Category: Close speed

A seed-stage founder we work with opened Puzzle one Tuesday and saw a clean dashboard: $2.1M cash, $140K net burn, runway in the green. Three days later his lead investor asked how he was recognizing a $480K annual prepay that had landed that month. The software had booked the cash correctly and flagged the deferred-revenue line.

What it could not do was decide how the contract’s setup fee, usage tiers, and renewal option split into performance obligations under the revenue standard. That decision sat with a person. The same gap surfaces for every founder running modern accounting software: the numbers are current, and the judgment underneath them is still open.

Puzzle is the strongest real-time accounting platform a startup can operate beneath its books, and we partner with it deliberately for that capability. This piece is not a critique of Puzzle. It examines where any software reaches its boundary and a human accounting team assumes responsibility — so founders provision that boundary intentionally rather than encountering it the week before a board meeting or a statutory tax deadline.

What does Puzzle actually do for a startup?

Puzzle automates the mechanical layer of accounting and keeps it current. Its AI categorizes transactions from day one, connects natively to Stripe, Brex, Mercury, Ramp, Deel and Gusto, and produces cash, burn, runway, and revenue-recognition figures that update continuously rather than at month-end. On its own site, Puzzle reports up to 98% automation and a close up to 50% faster, every month.

One detail matters more than the rest for founders who raise capital. Puzzle keeps cash and accrual books from a single data model, so you get a GAAP-basis accrual view alongside the cash view — the version investors and your tax preparer want.

That is genuinely hard, and it removes most of the manual categorization a junior bookkeeper used to grind through. The data is no longer the bottleneck. The judgment on top of it is.

Where does accounting software stop?

Software stops at the line between recording a fact and forming an opinion. It records that $480K hit the bank. It cannot decide, under the revenue standard, how much of that is earned this month. Three areas force that kind of judgment, and they recur at almost every startup.

The first is revenue recognition. ASC 606 runs on a five-step model: identify the contract, identify the performance obligations, set the transaction price, allocate it across the obligations, then recognize revenue as each is satisfied. Steps two and three carry real judgment.

Estimating variable consideration, applying the constraint on reversals, and deciding what counts as a distinct obligation are determinations a person makes and signs. A tool can hold the answer once it exists. It cannot reach the answer for a usage-based contract with tiers and a renewal option.

The second is stock compensation. Under ASC 718, you recognize award cost at grant-date fair value. For private startup equity there is no market price, so fair value gets estimated with a model that needs assumptions — volatility, expected term, a forfeiture policy. Those inputs are choices. Get them wrong and your option expense and net loss both move. No software picks those assumptions and owns the result.

The third is the close itself. A continuously updated ledger is not a closed month. The close is the disciplined sequence where a controller reconciles every account, books accruals for work delivered but not yet invoiced, trues up estimates against actuals, and certifies the financial statements as complete.

APQC’s benchmarks situate a representative monthly close at approximately six calendar days, even under substantial automation, because that review-and-estimation layer remains fundamentally human. Puzzle compresses the data preparation dramatically and hands a controller a clean, current ledger to close from — the real-time data on one side, the reviewed sign-off on the other.

Software handles vs. a human team handles

The split is clean once you draw it. Anything mechanical and rule-following belongs to the software. Anything that requires an estimate, a policy choice, or a signature belongs to a person. Here is how that maps across the work a startup actually generates each month.

Task	Puzzle (software)	A human accounting team
Transaction categorization	AI categorizes from day one, up to 98% automated	Reviews edge cases, sets the chart-of-accounts policy
Bank and tool sync	Native Stripe, Brex, Mercury, Ramp, Deel, Gusto feeds	Confirms completeness, investigates gaps
Cash, burn, runway dashboards	Updated continuously, not at month-end	Interprets, sanity-checks the inputs, frames for the board
Revenue recognition (ASC 606)	Records the booked revenue once decided	Identifies performance obligations, estimates variable consideration
Stock comp (ASC 718)	Holds the expense schedule	Estimates grant-date fair value, sets forfeiture policy
Month-end close	Cuts data-prep time up to 50%	Owns reconciliation, accruals, the final sign-off
Board and investor reporting	Feeds the raw numbers	Builds the narrative, ties variances to decisions
Corporate tax filing	Produces clean GAAP-basis books	Files Form 1120, claims credits, plans the position

Does Puzzle file my startup’s taxes?

No. Puzzle keeps your books; it does not discharge a tax filing. Per the IRS, a domestic corporation uses Form 1120 to report income and figure its tax liability — a statutory obligation no bookkeeping tool satisfies on your behalf. The clean GAAP-basis books Puzzle produces make that filing far easier, which is the point of pairing the two.

The R&D credit makes the gap concrete. Early startups claim it on Form 6765 to offset payroll tax. Deciding which engineering salaries and contractor costs are qualified research expenses is an analysis, not a category. A person reads the work, applies the rules, and defends the number if it is ever questioned. Software cannot make that call.

When will a startup outgrow software-only books?

The moment someone outside the company has to rely on your numbers. A priced round, a credit facility, a first audit, or an acquisition all demand GAAP financials a human stands behind. SEC rules under Regulation S-X require filings to be prepared in conformity with U.S. GAAP and treat non-conforming statements as presumed misleading — and registration brings audited financials into the picture.

Real-time books are a strong foundation. They are not, by themselves, the audit-grade statements a diligence process expects.

This is the case for software plus a team rather than software alone. Puzzle keeps the ledger current and the data honest. A team owns the close, the GAAP estimates, the board narrative, and the tax position. In the engagements Kevin Cahill and the Debit & Co. team run, that division is what lets a founder trust the dashboard on a Tuesday and still pass diligence in a quarter.

Our Puzzle accounting partnership is built around that exact split, and for later-stage finance leadership our outsourced CFO services carry the judgment layer further.

Frequently asked questions about Puzzle and a human accounting team

Does Puzzle replace an accountant?

No. Puzzle automates bookkeeping, categorization, and real-time dashboards, and it does that very well. The estimates, the close sign-off, board reporting, and tax all need a human accountant. The strong setup runs both together.

Does Puzzle produce GAAP financials?

Puzzle keeps cash and accrual books from one data model, so you get a GAAP-basis accrual view. The judgment-heavy GAAP areas, ASC 606 revenue and ASC 718 stock comp, still need a person to set the estimates and own them.

Can Puzzle close my month for me?

Puzzle cuts the data-prep time up to 50 percent, which shortens the close. It hands a controller a clean, current ledger to close from — reconciliation, accruals for delivered-not-invoiced work, and the final certification stay with that person.

Does Puzzle file my taxes?

No. A corporation still files Form 1120, and credits like the R&D credit on Form 6765 require a human to decide which expenses qualify. Puzzle’s clean books make the filing easier, but they are not the filing.

When do I need a human team on top of Puzzle?

The moment an investor, lender, or auditor relies on your numbers. A priced round, a credit facility, or a first audit all need GAAP financials a person stands behind. As of 2026, that threshold arrives earlier than most founders expect.

A report that lands on the 25th is a history lesson, while a report that lands on the fifth business day is a steering wheel you can still turn. Most month-end work fails for the same reason. The close runs as an improvised list of chores instead of a fixed sequence, so the numbers arrive late, shift after you read them, and lose your trust.

In the engagements run by senior controller Aaron Ressel and the Debit & Co. team, the close that earns trust is boring on purpose: the same steps, in the same order, finished by the same business day.

What is a month-end close, and why does the order matter?

A month-end close is the sequence that turns a month of raw activity into financial statements you can decide on. It matters in this order because each step depends on the one before it. You cannot trust the income statement until accruals are booked, and you cannot book accruals until cash and balances reconcile. Run it out of order and you redo work. Run it in order and the close flows in one direction.

The goal is a clean set of books: financials that are complete, accurate, and final, early enough in the following month to still drive a decision. The accrual basis is the reason the sequence exists.

Under GAAP, transactions are recorded in the period their effects occur, even when the cash moves in a different month. That single rule separates real books from a bank-balance narrative, and it is why cutoff and accruals sit in the middle of the checklist rather than the end.

What are the steps in a month-end close checklist?

Six steps, run in order. Reconcile the real-money accounts, set cutoff and clean AP/AR, record accruals and deferrals, review the balance sheet, review the P&L, then lock the period. The table below pairs each step with what it catches and when to run it. We work this in QuickBooks Online, but the order holds in any system.

Step	What it catches	Timing
1. Reconcile cash, cards, and debt	Missing, duplicated, or miscleared transactions; a wrong cash balance that corrupts everything downstream	Days 1–2 (bank feeds reconcile rolling, pre-close)
2. Set cutoff, clean AP and AR	Bills and revenue booked in the wrong month; expenses incurred but unrecorded	Day 2
3. Record accruals, deferrals, prepaids, depreciation	Cash-basis distortion, with costs and revenue landing in the wrong period	Day 3 (recurring entries templated ahead)
4. Review the balance sheet, account by account	Stale prepaids, unswept clearing accounts, suspense balances, errors that hide off the P&L	Day 3–4
5. Review the P&L against prior month, prior year, and plan	Miscategorized entries and missed accruals surfacing as variance	Day 4
6. Lock the period, issue statements	Numbers that quietly change after the fact; a close that was really a draft	Day 5

Why are reconciliations the first step?

Because nothing downstream is trustworthy until cash ties. Every bank account, every credit card, and every loan or line of credit gets reconciled to its statement. The balance in the books then matches what the institution reports for that month.

This is the step most often skipped or half-finished, and it quietly corrupts everything after it. A wrong cash balance makes your margins wrong and your runway wrong, and you would not know it. Reconcile to the statement, not to a number that looks about right, and chase every unmatched line first.

How does cutoff change the picture?

Cutoff draws a hard line in time and holds it. It decides which transactions belong to the month that just ended and which belong to the next one. Review accounts payable for bills tied to work delivered in the period, even if they are unpaid, because the expense belongs to the month it was incurred.

Do the same with receivables: revenue earned this month counts this month, whenever the customer happens to pay. Cutoff is what separates accrual-quality books from a glorified checkbook register, and a lot of the real picture lives there.

Accruals, deferrals, and prepaids finish the job that cutoff starts. Accruals capture costs incurred but not yet billed, along with revenue earned but not yet invoiced. Deferrals spread an annual software bill paid in January across the months it actually covers.

Prepaids and depreciation spread cost over the periods they serve. None of it is glamorous, yet it is the difference between a P&L that reflects how the business performed and one that merely narrates the bank account.

What does a good month-end close look like?

A good close is complete, repeatable, reviewed, and locked. Complete means every account reconciled and every accrual booked, with nothing left to revisit later. Repeatable means the same steps in the same order, driven off a written checklist so the close does not live in one person’s head.

Reviewed means someone who knows what the numbers should be doing actually walked the balance sheet, rather than letting the software generate it unread. Locked means the period is closed, so the figures cannot shift underneath you after the fact.

Speed is the fourth trait, and it is measurable. The typical organization closes its monthly books in 6.4 calendar days, while top-quartile performers close in 4.8 days or fewer, a benchmark APQC drew from 2,300 organizations.

As of 2026, the practical SMB standard is a fixed business day: close by Day 5, every month. The specific day matters less than the commitment behind it. “Close by the fifth” is a standard, whereas “whenever the books are ready” is not.

How do fast teams shorten the close?

They move work earlier. Much of the close, including recurring journal entries, standard allocations, reconciliations, and routine data checks, can run before month-end rather than waiting for Day 1.

The gap is wide: on period-end management reports, APQC puts the top quartile at 6 days, the median at 10, and the bottom quartile at 15. Bottom performers take roughly 2.5 times as long, mostly because they batch everything into the post-close crunch instead of reconciling cash and templating entries steadily through the month.

Review structure carries the rest. Books reviewed by one person leave a single point of failure, and the work degrades the moment that person is busy, on vacation, or gone. The federal internal-control standard makes the principle explicit. Management should segregate incompatible duties across authority, custody, and accounting to help prevent fraud, waste, and abuse (GAO Green Book, paras 10.12–10.14).

Small teams rarely have the headcount to split every duty, and the same standard says so. Where segregation is not practical, management should design compensating controls: a second-set-of-eyes review, an approval threshold, or a monthly close packet a controller signs off.

What are the most common month-end close mistakes?

The failures repeat across the businesses we see. Reconciliations skipped or rushed, so the foundation is unstable. No real cutoff, so revenue and expenses bleed across months and trends read as noise. Accruals ignored, leaving cash-basis books dressed up as accrual. A balance sheet never reviewed, so errors compound for quarters before anyone notices. No locked period, so figures keep shifting and nobody can say what last month actually was.

The most common mistake of all is no checklist. Without one, the close depends on a single person recalling every step, so it quietly breaks the first month they are out. When month-end becomes a scramble of late, shifting numbers you no longer trust, the work has usually outgrown bookkeeping.

A disciplined close is exactly the recurring, judgment-heavy work that outsourced controller services are built to own, and it anchors our small business accounting services. The checklist is the standard; running it the same way, on time, every single month is the part worth handing to someone whose job is to never miss it.

Frequently asked questions

How many days should a month-end close take?

The typical company closes in 6.4 calendar days and top performers in 4.8 or fewer, per APQC’s benchmark of 2,300 organizations. For most SMBs, the workable target is a fixed business day: close by Day 5 of the following month, every month.

What is the difference between a close and just categorizing transactions?

Categorizing the bank feed is one input. A close is the full sequence, from reconciling and setting cutoff to booking accruals, reviewing both statements, and locking the period, that produces financials final enough to decide on. Without cutoff and accruals, you have a cash-basis approximation, not a close.

Why lock the period after closing?

Locking stops the numbers from changing after you have read them. When last month’s reports can be edited indefinitely, you never had a close; you had a draft. Locking the period in QuickBooks Online makes each month a fixed record you can compare against.

Can a small team close the books without segregation of duties?

Yes, with compensating controls. The GAO Green Book notes that where splitting duties is not practical, management should design alternative controls to cover the same risk. A documented second review, approval thresholds, or a close packet a controller signs off all qualify.

By the time a business hits roughly 30 people, the same person who enters the bills often pays them and reconciles the bank account too. The books still close — eventually. Last month closed on the nineteenth.

The owner reads a profit figure and cannot say what moved it, where it landed against plan, or whether anyone checked that what was billed matched what was delivered. The bookkeeping is fine. The oversight around it never scaled.

That gap has a name. It is the work a controller owns and a bookkeeper does not, and most operators feel it long before they can describe it.

In the engagements Aaron Ressel runs, the upgrade signal is almost never a revenue number. It is a slow close, undefended controls, and decisions made on a P&L that answers “what happened” but never “what now.”

When does a business outgrow a bookkeeper?

You outgrow a bookkeeper when the work shifts from recording transactions to governing them — usually somewhere between 5 and 80 employees. The common rule of thumb puts it near $5M in revenue, but revenue is a weak signal.

The real triggers are structural: more people touch the money, more decisions ride on the numbers, and the close starts slipping. A 12-person SaaS business on a single product can run on clean bookkeeping. A 40-person firm with multiple revenue streams, a lender covenant, and three people in the payment workflow cannot.

Bookkeeping and controllership are different jobs. One keeps the ledger accurate. The other turns that ledger into something you can steer by and puts guardrails around how the money moves. The second job stops being optional once a single mistake — a misclassified period, an unbooked accrual, an unapproved payment — costs more than the oversight would.

What does a controller own that a bookkeeper doesn’t?

A bookkeeper owns accurate transactions. A controller owns the system around them: a repeatable monthly close, segregation of duties, GAAP-aligned books, and management reporting built for decisions. The bookkeeper keeps doing the day-to-day; the controller sits above it and is accountable for the integrity of the whole.

Dimension	Bookkeeper	Controller
What they own	Transactions, categories, bank reconciliation, AP/AR data entry in QuickBooks Online	The monthly close, accruals, GAAP adjustments, internal controls, and management reporting
Question they answer	“What happened last month?”	“What moved it, where did it land against plan, and what’s the move?”
Close discipline	Records as data arrives; no locked date	Reconciliations, accruals, and a hard close by a fixed day each month
Controls	One person often enters, pays, and reconciles	Separates authorize / record / review so no one controls a full transaction
Reporting	Tax-ready P&L and balance sheet	Profitability by line, client, or location, with commentary tied to a decision
Typical cost	$300–$2,000/mo for a bookkeeper	$161,700 median full-time wage, or a fraction of it outsourced

How fast should the monthly close be?

Use the APQC benchmark as your yardstick. Across roughly 2,300 organizations in APQC’s General Accounting benchmark, the median monthly close takes 6.4 calendar days. The top quartile closes in 4.8 days or fewer; the bottom quartile takes 10-plus.

If your books routinely close past two or three weeks — or never formally close at all — you are well outside the bottom quartile and operating without close discipline.

The cost of a slow close is decisions made on stale numbers. The December close that lands in mid-January arrives three weeks into Q1, after the pricing call and the hiring call are already made.

A controller pins the close to a fixed date so every report rests on the same locked foundation. This is the core of the Continuous Close Method™ we run: most of the close work happens before month-end, not after it.

Why does segregation of duties matter at 5–80 employees?

Because one person controlling a full transaction is a control failure, not a convenience. Federal internal-control standards are explicit. Key duties should be divided so that no single individual controls all key aspects of a transaction: separating who authorizes, who records, and who reviews, plus custody of the related assets. The GAO Green Book states this as Principle 10, and it is built on the same COSO framework that underpins public-company internal-control rules.

At small scale, the bookkeeper-does-everything setup is not a character problem; it is a structure problem that quietly leaks money or invites error. The stakes are real even outside public markets.

Among U.S. public companies, serious “Big R” restatements run at about 3% of companies per year — the kind of failure controller-level review exists to catch before it reaches a statement. A controller designs the approval steps and review routines that keep an honest business honest.

How much does a controller cost, and how do you add one without a full-time hire?

A full-time controller is expensive, and most growing businesses do not have 40 hours a week of controller work. The U.S. Bureau of Labor Statistics puts the median wage for financial managers — the category that includes controllers — at $161,700 a year as of May 2024, before benefits and payroll taxes. A 5–80-employee business usually has 10–15 hours a week of genuine controller work, not a full seat.

That mismatch is why so many companies either overpay for an underused full-time hire or keep limping along with no oversight at all. The cleaner path is fractional. With outsourced controller services, you get the close discipline, the controls, the GAAP-aligned books, and the management reporting scaled to the hours the business actually needs.

It layers on top of your existing small-business bookkeeping: the day-to-day stays where it is, and the controller sits above it, owning the close, the controls, and the reporting. You get the oversight of a finance department without building one.

You’ve outgrown a bookkeeper when…

If several of these are familiar, the question is not whether you need more oversight — it is how to add it. The checklist below maps to the four things a controller owns.

• Your monthly close runs past 6.4 days, or the books never formally close and you work off a moving target.
• One person enters bills, pays them, and reconciles the account, with informal approvals and no one checking billed-against-delivered.
• You make pricing, hiring, or product calls on instinct because the reports won’t show profit by line, client, or location.
• A lender, investor, or buyer asks for GAAP statements and your cash-basis shortcuts won’t hold up.
• No one builds a budget, watches cash forward, or flags the tight month before it arrives.
• You have more revenue streams, more people touching money, and bigger decisions than your accounting was built for.

Common questions about upgrading to a controller

Do I replace my bookkeeper when I add a controller?

No. The bookkeeper keeps doing accurate day-to-day work in QuickBooks Online. The controller sits above it and owns the close, the controls, and the reporting. The two roles complement each other; one records, the other governs.

What revenue means I need a controller?

There is no hard threshold. A common rule of thumb is around $5M. The better signals are operational: a close that slips past the 6.4-day median, one person controlling a full transaction, multiple revenue streams, or a lender requesting GAAP statements.

How much does a fractional controller cost versus a full-time one?

A full-time controller carries a $161,700 median wage (BLS, May 2024) plus benefits. A fractional or outsourced controller is priced to the 10–15 hours of controller work a 5–80-employee business actually generates, so you pay for oversight without the full seat.

What is segregation of duties, and why does it matter?

It means no one person authorizes, records, and reviews the same transaction. Federal internal-control standards (the GAO Green Book, Principle 10) require dividing those duties so a single individual cannot control a full transaction. A controller designs that separation into your workflow.

The monthly close has become the most misunderstood ritual in finance. Operators believe a slow close is a staffing problem. Investors believe it is a discipline problem. In practice, it is almost always a design problem — specifically, a series of compounding bottlenecks no one ever stopped to map. This piece walks through the seven we encounter most often at $5M–$50M B2B firms, and what a 7-day close looks like once they are removed.

The 22-day baseline

When we onboard a new client at the $5M–$50M revenue band, the monthly close runs 22 days on average. Some firms run 18. Some run 30. The number itself is less interesting than the pattern: the close is not a single bottleneck. It is a chain of small frictions that compound, and the cumulative latency is what produces the three-week timeline.

The cost of this is rarely measured. Owners and CFOs speak in vague terms about “not getting reports until the 20th.” What they mean, when pressed, is that decisions in weeks one through three of each month are made on intuition. Pricing decisions. Hiring decisions. Vendor decisions. Cash-deployment decisions. These are not minor. They compound across the year into 7–12% of operating margin in our experience.

The symptoms we see most frequently:

• Monthly statements arrive in the third week of the following month, sometimes the fourth.
• Variance from prior period is impossible to explain in real time — only retrospectively, two months later.
• Owners begin operating from the bank balance instead of the income statement.
• The finance team spends more time reconciling than analysing.
• Year-end requires a full reconstruction because monthly closes were never actually closed.

Each of these is downstream of the same structural problem: there is no single owner of the close, no fixed cadence, and no operating definition of “done.” What follows is a tour of where the time actually goes.

---

Bottleneck 1 — Subledger lag

Where bookkeeping debt accumulates

Every close starts with subledgers: accounts payable, accounts receivable, payroll, fixed assets, inventory. In a 22-day close, these subledgers are themselves running 5–7 days behind by the time the general ledger close kicks off. The team is not closing the month — they are catching up on the month before they can close it.

The fix is rarely “hire more bookkeepers.” It is almost always a process change: continuous subledger maintenance throughout the month, not in a batch at month-end. Bills get coded the day they arrive. AR gets reconciled weekly. Payroll integrates directly into the GL, not via a copy-paste summary. The general ledger close becomes a roll-forward, not an excavation.

Benchmark of subledger lag we typically see at clients before engagement, versus what is achievable in the first 60 days:

Subledger	Before (days lag)	After 60d (days lag)
Accounts Payable	5–7	0–1
Accounts Receivable	4–6	1–2
Payroll integration	Manual JE	Direct integration
Bank feed sync	Weekly batch	Daily auto-pull
Fixed asset roll-forward	Once at year-end	Monthly

The point is not that any one of these is hard. None of them is. The point is that all five compound, and each day of subledger lag tacks a day onto the GL close.

Bottleneck 2 — Bank reconciliation chaos

Bank reconciliation should be the cleanest part of the close. Cash is auditable, dated, and deterministic. In practice, at most $5M–$50M firms, bank reconciliation absorbs three to five days of close time because the underlying chart of accounts is misaligned with the bank feed structure, and matching rules are negotiated post-hoc rather than codified.

The fix is a set of explicit auto-categorization rules. When the bank feed lands, 80% of transactions should self-categorise. The remaining 20% — the judgment calls — get human review. Sample of a categorisation rule we typically encode for a $20M services firm:

if vendor LIKE 'STRIPE%' AND amount > 0
  category = 'Revenue — Stripe deposit'
  subledger = 'AR sweep'
  approval_required = false

elif vendor LIKE 'STRIPE%' AND amount < 0
  category = 'Processing fees'
  subledger = 'COGS — payment processing'
  approval_required = false

elif vendor LIKE 'GUSTO%'
  category = 'Payroll'
  subledger = 'Payroll JE'
  approval_required = false

else
  flag_for_review()

Once 80%+ of the cash feed is automated, the reconciliation moves from a multi-day exercise to an exception-handling task that fits comfortably inside a single afternoon.

---

Bottleneck 3 — Accruals as art form

The judgment-call problem

Accruals are where the close stops being procedural and starts being judgmental. How much sales commission to accrue? How much of the consulting retainer is deferred revenue versus earned? How do you treat the unbilled but delivered work-in-progress on the design engagement? Each of these questions has an answer, but the answer is rarely written down. So every month, someone re-derives it.

Re-deriving accrual logic each month is what we call tacit-process debt. The team is not doing the work slowly; they are reinventing the framework. When the senior bookkeeper resigns, the framework leaves with them. The next person re-invents it again, slightly differently. Variances appear. Audit questions appear. Investor questions appear.

The point of a close calendar is not to do the work faster. It is to stop re-deriving the same answers every month.

Debit & Co operating principle

The fix is documented accrual policy. Not a textbook. A two-page operating doc that lists every recurring accrual, the trigger that causes it, the formula, the reviewer, and the GL coding. We call it the Continuous Close Method playbook. The first version takes a week to write. The hundredth month it saves three days per close.

Bottleneck 4 — The review-loop hell

Most $5M–$50M close cycles have at least four review checkpoints, and the gap between checkpoints is where the days go. Bookkeeper produces draft. Controller reviews. Items go back. Bookkeeper revises. Controller re-reviews. CFO or fractional CFO reviews. Items go back again. Owner reviews the final. Items go back a third time. Each cycle introduces 24–48 hours of latency.

The standard review loop, as we observe it before engagement:

1. Draft trial balance produced by bookkeeper (day 5–7).
2. Controller-level review (day 9–11) — surfaces 15–25 items.
3. Bookkeeper rework (day 12–14).
4. Second controller pass (day 15–16) — surfaces 5–8 residual items.
5. Final rework (day 17–18).
6. CFO / fractional CFO review (day 19–20).
7. Owner review (day 21).
8. Distribute (day 22).

Two interventions collapse this. First, structured review templates so that the controller reviews deterministically, not exploratorily. Second, daily mini-reviews during the month so the month-end review isn’t a debugging session. The close becomes a sign-off, not an investigation.

---

Bottleneck 5 — Spreadsheet handoffs

Spreadsheets are not the enemy. Spreadsheets at the close handoff are. When the controller exports the GL to Excel, performs an analysis, and emails it back to the bookkeeper to update the GL, two problems appear: version drift and round-trip latency. The bookkeeper may have changed entries in the GL after the export. The reconciled spreadsheet now references stale data. Three days disappear into reconciliation of the reconciliation.

Failure modes we see in spreadsheet-mediated closes:

• Email-as-database. The current version of the working trial balance lives in someone’s inbox attachment.
• Copy-paste fidelity loss. Numbers move between sheets with manual paste, dropping formulas mid-flight.
• Reviewer not in source. The controller reviews the export, not the live GL — comments don’t propagate back.
• Multiple sources of truth. By day 15, three working files have meaningfully different totals. No one is sure which is current.
• No version history. A breaking change is impossible to trace because the “old version” was overwritten last Thursday.

The fix is not the absence of spreadsheets — it is the elimination of round-trips. Review happens in the GL itself. Spreadsheets become outputs for analysis, not the medium of revision.

Bottleneck 6 — Chart of accounts drift

Over five years of operation, every chart of accounts drifts. New revenue streams add new accounts. Departments split. Someone creates “Misc. Operating Expense” because they didn’t know where else to code it. By year five, the chart has 240 accounts, of which 70 are inactive, 30 are near-duplicates, and 12 contain coding errors no one has had time to fix.

A drifted chart of accounts adds latency to the close in three ways. First, transaction coding requires judgment calls that should be automatic. Second, financial statements produce categories that don’t match management’s mental model, requiring reformat work each month. Third, the GL reconciliation process spends time matching accounts that should be merged.

After Debit & Co rationalised our chart of accounts, what used to be a three-day reformat exercise became a one-page report we look at the morning of day six.

CFO, $28M B2B services firm (anonymised)

A chart cleanup is one of the highest-leverage interventions we do. It takes about ten hours of concentrated work. It saves two to four days every month thereafter, indefinitely.

Bottleneck 7 — Confusing the dashboard with the close

The final structural error: assuming that a real-time dashboard is a close. It isn’t. Dashboards show transactional state. A close produces accrual-basis financial statements with judgment calls baked in. The two operate at different layers of the stack, and they answer different questions.

A dashboard answers: what is the bank balance right now? A close answers: what is the operating profit for the month, after recognising deferred revenue, accruing payroll, and matching vendor expenses to the periods they belonged to? The dashboard is necessary — it tells you cash position daily. But it doesn’t replace the close.

When operators replace the close with the dashboard, they get fast feedback but inaccurate margin. Pricing decisions get made off cash inflow without recognising that 30% of that inflow was deferred revenue, or that the COGS hasn’t hit yet. Decisions are confident and wrong — the worst combination.

---

What a 7-day close looks like

A 7-day close is not a faster version of a 22-day close. It is a structurally different process. The work compresses from twenty-two sequential days into seven by being done throughout the month, not at the end of it. By month-end, 80% of the work is already done.

Day	22-day close (before)	7-day close (after)
Day 1	AP catch-up	Daily cash close + open subledgers
Day 2	Continued AP	AR aging review + commission accrual
Day 3	AR catch-up	Inventory + WIP roll-forward
Day 4	Payroll JE	Trial balance review (controller)
Day 5	Bank rec start	CFO-level analytic review
Day 6	Bank rec continued	Owner sign-off
Day 7	Bank rec finalisation	Distribute
Day 8–14	Accrual derivation	—
Day 15–18	First review pass	—
Day 19–22	Subsequent review passes + distribute	—

Three structural shifts make this collapse possible. One: the subledger work is continuous, not batched. Two: the chart of accounts is rationalised and stable. Three: review is deterministic, not exploratory. Each is a discipline. None is a headcount lift.

What changes when you have a 7-day close, beyond the calendar?

• Pricing decisions become real-time. When the income statement lands on day 7, the team has 21 days of the next month to act on what it says.
• Cash deployment is informed. The owner knows whether margin is intact before authorising the next major spend.
• Board meetings stop being archaeological digs. The numbers in the deck reference the current month, not three months ago.
• The finance team starts to do finance. Variance analysis, forecasting, capital allocation — the work that compounds.
• Audits become trivial. When monthly closes are actually closed, year-end takes days, not weeks.

The math

Compressing the close from 22 days to 7 days is a 68% reduction in close cycle time. The compounding effect is larger than the headline number suggests. Three multiplier effects matter.

1. Decision lead time. An additional 15 days per month of operating decisions made on real data, not intuition. Compounded over 12 months: 180 decision-days reclaimed.
2. Team capacity. 15 days per month freed from reactive close work means 180 days/year released for forward-looking analysis — effectively a second senior hire at no incremental cost.
3. Margin recovery. Operators making real-time-informed pricing and vendor decisions recover 7–12% of operating margin within 12 months. On a $20M revenue book, that is $1.4M–$2.4M annually.

The cost side is small. The interventions we’ve described — subledger continuity, categorisation rules, accrual policy doc, review templates, chart cleanup — take roughly 90 days of focused work. The ROI hits month four and compounds from there.

The close is not a finance problem. It is an operating system for the business. When the operating system runs slowly, every decision downstream of it runs slowly too.

Debit & Co takeaway

If your close is taking longer than ten days at $5M–$50M revenue, the cost is rarely visible because it shows up in the decisions you didn’t make, the variances you couldn’t explain, and the investor questions you couldn’t answer. The interventions are knowable. The discipline is buildable. The difference, after a quarter, is operationally transformative.

—

If you are interested in mapping where your own close is spending time, we offer a 30-minute diagnostic call. We will walk through your current cycle, identify the top three bottlenecks specific to your business, and tell you honestly whether it is a fit for our Continuous Close Method™ or whether the work belongs in-house. Book a discovery call.