The monthly business review starts with three versions of headcount. Payroll has one number, the HR system another, and the cost centre report in the ERP a third. Sales brings bookings from the CRM, which do not tie to revenue in the ledger. Operations brings on-time delivery from a system finance cannot see. Much of the meeting goes on reconciling numbers before anyone discusses what to do.
The financial results were never the problem. The ERP holds them, and the close produces them. What the room lacks is the operational KPIs that explain them, built on agreed definitions and sitting next to the financials in one place.
What the ERP cannot give you
An ERP records transactions with a financial effect: invoices, payments, journals, inventory movements. It answers "what happened to the money" well. It was not built to answer "why".
The drivers sit elsewhere. Pipeline, win rates and renewals are in the CRM. Headcount, hires, leavers and hours are in HR, payroll and timesheet systems. Throughput, on-time delivery, yield and capacity are in operations or project tools. The ratios a CFO actually manages with combine the two sides: revenue per employee, cost per order, margin per billable hour, cash conversion against delivery dates.
Most groups try to solve this with a BI tool and a data project, or with a spreadsheet that someone rebuilds every month. The first is often slow. The second depends on one person and rarely survives their holiday.
Operational KPIs need definitions before dashboards
Most of the friction in combining financial and operational KPIs is about meaning, not technology.
Take headcount. It can mean heads or full-time equivalents, at month end or as an average, with or without contractors, interns and people on leave. Each choice is defensible. Mixing them in one ratio is not.
Financial measures have the same issue. APQC's benchmark definition of days sales outstanding uses average gross accounts receivable divided by gross annual sales over 365, and excludes unbilled receivables. On that definition its cross-industry median is 38.0 days. A business that includes unbilled work, or uses period-end rather than average receivables, will produce a different number from the same ledger. Neither is wrong, but a comparison across them is.
The fix is a definitions register that sits above every system.
| Field | Example for "revenue per billable FTE" |
|---|---|
| Definition | Recognised revenue for the month divided by average billable full-time equivalents in the month |
| Numerator source | ERP, revenue accounts in the group chart |
| Denominator source | HR system, employees flagged billable, FTE basis, averaged over the month |
| System owner | Finance for revenue; HR for the billable flag |
| Metric owner | Head of delivery |
| Refresh | Monthly, after close |
| Tolerance | Range agreed with the metric owner; outside it triggers a review |
The data ownership problem. A system owner keeps the data accurate. A metric owner answers for the number and acts on it. They are rarely the same person. Finance can own the register and the calculation without owning every input, but someone must be named for each.
A short list of KPIs by business model
A long dashboard dilutes attention. Pick a few financial measures and the operational KPIs that drive them.
| Business model | Financial | Operational |
|---|---|---|
| Professional services | Gross margin by project, DSO, revenue per billable FTE | Utilisation, realisation, bench time, attrition |
| Subscription software | Recurring revenue, gross margin, cash burn | Net revenue retention, logo churn, pipeline coverage, support backlog |
| Manufacturing and distribution | Gross margin by product, inventory days, cash conversion | On-time in full, scrap or yield, capacity use, order backlog |
| Multi-location services | Margin by location, labor cost as a share of revenue | Sales per labor hour, occupancy or utilisation, staff turnover by site |
The pairings matter more than the list. Each financial line should have at least one operational measure that moves before it does.
The single-view approach
The approach we take with CFOs is a single view built as a Command Center: a management layer that reads each system as it is and does not replace any of them. It applies the definitions register, shows financial and operational KPIs side by side, and marks which are outside tolerance, who owns each one and how long it has been open. It keeps a record of what was flagged and what was done.
Three design choices keep it honest. Read-only access, so the view never changes source data. Every figure traceable to the system it came from. And a visible mismatch when two systems disagree, such as payroll headcount against cost centre headcount, rather than a silent choice of one.
Where AI helps. It can draft commentary that links a financial movement to the operational change behind it, using figures taken from the systems rather than generated. It can flag when a definition breaks, for example a new employee type that the billable flag does not cover. And it can answer a follow-up question with the figures and their sources. A person reviews anything that goes to the board.
What one view will not fix
It will not settle disagreements about definitions. If sales and finance disagree on what counts as a booking, the view will show both until management decides.
It will not make poor source data reliable. If timesheets are filled in late or the CRM is updated before quarter end only, the view shows those gaps faster. Fixing them is a management task.
Frequently asked questions
What is the difference between financial and operational KPIs?
Financial KPIs measure results recorded in the ledger, such as margin, DSO and cash conversion, while operational KPIs measure the activity that drives them, such as utilisation, on-time delivery and pipeline coverage.
Why can't the ERP report operational KPIs?
The ERP records transactions with a financial effect, and most operational data sits in the CRM, HR and payroll, timesheet and operations systems. Combining them needs a layer above the ERP with agreed definitions.
Who should own KPI definitions?
Finance should usually own the definitions register and the calculation, with a named metric owner in the business for each KPI and a system owner for each data source.
How many KPIs should a management team track?
A short list, typically a few financial measures each paired with the operational KPIs that drive them. More than that tends to dilute attention in the monthly review.
Can AI build a KPI dashboard on its own?
No, AI can draft commentary, flag definition breaks and answer questions with sourced figures, but the definitions, owners and tolerances have to be agreed by people first.
Where CFOLogic fits
We design and build single views over the ERP, CRM, payroll and operations systems a business already runs, read-only by default, with a definitions register agreed with the management team. We usually start with a proof of concept covering one business unit and a short KPI list. Our CFO Success Partners page sets out how we work with CFOs on this.