KPI Reporting for Lower-Middle-Market Companies: Picking the 8 That Matter

How many KPIs should a board report?

Eight to twelve. Fewer and the board cannot see the business; more and nothing is prominent, which produces the same result as reporting nothing. The constraint is attention rather than data — a board meeting has a fixed number of minutes, and thirty metrics divides that attention into portions too small to act on.

The selection test is a single question: would a plausible movement in this metric change a decision this board makes? If a 15% swing would prompt no action, the metric belongs in an operating review, not the board package.

Definitions matter more than selection

Before choosing metrics, settle what they mean. This sounds procedural and it is the most valuable hour in the exercise.

Where sales counts bookings on verbal commitment and finance counts them on signature, the board is shown two truths and every discussion becomes an argument about the definition rather than the result. Where "recurring revenue" includes a contract with a 90-day termination clause, the metric is describing something less durable than the word implies.

Each metric needs a written definition covering the formula, the data source, the owner, the frequency, and the treatment of edge cases. When a definition changes, restate the history and say so in the package. An unannounced definition change is indistinguishable from performance, which is why it corrodes trust in every other number.

The eight that apply to most companies

1. Revenue growth, year over year

Growth % = (Current period revenue ÷ Prior year same period) − 1

Compare against the same period a year earlier, not the prior month, so seasonality does not masquerade as a trend. Report organic and acquired separately for sponsor-owned companies — a platform showing 40% growth of which 34 points came from acquisitions is a different business from one growing 40% organically, and blending them hides the distinction the board most needs.

2. Gross margin

Gross margin % = (Revenue − Cost of sales) ÷ Revenue

Margin is where operational problems appear before they reach the bottom line. Decompose the movement rather than reporting the level. Consider margin falling from 42.0% to 39.6% on $12M of quarterly revenue:

Four different problems with four different owners, where "margin down 240bps" was one undifferentiated fact.

3. Adjusted EBITDA and margin

EBITDA margin % = Adjusted EBITDA ÷ Revenue

Use the credit agreement's definition of adjusted EBITDA, including its caps on add-backs, rather than a management-convenient version. Reporting one EBITDA internally and a different one to the lender guarantees a reconciliation conversation nobody enjoys. Covenant compliance reporting works through where the definitions diverge.

4. Cash conversion

Cash conversion % = Operating cash flow ÷ Adjusted EBITDA

This is the metric that catches profitable companies running out of money. Sustained conversion below roughly 70% means earnings are being absorbed by working capital or capex. A company reporting record EBITDA and 45% conversion is growing itself toward a liquidity problem, and this single ratio is the earliest visible signal.

5. Days sales outstanding

DSO = (Accounts receivable ÷ Revenue) × Days in period

Worked: $8.2M of receivables on $46M of annual revenue.

DSO = ($8,200,000 ÷ $46,000,000) × 365 = 65.1 days

Against 45-day terms, that is twenty days of drift. Each day is worth roughly $126,000 of cash ($46M ÷ 365), so closing the gap by ten days releases about $1.26M. That is a larger sum than most cost programmes produce, and it requires no new revenue.

6. Net leverage

Net leverage = (Total funded debt − Unrestricted cash) ÷ TTM Adjusted EBITDA

Report it against the covenant maximum and express the gap as headroom in EBITDA dollars rather than as a ratio alone: "2.52x against 3.25x, headroom of roughly $1.1M of EBITDA." That framing tells the board how much performance deterioration the balance sheet can absorb, which is the actual question.

7. Backlog or pipeline coverage

Coverage = Contracted or qualified pipeline ÷ Remaining period revenue target

The forward-looking metric on the list, and the only one that says anything about next quarter. Coverage of 1.0x means the target requires everything in the pipeline to convert, which never happens. What good looks like depends entirely on the historical conversion rate — which is why that rate should be reported alongside it rather than left implicit.

8. Revenue or gross profit per employee

Per-employee = Revenue (or gross profit) ÷ FTE headcount

The clearest single measure of whether growth is being bought with headcount. A company growing revenue 25% while this metric falls 15% is adding people faster than it is adding output, and that is a scalability question worth raising before it becomes a margin question.

Metrics worth adding by business model

The eight above are close to universal. Beyond them, add two to four that reflect how the business actually works:

Resist adding a metric per function to satisfy internal politics. A KPI set assembled so every department is represented reports the organisation chart rather than the business.

Presenting them

Each metric needs three things beside the current value: a target, a trend, and a comparison.

The value alone says nothing. DSO of 65 days is a problem if the target is 45 and an achievement if it was 78 last quarter and the target is 60. Twelve months of trend is usually the right window — enough to distinguish direction from noise, short enough to fit legibly.

Colour should encode a value consistently across the whole package. If green means ahead of target on one page, it cannot mean "increased" on the next. That inconsistency is small and it quietly makes the document harder to read than a table with no colour at all.

What to do when a metric is disputed

Escalate it and settle it in one session with the people who own the underlying data. Write the agreed definition down. Restate history onto it. Disclose the restatement in the next package.

The instinct is to report both versions until the disagreement resolves. It does not resolve — it becomes permanent, and the board learns that the company's own numbers are a matter of opinion. One agreed definition, even an imperfect one, is worth considerably more than two defensible ones. The rest of the package structure is covered in the section-by-section build.

Building the definitions document

The definitions document is the artifact that makes the KPI set durable. It is short, it is boring, and it prevents most of the recurring arguments. One row per metric:

The last row does more work than it appears to. A dated change log on each metric means that when a trend line moves, anyone can check whether the business changed or the definition did. Without it, that question is unanswerable a year later, and the usual resolution is to distrust the whole series.

Keep the document with the board materials rather than in a folder somebody owns. A definition nobody can find is functionally the same as no definition.

How often to revisit the set

Once a year, and not more often.

Changing the KPI set quarterly destroys the thing that makes it valuable — the trend. A metric with three data points is not measuring anything, and a board that sees a different scorecard each quarter never develops a sense of what normal looks like for the business.

The annual review should ask three questions of each metric. Did anyone use it to make a decision this year? Has the business changed in a way that makes it less relevant? Is there something we repeatedly asked for ad hoc that should have been on the list?

That last question is the useful one. Recurring ad-hoc requests are the clearest available evidence of a gap in the standing set, and they are already documented in the email trail. When a metric does come off the list, keep reporting it in the appendix for a year rather than deleting it — that way the history remains available if the question returns, which it often does.