Every company closes its books. On a schedule, someone reconciles the accounts, ties the balances, and signs off on a statement of what happened. The discipline is so standard that skipping it would read as negligence.

Recurring revenue almost never gets the same treatment. The MRR figure updates in a dashboard, the chart moves, and the month ends without anyone establishing what actually happened or being able to prove it. When a board member, a lender, or a client asks why the number moved, the answer gets rebuilt from scratch in a spreadsheet, and rebuilt differently each time.

A recurring-revenue close applies the discipline of a book close to the subscription base. This guide defines the close, walks through its five steps, and describes what separates a close that holds up under questioning from a report that merely looks finished.

What a recurring-revenue close is

A recurring-revenue close is a monthly process that reconciles recurring revenue from opening balance to ending balance, explains every movement in between, and ties each figure to the source records that produced it. The output is not a chart. It is a settled account of the month: the starting position, the revenue that arrived, grew, shrank, or left, the ending position, and the evidence behind each leg.

A finished close produces four things:

• A movement statement: opening balance, new, expansion, contraction, churn, and reactivation, resolving to the ending balance on a stated basis.

• A reconciliation: proof that the movements sum to the ending balance and that the figures tie back to the source data.

• An exception list: the customers and transactions that need a human decision before the numbers settle.

• A written read: what happened this month, in words, connected to the figures rather than floating above them.

Why the topline is not a close

A number can be right and still hide the month. An MRR figure that is up 3.9 percent is a true sentence that says nothing about the one enterprise expansion that drove it, or the eight smaller customers who quietly left underneath it. Both months exist inside the same topline.

A dashboard reports position. A close explains movement. The difference shows up the moment someone asks a follow-up question. Position can be restated but not defended; movement, properly closed, comes with its own evidence.

The five steps of a close

The steps below are ordered so that each one settles what the next depends on. Skipping ahead is how closes quietly go wrong.

1. Confirm the source shape

Before any figure is computed, confirm what the source data actually is: which columns identify the customer, the date, and the amount, what period the file covers, whether currencies are mixed, and which payment types are present. A close computed on a misread file is wrong before it starts, and the error is invisible downstream. What the file contains depends on which door you came through: the five routes out of Stripe do not all carry a service period, and one of them cannot answer a period question at all.

2. Settle the run plan

Decide, explicitly, how the month will be read. The reporting basis: cash collected, or normalized recurring revenue. The scope: which rows count as recurring, and what is excluded as one-off, credit, or refund. The treatment of prepaid revenue: spread across the service period, or recognized as collected, and on what plan terms. None of these are calculations. They are decisions, and a defensible close records them as decisions rather than burying them in formulas.

3. Compute the movement

With the shape confirmed and the plan settled, compute the movement statement: starting balance, new revenue, expansion, contraction, lost revenue, and reactivation, resolving to the ending balance. Each leg should attribute to the customers that produced it, so the statement can open into names and transactions rather than staying an aggregate. Which leg a change belongs to is a rules question, not a judgement made fresh each month, and the classification logic is worth settling once.

4. Reconcile

Check that the close ties. The movement legs must sum from opening to ending balance exactly. Customer counts should move consistently with the revenue story. Source coverage should be stated: what share of revenue the confirmed decisions cover, and what remains assumed. A reconciliation that fails is information, not embarrassment. It means the month is not yet understood.

5. Write the read

The last step is the written account: what happened, which customers and segments drove it, what the headline conceals, and what deserves watching next month. The read is written from the verified figures, not alongside them. If a sentence in the read cannot be traced to a figure in the close, it does not belong.

What makes a close defensible

Defensible does not mean perfect. It means the close survives follow-up questions. Four properties carry most of that weight.

The basis is stated. Every figure says whether it is cash collected or normalized recurring revenue. A retention rate without a stated basis is not a metric, it is a guess wearing one.

Figures trace to source. Any number in the close can open into the accounts, transactions, and rules that produced it. The evidence stays attached instead of being summarized away.

Assumptions are surfaced. Where a plan term was inferred or a payment type was classified by judgment, the close says so. A close that reports confirmed decisions for 94 percent of revenue is stronger than one that implies certainty it cannot deliver.

It withholds rather than guesses. When the data cannot support a figure, the defensible move is to say so and stop. A withheld number costs a moment of discomfort. A guessed number that later fails costs the credibility of every figure around it.

The cost of skipping this is not usually a wrong number. It is an unexplainable one, which is what happens when automation hands over the figure and quietly keeps the reasoning.

A close is not a dashboard

Dashboards are continuous, aggregate, and silent about method. A close is periodic, settled, and explicit about method. Both have a place. The failure mode is using a dashboard where a close is owed: board reporting, lender conversations, client reviews, and any setting where someone else's decision depends on the number being right and provable.

Where Morevy fits

Morevy runs this process as a product. It confirms the source shape and asks for the run-plan decisions before computing anything, produces the movement statement and reconciliation on a stated basis, keeps every figure traced to its source rows, and drafts the read from verified results for you to edit. When a basis cannot be established for a material share of revenue, it withholds the dependent figures rather than approximating them. The close is currently in private beta at morevy.ai.

Related documentation: Your first monthly review and What reconciled actually means.