The 8 HR + delivery metrics an IT services company should actually track

The vanity-metrics trap

Most HR dashboards report what's easy to count: headcount growth, training hours, 1-on-1s held, an annual "engagement score" nobody remembers by February. The numbers move, the slides look busy, no decision changes.

The test for whether a metric deserves a slot is simple: it has an owner, and there's a decision it can change this month. If nobody would act differently when the number moves, it's decoration.

At an IT services company there's a second trap: treating HR metrics and delivery metrics as separate worlds. They aren't — your people are the P&L. Utilization is an HR outcome and a revenue driver; attrition is a culture signal and a margin event. The eight below deliberately mix both.

The 8 that matter

1. Billable utilization (net of PTO)

The share of your billable people's available capacity on billable work — billable people only, billable projects only, vacation and sick leave out of the denominator. The single strongest predictor of whether the month makes money. Bad: sustained drift below the mid-60s (the bench is eating your margin) — or above ~90% for months, which means no slack and burnout doing the cutting for you. Full math and bands in the utilization guide.

2. Bench cost

The bench in money, not heads: each under-allocated billable person's idle share times their monthly cost. Two people at 50% never show up in a headcount view — they show up here. Bad: a rising trend while everyone "feels busy".

3. Margin per project

Revenue minus the pro-rated cost of the people delivering, per engagement. Company-level profit can hide a flagship client that's underwater, subsidized by two small ones. Bad: finding that out at annual accounting instead of next month.

4. Revenue per employee

Total revenue over total headcount — blunt, and useful precisely because it's blunt. Bad: headcount growing while the ratio falls — you're hiring ahead of (or instead of) revenue.

5. eNPS trend

Not the absolute score — at small headcounts one person swings it by several points — but the direction across surveys, and the response rate. Bad: the response rate falling before the score does; people who stop answering have stopped believing anything changes. Running it without destroying trust is its own guide.

6. Attrition and tenure

Twelve-month turnover (leavers over average headcount), split by reason, next to who's joining. Bad: regretted exits clustering in one team, under one manager, or in one quarter — a pattern individual exit conversations never quite reveal.

7. Time-to-hire

Days from application to acceptance, and where in the pipeline they accumulate. Speed is a competitive weapon for talent. Bad: candidates consistently lost at the offer stage — the pipeline is fine, your decision-making is slow.

8. PTO liability

Untaken leave accruing on your books — a cash exposure (unused days are typically paid out on exit) and a burnout signal in an accounting costume. Bad: your most senior people sitting on 25+ unused days — exactly who you can least afford to have crash or leave expensively.

Leading vs lagging — know which is which

Half the value of a metrics list is knowing which numbers predict and which confirm.

Leading (they change your next 4–8 weeks): utilization, bench cost, time-to-hire, eNPS response rate. When utilization dips, next month's margin is already decided — unless you act this week: resell the capacity, pull sales forward, pause a hire.

Lagging (they tell you what already happened): margin per project, revenue per employee, attrition. By the time attrition rises, the causes are months old. These are steering checks, not steering wheels.

The operating rule: manage by the leading metrics weekly-to-monthly; review the lagging ones monthly-to-quarterly to verify the steering worked. A company watching only lagging metrics is driving by the rear-view mirror.

The monthly 30-minute metrics ritual

The eight numbers earn their keep in a short, fixed ritual — not on a dashboard someone glances at:

1. Same day every month, 30 minutes, founder plus whoever owns delivery and HR. No deck — the eight numbers on one page.
2. Start with last month's decision. What did we say we'd do, did we, did the number move? This step is the whole difference between a ritual and a slideshow.
3. Read deltas, not absolutes. "Utilization 74%" means little; "down 6 points since the roll-off" is a decision waiting.
4. One named action per red metric. Not five initiatives — one owner, one action, reviewed next month.
5. Swap, don't add. If a ninth metric wants in, something leaves.

And if assembling the page takes half a day of spreadsheet archaeology, that — not the meeting — is the problem to fix first.

How Helia HR does this

Most of the eight are live views in Helia, not exports you assemble:

• Utilization and bench — on the dashboard, computed honestly: average billable allocation across available billable people, net of PTO; bench = billable people under 100%. Trend chips compare today against a stored snapshot from about a week or month ago — and with no history yet there's no arrow, never a fabricated delta.
• Margin and bench cost — on the profitability page (admin-gated): the billable revenue run-rate from live assignments and rates, monthly workforce cost (actual compensation where recorded, a blended average otherwise), margin %, and bench cost as the idle share of each billable person's cost — a 75%-booked person contributes 25% of theirs. Multi-currency revenue is FX-consolidated, with honest exclusions when a rate is missing.
• Margin per project, per client, and per person — three lenses on the same live data, so you see where money is made, not just that it is.
• Revenue per employee — the components sit side by side on the profitability page (revenue run-rate and headcount); the division is yours.
• Attrition — the turnover report: joiners and leavers by month over the last 12 months, the turnover percentage, and exit reasons fed by offboarding.
• eNPS trend — the Engagement pack: anonymous pulses with the score, trend across cycles, response-rate tracking, and a minimum-response floor before anyone sees results.
• Time-to-hire — recruiting analytics in the Recruitment pack: average time-to-hire, funnel conversion per stage, and time-in-stage so you see where candidates stall.
• PTO liability — the inputs are tracked: per-person balances and carryover, exportable with payroll inputs. The money valuation happens in payroll with your accountant; Helia keeps the day counts honest.

Money views are role-gated like everything else — margin data is for admins, not the whole company.

FAQ

How often should we look at these?

The monthly ritual is the backbone. Utilization and bench deserve a weekly glance in volatile periods — a project ending, a ramp-up — because they're the ones you can still act on before the month closes.

Should we set individual utilization targets per engineer?

Carefully, if at all. Individual utilization is mostly a staffing outcome, not a personal choice — punishing an engineer for a bench their manager created teaches people to inflate timesheets, which destroys the metric you manage by. Manage utilization at company and project level; manage individuals through 1-on-1s.

We don't track costs yet — where do we start?

Start with the metrics that need no cost data: utilization, bench headcount, time-to-hire, eNPS. Then add a single blended monthly cost per employee — one number, not a payroll integration — and margin and bench cost unlock immediately.

Is eight too many for a 10-person company?

Probably. Start with three: utilization, bench, and eNPS once there's enough trust to survey. Add margin per project when you run more than a couple of concurrent engagements, and time-to-hire when hiring becomes regular.