The spreadsheet that runs the close
Ask a finance team where their numbers really live and the honest answer is often a spreadsheet. Not the ERP, not the BI tool — the working file where actuals from three systems get pulled together, adjusted, reconciled, and turned into something a board can read.
There's nothing wrong with that. Spreadsheets are the most flexible modeling tool finance has ever had. They're fast, they're universal, and they let a smart analyst build exactly the view they need without waiting on IT. Every finance team on earth runs on them, and that isn't going to change.
The problem isn't the spreadsheet. It's what happens to spreadsheet reconciliation as a company grows. At ten customers, tying ARR to billing to the general ledger is a twenty-minute check. At two hundred customers across annual and monthly billing, multiple entities, and a revenue-recognition policy with real edge cases, that same reconciliation becomes a multi-day effort that a single broken formula can quietly corrupt.
This piece is about the cost of that work — not the visible hours, but the hidden ones. The delayed reports, the executive rework, the erosion of confidence, the audit friction. And it's about why the fix isn't abandoning spreadsheets. It's treating reconciliation as a governance problem, which is what it actually is.
Why spreadsheets stay — and why they get riskier
Spreadsheets survive in finance for good reasons. They're the fastest way to model a new scenario, restructure a report, or answer a board question nobody anticipated. A capable analyst can build in an afternoon what a software project would take a quarter to ship. That flexibility is genuinely valuable, and any honest conversation about reconciliation has to start by acknowledging it.
But the same properties that make spreadsheets flexible make them fragile at scale. A spreadsheet has no enforced structure. Any cell can be overwritten. Any formula can be dragged one row too far. Any assumption can be hardcoded and forgotten. When one person builds and owns the file, they hold the whole mental model, and it mostly works.
Growth breaks that. More customers means more transactions to reconcile. More systems — a billing platform, a CRM, a general ledger, maybe a separate usage database — means more sources that have to agree. More people touching the file means more chances for two versions to diverge. The spreadsheet didn't get worse; the load on it got heavier, and its lack of guardrails started to matter.
That's the pattern: spreadsheets don't fail loudly. They quietly get riskier as the company scales, until one month the numbers don't tie and nobody can immediately say why.
Reconciliation is more than matching numbers
The word "reconciliation" makes it sound like a matching exercise: line up two columns, confirm they agree, move on. In practice, reconciling financial data in spreadsheets is a stack of distinct, error-prone jobs.
- Version control. Which file is authoritative?
Board_Deck_v7_FINAL_v2_updated.xlsxis a punchline because it's real. When the close lives across emailed copies and shared drives, confirming you're working from the current version is its own task — and getting it wrong means presenting stale numbers. - Manual adjustments. Top-side entries, reclasses, one-off corrections. Each is a judgment call typed directly into a cell. They're rarely documented in a way another person can follow, and they don't carry their own audit trail.
- Broken formulas. A reference that shifts when a row is inserted. A
SUMthat stops one line short. A paste that overwrites a calculation with a static value. These errors are silent — the cell still shows a number, just the wrong one. - Conflicting assumptions. The revenue model assumes one churn definition; the board deck assumes another. Both are defensible; they just don't match. When assumptions live in different files, nobody notices they've diverged until the totals don't.
- Disconnected source systems. Billing says one ARR figure, the CRM says another, the GL says a third. Reconciling them by hand — exporting, pasting, aligning customer IDs that don't quite match — is where most of the real time goes.
- Repeated validation. And because none of the above is guaranteed stable, the whole thing gets re-checked every close. Not once — every month, from something close to scratch, because last month's confidence doesn't transfer to this month's file.
That last point is the quiet killer. Reconciliation in spreadsheets isn't a task you complete. It's a task you repeat, in full, every period, because nothing about the process locks in the trust you earned last time.
The costs you don't see on the invoice
The obvious cost of manual reconciliation is hours. That's the least of it. The hidden costs are larger and harder to name in a budget line.
Delayed reporting. When the close depends on manually reconciling several systems, it takes as long as it takes. Board decks go out late, or go out on time but under-baked. Either way, the reporting cadence is hostage to the reconciliation cadence.
Reduced confidence. When you can't quickly trace a number back to its source, you present it with a caveat in your head. Finance leaders know the feeling of putting a figure on a slide they're 90% sure of. That last 10% is a tax on every decision the number informs.
Executive rework. A director asks a question the deck can't answer from its own data. Someone reopens the files, rebuilds a bridge, and circulates a correction. The board's first impression was wrong, and finance spends the next week un-ringing that bell.
Slower decisions. If leadership doesn't fully trust the numbers, they hesitate. Hiring plans, spend decisions, fundraising timing — all slow down when the underlying figures carry an asterisk. Slow reconciliation becomes slow decision-making.
Audit challenges. Auditors want to trace a reported number back to source. When the path runs through manual adjustments and undocumented formulas across several files, that trace is expensive to produce. What should be a quick tie-out becomes a scavenger hunt.
Lost trust. This is the one that compounds. Every unexplained discrepancy, every late correction, every "let me get back to you" chips at the board's confidence in finance. Trust is slow to build and fast to lose, and once a board starts double-checking your numbers, the meetings change character permanently.
None of these show up as a line item. All of them are real, and together they dwarf the hours.
It's a governance problem, not a spreadsheet problem
Here's the reframe that matters. Every issue above — version confusion, silent formula errors, undocumented adjustments, discrepancies that can't be traced — is a symptom of the same root cause. The reconciliation process has no governance.
Governance, in this context, means a few concrete things. That there's one authoritative version of the numbers, not five. That every figure can be traced back to its source. That the same inputs always produce the same outputs. That adjustments are recorded, not just typed. That once numbers are finalized for a period, they don't quietly shift underneath you.
A spreadsheet, on its own, provides none of these by default. You can impose discipline on top of one — locked tabs, naming conventions, a rigorous change log — and disciplined teams do. But that discipline lives in people's habits, not in the tool, so it degrades under deadline pressure exactly when it matters most.
So the goal isn't "stop using spreadsheets." It's "stop relying on spreadsheets to provide governance they were never built to provide." Keep the spreadsheet for what it's great at — modeling, flexibility, ad-hoc analysis — and move the reconciliation and the source of truth onto a foundation that's governed by design.
What a governed foundation looks like
Three properties separate a governed reporting layer from a manual one.
Deterministic calculations. The same inputs produce the same outputs, every time. Run this period's ARR waterfall twice and get an identical result. Determinism is what makes quarter-over-quarter comparisons mean something and what makes a number defensible when someone asks how it was computed.
Traceable outputs. Every figure drills down to its source. If ending ARR is $12M, you can see the movements that built it, and each movement ties back to the underlying customer records — not to a cell someone typed. Traceability turns "let me get back to you" into "here's exactly where that comes from."
Governed reporting. Once numbers are finalized for a period, they're stable. The figure in the board pack is the figure, and it doesn't move because someone refreshed a source after the deck shipped. Governance is what lets you present a number and stand behind it.
Put together, these turn reporting from an act of faith into an act of evidence. The number isn't trustworthy because a smart analyst vouches for it. It's trustworthy because it can be traced, reproduced, and defended.
Where SMPL.ai fits
SMPL.ai is built to give finance teams that governed foundation without taking away the spreadsheet.
SMPL reads and reconciles data from your source systems — billing, CRM, and the general ledger — into one governed operating model. It computes the SaaS metrics a board expects, drawn from a single reconciled base, so ARR, deferred revenue, recognized revenue, cash, and retention all tie to the same source instead of to five separate exports.
Critically, SMPL does not write back to your ERP or general ledger. It reads from your systems of record; it never posts to them. Your GL stays your GL, owned by your team and your auditors. SMPL is a reconciliation and reporting layer on top of your books — not a system that touches them.
The calculations are deterministic and every figure carries its lineage, so when a director asks how a number was built, you can trace it on screen to the contracts underneath. And where AI comes in, it plays a deliberately narrow role: it explains validated financial results rather than generating financial numbers. The narrative describes the movements the model actually computed. It doesn't invent a figure or assert a trend the data doesn't support. The numbers come from reconciled source data; the AI just puts them into words you can defend.
What finance teams can do today
You don't need to replace anything this week to start closing the governance gap. A few practical moves:
- Name one authoritative source per metric. Decide, explicitly, where the real ARR number lives — and make every deck pull from it. Most discrepancies start with two "sources of truth."
- Document your adjustments. Every top-side entry should have a one-line reason recorded next to it. Not for the auditors' sake — for the version of you reconciling this again next quarter.
- Freeze the numbers before you present them. Once a period's figures are finalized for the board, stop editing the file. If something changes, that's next period's revision, tracked as such.
- Reduce the manual hops between systems. Every hand-keyed export is a place errors enter. Wherever you can pull directly and reconcile once, do.
- Make traceability a requirement, not a nice-to-have. For every headline number in the board pack, ask: can I trace this back to source in under a minute? Where the answer is no, that's where your risk concentrates.
These steps don't require new software. They require treating reconciliation as governance — which is the shift that matters, tool or no tool.
See it on your own numbers
The real test is whether your ARR, revenue, and cash can be reconciled to one source and traced back to the contracts underneath — quickly enough to answer a director in the room.
Book a demo and we'll walk that reconciliation on data that looks like yours, so you can see what a governed close feels like next to the manual one.