Ask a finance and accounting team how the month went and you'll probably get volume: 1,200 journal entries posted, 340 vendor invoices processed, 85 wires sent. Those numbers are real and they reflect effort. But they don't answer the question a CFO actually cares about: is this team delivering reliable information, on time, without burning hours?
A team can process more documents than ever and, at the same time, close later, pile up unreconciled items, and redo half of last week's work. Volume measures activity. Efficiency shows up somewhere else: how long it takes data to reach a trustworthy state, how much is left unresolved, and how much work gets done twice.
1. Why Counting Tasks Isn't Enough
Volume metrics have three problems. First, they track the business, not the team: when sales grow, entries grow, and nobody has improved anything. Second, they reward visible effort over outcomes: a reconciliation redone three times counts as three tasks. Third, they don't point to the problem; they only tell you there was a lot of work.
Useful metrics follow a different logic. They measure flow (how long something takes to go from open to resolved), stock (what's still open, and since when), and quality (how much has to be fixed). Six metrics are enough to cover all three.
2. Six Metrics That Actually Describe the Team
Useful metrics follow a different logic. They measure flow (how long something takes to go from open to resolved), stock (what's still open, and since when), and quality (how much has to be fixed). Six metrics are enough to cover all three with practical, operational definitions:
| Metric | Formula / Measurement | What It Reveals | Warning Sign |
|---|---|---|---|
| 01 · Flow Days to close | Business days from month-end to approved financials |
How long it takes for the period numbers to be ready and usable for business decisions. | Wide month-to-month swings, or a specific close stage that consistently runs over schedule. |
| 02 · Stock Reconciliation rate | Reconciled items ÷ total (by count and dollar value) |
The percentage of transactional volume fully validated against banks, customers, or vendors. | High match rate by count but low by value: minor items are cleared while material items stall. |
| 03 · Stock Aging of open items | Open items broken down: 0–7, 8–30, 31–60, and 60+ days |
Whether unresolved items turn over cleanly or settle permanently at the bottom of the pile. | The 30+ day bucket expands while the overall reconciliation percentage appears stable. |
| 04 · Quality Unexplained variances | Σ |variance| with no documented root cause at close |
The absolute dollar volume lacking audit backup, even when the balance sheet ties out. | Suspense or miscellaneous adjustment accounts carrying growing balances month over month. |
| 05 · Flow Time to resolution | Days from exception detection to close: median & p90 |
How fast the department converts an operational breakdown into an accounted resolution. | Short median combined with a very long p90: blockers originating outside the finance team. |
| 06 · Quality Rework rate | Reopened items or reversed journal entries ÷ total processed |
How much team capacity is spent correcting past entries rather than executing current work. | Repeated errors clustered around specific invoice types, counterparties, or preparers. |
Three definition choices separate a useful metric from a decorative one:
- Measure the close by step, not just in total. Breaking out subledgers, reconciliations, accruals, review, and reporting shows you which step stretches. "We close in seven days" says nothing about where the seven days go.
- Add variances in absolute value. A +$40,000 variance and a −$38,000 variance net to $2,000, which looks immaterial. In reality, that's $78,000 nobody can explain. Netting is the most common way problems get buried in a reconciliation.
- Look at the p90, not just the average. With resolution times, a handful of items that take weeks distort the average and hide inside it. The median shows the typical case; the 90th percentile shows the case that holds up the close.
"A team that processes more documents than last year isn't necessarily more efficient. It may just have more volume and the same bottleneck, only bigger."
3. Aging Tells You What the Percentage Hides
The reconciliation rate is the most commonly reported metric and, on its own, the most misleading. A team can report 94% three months in a row while the problem gets worse underneath. What changes isn't how much is open, but how old it is:
What happens in a case like this is predictable: the team clears the easy, recent items first, because that's what keeps the percentage up, and the hard ones sink to the bottom of the pile. As we found when looking at where the time in a bank reconciliation actually goes, a small pile of difficult items can eat most of the hours. The older an item gets, the harder it is to rebuild its context, and the more likely it ends up as an unexplained adjustment.
4. How to Read the Metrics Together to Find the Bottleneck
A single metric describes a symptom. A combination of metrics points to the cause. These are the most common patterns and what they usually mean:
| Observed Pattern | Likely Bottleneck | First Move |
|---|---|---|
| Late close + elevated rework Reconciliation rate looks high, but errors surface at the deadline. | Late review & verification Transaction processing is quick, but quality checks occur only during month-end crunch, forcing teams to back out entries late. | Move validations upstream: establish automated intake checks at entry, rather than manual gatekeeping at close. |
| Steady reconciliation + climbing aging Overall match rate holds flat, but the 30+ day aging bucket is expanding. | Unassigned or orphaned exceptions The team quickly clears recent and straightforward items to protect the top-line KPI; difficult items sit without an owner. | Assign explicit accountability per aging bucket and review every item past 30 days during a weekly 15-minute sync. |
| Low median + extreme 90th percentile Most exceptions close in 24–48 hours, but a persistent minority takes weeks. | External cross-departmental dependencies Exceptions stall while waiting on feedback from Sales, customer remittance slips, or banking inquiries. | Agree on cross-functional turnaround SLAs and standardize incoming dispute documentation templates. |
| Unexplained variances trending up Suspense, clearing, or miscellaneous adjustment balances rise monthly. | Unwritten business rules or corrupted source feeds Application logic lives inside individual heads, or ERP export files arrive with missing or ambiguous identifiers. | Document clear matching rules and variance tolerances; clean and standardize ERP report output schemas. |
| Reconciliation stalled until month-end Very little progress is tracked during the first 25 days of the period. | Unnecessary monthly batching Transactional volume that could easily be validated continuously gets bunched into a high-pressure month-end sprint. | Shift high-volume bank and subledger reconciliations to a daily or weekly cadence to flatten close workload. |
The fourth pattern deserves special attention because it's often mistaken for a capacity problem. If unexplained variances keep growing, adding headcount rarely fixes it. In many cases the cause is how the data arrives, as we covered in how to prepare your financial data for automation without changing your ERP, or tax rules that create systematic differences, as happens with tax withholdings in customer accounts.
5. How to Start Measuring Without a BI Project
You don't need a new tool to start. You need a minimal log and the discipline to keep it up. A reasonable plan:
- Start an exceptions log. A shared spreadsheet with seven fields: date flagged, process, exception type, amount, owner, date resolved, and whether it was reopened. That gives you time to resolution, aging, and most of your rework metric.
- Log close dates by step. One line per step with a start and end date. Within two or three closes, the step that always runs long becomes obvious.
- Build a three-month baseline. There's no universal benchmark that fits every company; what matters is the trend against your own starting point. Once you have the baseline, set your own thresholds: what age or what unexplained amount triggers a review.
- Give every metric an owner and a threshold. A number that gets looked at but never triggers action is information, not a control. It's the same distinction we drew in the real difference between a dashboard and a control.
- Review them in 30 minutes a month. After each close, look at all six metrics together, pick one pattern to tackle, and check next month whether it moved.
One caution: these metrics describe the process, not individual performance. If they're used to single people out, the team learns to game them: match the easy items, net the variances, avoid reopening what was wrong. Positioning them as a way to find where work gets stuck, not who to blame, is part of what we discussed in how to talk to your team about automation.
Once part of the process is automated, measurement stops depending on the spreadsheet: every item is logged with its status, who reviewed it, and when. In the projects we run, that log is part of the audit dashboard, and exceptions go through human-in-the-loop review before they're closed, so time to resolution and rework come straight out of the workflow instead of being reconstructed at month-end.
Measure to Decide Where to Intervene
"Is the team running well?" gets a concrete answer once you swap volume for six numbers: how long the close takes and in which step, how much is reconciled by count and by value, how old the open items are, how much has no explanation, how long an exception takes to resolve, and how much gets redone. None of them is enough on its own. Together, and compared against your own baseline, they point to the exact spot where the process gets stuck, which is rarely where the team puts in the most hours and almost always where the work sits waiting. If you'd like to see what these metrics show for your operation, reach out and we'll walk through it on an actual close.