A repeatable month-end close starts with the documents

Ask ten firm owners when their books go out and you will hear ranges. The 8th, usually. The 12th on a bad month. One client that always lands on the 20th and nobody remembers why. A repeatable close, the kind where you can put a delivery date in the engagement letter and hit it eleven months out of twelve, does not come from working faster. It comes from a calendar built backward from delivery day, with a hard documents gate at the front. This post is that calendar, the one we run on our own client book at Scale CPA, plus notes on installing it.

Work backward from the day you deliver

Most close calendars are built forward. Statements come in whenever, reconciliation starts when someone has capacity, review happens when the reviewer surfaces, and delivery lands wherever it lands. Every stage absorbs the slippage of the one before it, silently, until the client asks where their financials are.

Building backward inverts that. Commit to a delivery day first, say business day ten. Then ask what must be finished the day before delivery (review fixes), the day before that (review itself), and so on until you reach day one. Each stage gets an entry condition and an exit condition. A stage that cannot enter on schedule raises its hand immediately instead of quietly eating the buffer.

When we mapped our own close this way, the uncomfortable finding was how little of the elapsed time was work. Reconciliation, adjustments, and review are hours per client. The days belonged to waiting: statements that had not arrived, logins that had silently broken, a client contact on vacation. Most slow closes are docs-gated rather than work-gated, and the breakdown of where firms lose time in the monthly close walks through why. The calendar below is designed around that finding: the scarce, unpredictable input goes first, where a miss is visible and cheap.

The close calendar

Days are business days after month end. Compress or stretch the bands to fit your delivery promise; the order and the gates are the part that matters.

Days 1 and 2, the documents gate. Every bank, credit card, and loan statement for every client either sits in the right folder or appears on an exceptions list with a person’s name next to it. Processor statements too: Stripe, PayPal, Square, Shopify payouts, wherever the client sells. Exceptions come in three kinds, and each kind gets handled differently. First, the statement has not posted at the bank yet, which is normal for mid-month cycles; check the account’s real cycle date before chasing anyone, because bank statements do not follow calendar months. Second, a connection or login broke and needs repair; connections break for predictable reasons, and several large banks (Wells Fargo, Chase, Bank of America, Citi) run security-hold flows that pause third-party access until the client acts. Third, the client has to send something themselves. Only the third kind needs a chase email, and now it is a specific ask on day one instead of a vague reminder on day six.

The gate rule that makes this work: reconciliation on a client does not start until that client’s gate status is known. Green means go. Red means the exception gets an owner and a due date, and the client’s slot in the reconciliation band moves, visibly, on a board everyone can see.

Days 3 through 6, reconciliation and adjustments. This is the schedulable middle. Because the gate ran first, you know on day two exactly which clients are workable, so you can batch them and level the staffing instead of guessing. Preparers work from complete inputs. Nobody opens a file, discovers a missing statement, closes the file, and context-switches away. That half-open file is where more hours die than most time reports admit.

Days 7 and 8, review. Reviewer availability is the second most common gate failure after documents, and it fails for the same structural reason: the review stage inherits whatever slippage came before, and reviewers have other jobs. Book review slots in advance like client appointments, and protect them. A review that slips a day pushes fixes into delivery day.

Day 9, fixes. Short by design. If fixes routinely need more than a day, the problem is upstream in preparation quality or review depth, and the calendar just made it measurable.

Day 10, delivery. With commentary, on the promised date. The whole point of a repeatable date is that the client learns to expect it.

Why the documents gate goes first

Because errors found early are cheap and errors found late are expensive, and a missing document is the single most common error in a close. A statement you learn is missing on day one costs an email. The same statement discovered on day six, when a preparer opens the file, costs the email plus a blown reconciliation slot plus a rescheduled review. Discovered on day nine, it costs the delivery date.

The gate also changes what “late” means. Without it, lateness is a feeling that accumulates until someone panics. With it, lateness is a list: two accounts, named owners, known reasons, checked every morning. Our own close stalled waiting on statements more months than we care to admit before we started treating the gate as a real stage with a real exit condition rather than a background hope.

Making the gate pass itself

Everything above runs on a spreadsheet grid and a fixed morning routine, and the complete guide to collecting client bank statements covers those manual mechanics. The gate is also the stage that automates most cleanly, because it is pure retrieval and checking, no judgment required. This is the part we ended up building software for. StatementFlow connects each client’s banks once through Plaid or Mastercard Open Banking, learns each account’s actual posting cycle, pulls the official PDF when the bank publishes it, verifies the download, and files it into the firm’s own Google Drive by client, year, and month. The coverage board is the documents gate rendered as a grid: every account by every month, with gaps, needed reconnects, and bank security holds flagged before close starts. It retrieves and files documents only. It does not extract data from the PDFs (Dext and AutoEntry cover that) and it does not touch the reconciliation itself. The product page has the full shape of it.

With retrieval automated, days one and two stop being a scramble and become a fifteen-minute read of the board: which cells are green, which exceptions are real, who owns each one.

Rollout notes

Month one measures your lag; treat it that way. The first time you run the gate, expect a sizable share of your book to fail it. Broken logins nobody knew about, accounts nobody inventoried, statements that never had a home folder. This is the gate doing its job. Log every failure, fix the systemic ones, and expect month two to look meaningfully better.

Business days, always. A calendar written in calendar days gets wrecked by weekends and holidays a few times a year, and the wreckage always lands on review week. Write the bands in business days and re-derive the actual dates each month.

Mid-month cycles need a decision, per account, up front. Some accounts genuinely cannot pass the gate on day two because the bank cuts the statement on the 15th. Decide the policy once: close on the prior statement plus activity, or carve those accounts into a second mini-gate mid-month. Either works. Re-litigating it every close does not.

Watch the exceptions list for rot. An exceptions list without owners and due dates turns into a graveyard within a quarter. If an item survives two closes, escalate structurally: change the access method for that account, or change the conversation with that client.

Revisit quarterly. Delivery promises change, staff changes, the client book changes. Check the bands against actual dates four times a year, and re-derive the whole calendar whenever the delivery day moves.

A repeatable close is mostly a repeatable day one. If the documents gate is where your calendar keeps failing, StatementFlow is in early access for US firms and we are onboarding from the waitlist: request early access.