirely in how much of the coming period is already settled.
Revenue divides three ways: how much arrives, in what form it arrives, and how much of it is known before the period begins. Recurring revenue is the third of those three, and a subscription is one of the forms the second can take. How much of a period is known before the period begins is a property of the arrangement behind the revenue rather than of the statement that reports it.
Two revenue lines can sit on two statements looking identical, carry the same total to the rupee, and hold completely different amounts of advance certainty. No line item anywhere on a statement identifies which of the two is present. Telling the two apart means going underneath the number to the contract that produced it.
What separates revenue that is contracted from revenue that merely repeats?
Two businesses on one street will do this faster than any definition. A milk vendor delivers to the same two hundred houses every morning and has done for nine years. Nobody has signed anything, ever. Two doors down, a small gym takes a membership for the year, collected in advance, and the members come in when they feel like it. Both businesses take money from the same neighbourhood, month after month, from people who keep coming back. One of them has revenue that repeats. The other has revenue that is contracted. Revenue that repeats is not the same as revenue that is contracted, and the only difference that matters is whether anybody has committed to anything.
Notice the direction each one is observed from. A habit is read backwards, out of what already happened: nine years of milk delivered is a complete and accurate record of the past and says nothing whatever about tomorrow morning. An agreement is read forwards, out of what has been settled: the gym can count its members and name the amount before the year starts. Both readings are honest. Only one of the two readings answers how much of next year is already decided.
The next part is where the confusion usually lives. A habit can be far more reliable than a contract. The two hundred houses may have a lower rate of departure than any gym in the city. Reliability and commitment are different properties, and the milk vendor may well have the more reliable of the two arrangements. The milk vendor has no way of knowing that reliability in advance. The gym has a number for it before the year starts.
Now put the case entity on it. Setu Bazaar is an invented marketplace with 50,000 buyers. In its transactional arrangement it takes a take rateThe share a marketplace keeps of everything that passes across it. How that share is measured, and what moves it, is built in the notes on business fundamentals and models. of 4.00 per cent on a gross flowThe whole value of what buyers and merchants trade on a marketplace in a twelve month period. What the marketplace itself collects is a small slice of that figure and is counted separately. of Rs 5,00,00,00,000/-, which is Rs 500 crore, and collects Rs 20,00,00,000/- in the year. Underneath that total sit 5,00,000 separate orders, ten from each buyer, each leaving Rs 400/- of revenue behind it. Not one of those 5,00,000 orders exists on the first day of the year. Every rupee of the Rs 20,00,00,000/- depends on a transaction nobody has performed yet.
Now run the same business as a subscription instead. Ask every one of the 50,000 buyers for Rs 4,000/- covering the whole twelve months, agreed and payable at the start. Fifty thousand at Rs 4,000/- is Rs 20,00,00,000/-. The identical total, from the identical buyers, doing identical work. The two arrangements cannot be told apart on the revenue line of a statement, and they are not remotely the same thing to anybody trying to work out what next year holds.
A milk vendor has delivered to the same two hundred houses every morning for nine years. Not one house has signed anything. Is that recurring revenue?
Why is predictable revenue sought after, and what does that preference not settle?
The word predictable invites a sentence that does not follow from it. Predictable revenue is sought after. Lenders prefer it, buyers of businesses prefer it, and the people running the business prefer it most of all. The preference is an observable fact about how people behave. Why predictable revenue is preferred can be described entirely in terms of what is known and when it becomes known; what predictable revenue is worth is a valuation, and a valuation needs assumptions of its own.
Working out what a stream is worth needs a set of assumptions about how long it lasts, what it costs to serve and what a rupee arriving later counts for against a rupee arriving now, and not one of those assumptions can be read off the revenue line. An answer that skipped straight to the conclusion would rest on inputs nobody had seen. So the honest version of the claim is narrower and more useful: predictable revenue reveals more, earlier.
Preferring it comes down, in practice, to three ordinary questions, none of which is about a figure. A lender wants to know how much money is already committed to arrive before the repayment falls due. Somebody deciding how many people to hire in April wants to know how much of the salary bill is covered by arrangements that already exist. Somebody planning warehouse space wants to know how many buyers are already signed rather than hoped for. All three ask what is known and how early it became known, and all three can be answered in full without naming a value.
Here is the shape of it on Setu Bazaar's two arrangements, and the shape is the point. Both finish the year at Rs 20,00,00,000/-. The transactional arrangement learns its own revenue as the orders arrive, so on day one it knows Rs 0/- and by the last day it knows all of it. The subscription arrangement knows Rs 16,00,00,000/- on day one and spends the year winning the rest. Draw those two as lines and they end in exactly the same place. The path differs, and the path is what somebody planning next year has to work with.
Two arrangements at Setu Bazaar raise the identical Rs 20,00,00,000/-, and one of them has Rs 16,00,00,000/- of it contracted before the year starts. What does that establish about what the contracted one is worth?
How much of the coming year is already settled?
Getting that number takes three steps. First, count the contracted base, meaning everything signed and running. Apply what is known about how much of that base continues into the next period. State the answer twice, once as a rupee amount and once as a share of the coming period. Either one can then be rebuilt from the other.
Run it on Setu Bazaar's subscription arrangement. The base is 50,000 buyers on annual agreements at Rs 4,000/- each, a total of Rs 20,00,00,000/-. Retention has been observed at 80.00 per cent, so 40,000 of those buyers continue and 10,000 do not. Forty thousand at Rs 4,000/- is Rs 16,00,00,000/-, and the ten thousand who leave take Rs 4,00,00,000/- with them. The two parts add back to Rs 20,00,00,000/-, and doing that addition every time catches the commonest slip. The slip is putting the contracted part where the part still to be won belongs. Rs 16,00,00,000/- is contracted before the year starts and 20.00 per cent of the year has to be won again, and that pair of numbers says more about next year than the annual total does.
The 80.00 per cent is doing a great deal of work in that sentence, so be careful about what it actually is. The figure was observed on a cohortCustomers bundled by the period in which they first signed up, then tracked as one group so a later intake can be measured against them. Why customers stay or go is taken up in the notes on customers and brands. that has already been through a renewal, and it is being pointed at a period that has not happened. Pointing it forward is an assumption and should be written down as one. Quietly turning last year's observation into next year's fact smuggles a forecast in under the cover of arithmetic. The rupee amount is only as firm as the share that produced it.
The top part of that split is read wrongly more often than the contracted part. The Rs 4,00,00,000/- still to be won is not a smaller version of the same money. The Rs 4,00,00,000/- is unagreed money. If none of it is won, the year finishes at Rs 16,00,00,000/- and not at Rs 20,00,00,000/-, and the difference between those two outcomes is a fifth of the business. Money collected in advance also has a place of its own in the accounts, held as deferred incomeMoney taken in before the thing it pays for has been supplied. It waits in the accounts as a duty to supply, and turns into revenue only once that duty has been met. until whatever was promised has actually been supplied, and that is a separate matter from whether the arrangement behind it is firm.
India, for the accounting words only
The accounting words used above are the Indian ones. Whether an amount collected in advance sits as an obligation or as revenue, and when a performance obligationA promise inside a contract to hand over a distinct good or service. The accounting standards use the phrase to decide when revenue may be recognised, and they define it precisely. has been discharged, are decided by the Indian accounting standards rather than by any reasoning here. The standard on revenue from contracts with customers also asks certain entities to disclose what they have agreed to deliver and have not yet delivered. The disclosure is the closest thing in a published statement to the contracted amount described above. Which entities that applies to, what has to be said and over what periods are read at the standard's own source before anybody relies on it.
Setu Bazaar's subscription raises Rs 20,00,00,000/- from 50,000 buyers at Rs 4,000/- each, and retention has been observed at 80.00 per cent. How much is contracted before the year starts?
How to Assess the Durability of Recurring Revenue: what three things are in question?
Three, and they are worth naming separately because a business describing its own arrangement will usually quote one of them and stay quiet about the other two. How long the agreements run. The effort it takes to get out of one. And whether continuing is the default or has to be won every time. Everything else worth asking about a recurring line hangs off those three.
The length gets quoted most, so start there. An agreement written for five years sounds firmer than one written for a year, and a business with a book of five year agreements will say so. The stated term is worth holding lightly. A five year term gives the longest the arrangement could possibly run, and nothing at all about the shortest.
The second thing is where the length gets tested. Ask what a buyer actually has to do to stop paying. A notice period is the usual answer: give a month, give a quarter, give a year. Whatever that period is, it is the stretch of time a buyer who decided this morning to leave must still keep paying for. Set the two side by side on the same contracted base and the difference is not subtle. Setu Bazaar's Rs 16,00,00,000/- under a year's notice means a buyer who decides today still owes the full twelve months, so the whole Rs 16,00,00,000/- is held by the notice. The same Rs 16,00,00,000/- under a month's notice holds one month of itself, Rs 1,33,33,333/- to the nearest rupee, and leaves Rs 14,66,66,667/- resting on nothing more than buyers choosing to stay. Nothing about the size of the arrangement moved between those two sentences. A long agreement that anybody may leave at a month's notice is a one month agreement. The stated term is the least informative of the three.
The third is quieter and easy to miss. Does the arrangement carry on unless somebody stops it, or does it end unless somebody renews it? The two look almost identical on paper and behave completely differently. Each puts the effort of continuing on the opposite side of the table. An arrangement that continues by default survives inattention. An arrangement that must be actively renewed has to win a fresh decision from every single buyer, on a date, every period, and inattention kills it. The notice periods of a month and a year used above are illustrative, picked so that the arithmetic shows up clearly, and neither is drawn from any contract anywhere.
Durability rests on three things. Which set names all three of them?
Which one of the three gives least when it arrives on its own?
The same Rs 16,00,00,000/- contracted base, once with a year's notice to leave and once with a month's. Why is the second one weaker?
Move retention, then change what it takes to leave, and watch how much of the year is settled before it starts.
The panel opens on the worked example above: Rs 20,00,00,000/- of subscription, retention at 80.00 per cent, Rs 16,00,00,000/- contracted, and a year's notice needed to leave. Move the slider first and watch the solid part of the year grow and shrink while the outline around the whole year never moves. The total is held fixed at Rs 20,00,00,000/- throughout. Then leave retention where it is and change the notice period instead, and watch a large contracted amount that nothing about the arithmetic has touched turn pale from the right hand end inwards. The second control is the durability question. The first moves how much is agreed, and the second moves how much of the agreement anybody is actually held to.
Where does recurring revenue hide its risk?
Somewhere very specific, and it is not where most readers look for it. Set the two Setu Bazaar arrangements side by side and count the decisions rather than the rupees. In the transactional arrangement, 50,000 buyers place 5,00,000 orders in a year, about 41,667 buying decisions a month to the nearest order, spread evenly across every month there is. Any one of those decisions can go badly and none of them can go badly enough to matter on its own. In the subscription arrangement, the same 50,000 buyers make 50,000 decisions in the year. Fifty thousand decisions is one tenth as many, and that sounds like an unambiguous improvement until somebody asks when the decisions happen.
The decisions happen on renewal dates. If every agreement was signed at the same time and runs to the same anniversary, all 50,000 of those decisions land inside a single month, and the eleven months either side of it carry none at all. Contracted revenue does not remove the buyer's decision. Contracted revenue postpones the decision and then concentrates it. A concentrated risk is a different risk, not an absent one.
Sit with what that does to a year. For eleven months the arrangement looks immovable and the reported churnThe share of customers who stop paying over a period, the mirror image of retention. What drives it, and how to reduce it, is taken up in the notes on customers and brands. reads as nil, because nothing has come up for decision. In the twelfth month the entire book is in play at once. Nothing about the business changed between month eleven and month twelve, and every number a reader might have been watching says it did. A transactional business gets its bad news in small amounts, continuously, and small continuous bad news is unpleasant and impossible to ignore. A contracted business gets its bad news in one piece, on a date, and can look completely settled the day before.
The useful move, once that is seen, is to stop asking how much is contracted and start asking when it comes up. A business whose renewals are spread evenly across twelve months has the same contracted amount as one whose renewals all fall in March, and the two are not in the same position at all. Spread renewals give eleven chances to notice a problem. A single anniversary gives one.
Setu Bazaar moves its 50,000 buyers onto annual subscriptions. Does that remove the decision those buyers were making?
What does calling revenue recurring not establish?
Four things, and each one is claimed on the strength of the label often enough to be worth stating flatly. The label does not establish that customers will stay. An agreement is a claim on somebody rather than a fact about them, and every one of those agreements ends on a date. The label does not establish that the amount will hold. The same buyer can renew at a smaller commitment and the arrangement still counts as recurring. The label does not establish that the business earns anything. What is left after serving those buyers depends on contributionWhat is left out of a sale after the costs that move with volume have been taken off. It is built, with the fixed costs it has to cover, in the notes on unit economics. and on the fixed costs behind it, and Setu Bazaar's own worked figures ran to a loss at exactly this revenue. And the label does not establish what any of it is worth. Worth is a valuation and a separate exercise.
Recurring describes an arrangement and not an outcome. That is the whole of the caution. The word says something real and useful about the mechanism by which money arrives, and nothing whatever about whether the money will keep arriving, how much of it there will be, or whether any of it is left at the end. A reader who takes the label as a verdict has read a description of plumbing as a description of results.
What does a lender or an analyst ask before any of the definitions matter?
People who read these arrangements for a living do not start from the definition. One question goes underneath the label and comes back with a number, and the question is worth stealing.
The question that cuts through the label
How much of next year could walk away before the next invoice? The question forces the notice period and the renewal dates into the open in the same breath, so it separates a contracted base from a hopeful one faster than any definition. A lender deciding what to advance against next year's collections asks it because the answer bounds what can be counted on before the repayment falls due. An analyst asks it because a business quoting a contracted figure has given the size of the base and nothing about how firmly it is attached. A lease is a fixed obligation and the money meant to cover it may not be one, so somebody running the business asks it before committing to a lease.
Ask it in three parts, in order, and the answer is arithmetic rather than opinion. Is anything actually signed, or is this a habit being described in contract language? When do those agreements next come up, month by month rather than as an average term? And what does a buyer have to do to stop paying? On Setu Bazaar's Rs 16,00,00,000/- base, the first two questions can both be answered comfortably and the third still moves the answer by Rs 14,66,66,667/-, depending only on whether the exit takes a month or a year. Notice that the third question is also the one a business is least likely to volunteer, and the one that survives the shortest invoice cycleThe regular interval at which a supplier raises its next bill, often monthly and sometimes quarterly. It governs the timing of a demand for payment and not the size of one. the least well.
What does the whole of it look like on one set of numbers?
Every component is set out below, and every total can be rebuilt from the components. The moment the totals differ the comparison stops being about certainty and starts being about size, so the subscription arrangement and the transactional arrangement are held at the identical Rs 20,00,00,000/-.
| Component | How it is built | Amount |
|---|---|---|
| Buyers on subscription | Carried from the earlier reading on this business | 50,000 |
| Subscription each buyer pays for the year | Carried from the earlier reading on this business | Rs 4,000/- |
| The year's subscription revenue | 50,000 buyers at Rs 4,000/- each | Rs 20,00,00,000/- |
| The same year, collected transactionally | 5,00,000 orders leaving Rs 400/- of revenue each | Rs 20,00,00,000/- |
| Retention on the buyers now on subscription | An assumption, observed on an earlier set and pointed forward | 80.00 per cent |
| Buyers continuing into the coming year | 50,000 at 80.00 per cent | 40,000 |
| Contracted before the year starts | 40,000 buyers at Rs 4,000/- each | Rs 16,00,00,000/- |
| Buyers not continuing | 50,000 less 40,000 | 10,000 |
| Still to be won again during the year | 10,000 buyers at Rs 4,000/- each, being 20.00 per cent | Rs 4,00,00,000/- |
| The two parts add back to the year | Rs 16,00,00,000/- plus Rs 4,00,00,000/- | Rs 20,00,00,000/- |
The durability test now goes to that same base, without a single figure in the table above changing. The only thing that moves is the notice a buyer must give, and the notice periods are illustrative rather than terms taken from anywhere.
| Notice a buyer must give | How it is built | Still payable after a buyer decides to leave |
|---|---|---|
| A year | Rs 16,00,00,000/- for twelve of twelve months | Rs 16,00,00,000/- |
| Six months | Rs 16,00,00,000/- for six of twelve months | Rs 8,00,00,000/- |
| Three months | Rs 16,00,00,000/- for three of twelve months | Rs 4,00,00,000/- |
| One month | Rs 16,00,00,000/- for one of twelve months, to the nearest rupee | Rs 1,33,33,333/- |
| Resting on buyers choosing to stay, at a month's notice | Rs 16,00,00,000/- less Rs 1,33,33,333/- | Rs 14,66,66,667/- |
Setu Bazaar and every figure above were built for teaching. The 50,000 buyers, the Rs 4,000/- each, the Rs 20,00,00,000/- and the 5,00,000 orders are carried unchanged from the earlier reading on this business rather than recalculated here. Retention of 80.00 per cent is an assumption stated as one, and the notice periods are illustrative. Every amount is an exact whole number of rupees except one. The one month figure does not divide the year evenly and is shown to the nearest rupee.
What goes wrong when a recurring label is read as settled revenue?
The contracted figure is correct and the conclusion is wrong
Here is how it happens, and it happens to careful readers rather than lazy ones. A business reports Rs 16,00,00,000/- of contracted recurring revenue against a total of Rs 20,00,00,000/-, or 80.00 per cent. Every word of that is accurate. A reader writes next year down as substantially settled, plans against it, and in a lender's case advances money against it. Then the renewal schedule turns up and all 40,000 of those agreements come up inside the first quarter, and every one of them can be ended on a month's notice.
Count what is actually held at that point. One month of the contracted base is Rs 1,33,33,333/-. The other Rs 14,66,66,667/- is resting on 40,000 separate decisions, all of which arrive within three months of each other. The 80.00 per cent was never false. The 80.00 per cent described the size of the base and said nothing about how long the base was attached, and the reader supplied the missing half of the sentence without noticing.
The fix is two questions and both are answerable: what does it take to leave, and when does the next renewal land. A contracted amount whose agreements all end soon is a description of the year just gone rather than the year ahead. An average term of eleven months is exactly what forty thousand agreements ending in March look like from a distance, so ask for the renewal dates month by month rather than as an average.
A statement reports Rs 16,00,00,000/- of contracted recurring revenue, being 80.00 per cent of the total. A reader writes next year down as settled. What did that reading miss first?
The worth of predictable revenue is a separate exercise, resting on assumptions about how long a stream lasts, what it costs to serve and what a rupee arriving later counts for against one arriving now. Whether Setu Bazaar's buyers will actually stay, and what would make them stay, is covered under customers and brands. Why a business is able to hold its charge while its buyers keep paying is covered under competitive advantage and moats. How the timing of an invoice moves cash rather than revenue is covered under working capital and the cash conversion cycle.
Which sources settle the accounting words named here?
Four documents hold the parts of this subject that notes describing an arrangement have no business settling on their own. Two of them are the standard on revenue from contracts with customers, in its Indian form and its international one. The standard is where the words contract and obligation stop being loose and acquire tests. One is the guidance written for the people who have to apply it. The last is the rulebook deciding what a listed entity publishes and how often. Without that rulebook, no contracted figure would be visible to anybody outside the business at all.
| Who holds it | What to look for | Site |
|---|---|---|
| Ministry of Corporate Affairs | Indian Accounting Standard 115, Revenue from Contracts with Customers, as notified, including what it asks an entity to disclose about obligations taken on and not yet performed | mca.gov.in |
| Institute of Chartered Accountants of India | The educational material on that standard, written for the people applying it rather than for the people drafting it | icai.org |
| International Financial Reporting Standards (IFRS) Foundation | IFRS 15, Revenue from Contracts with Customers, the international standard the Indian one is built alongside | ifrs.org |
| Securities and Exchange Board of India | The Listing Obligations and Disclosure Requirements Regulations, for what a listed entity has to publish about its results and how often | sebi.gov.in |
Setu Bazaar is invented.
Educational material. Not advice on any investment, tax, budget or market position.