Agency capacity planning for teams that hate timesheets

TL;DR

Agency capacity planning fails for the same reason timesheets fail: both demand task-level precision from people whose week is interruptions. The workable version needs three rough numbers instead. Supply — each person's honest deliverable hours per week (25–30 out of a 40-hour week, not 40). Demand — each client's weight class (S/M/L hours-per-week, calibrated once a quarter against a single sampling week, not tracked daily). The board — committed hours over deliverable hours, per person per department, reviewed weekly in ten minutes. Healthy is 70–85% committed; the unplanned 15–30% is where reactive client work actually lives, so a board showing 100% isn't full, it's already failing. Sustained weeks above 85% with pipeline behind it is the hiring trigger; spikes get the freelance valve instead. None of it needs a timesheet.

---

Most agencies discover their capacity the way you discover a cliff edge — by going over it. A good quarter of sales lands three new clients, delivery quietly saturates, quality slips before anyone has data showing why, and the agency lurches into panic-hiring just as the next quarter softens. The whole cycle is a visibility problem, and the standard fix — everyone tracks their hours now — is the one cure delivery teams reliably reject.

Why timesheets are the wrong tool anyway

Set aside the morale cost and timesheets still don't answer the capacity question, because they face backwards. A timesheet records where last month's hours went; capacity planning asks whether the next six weeks of commitments fit the team you have. By the time retrospective data shows the overload, you've been living it for weeks.

And the precision is fake. Agency timesheet data degrades within a quarter of mandating it — end-of-week reconstruction, round numbers, the same 7.5 hours of "client work" daily. You end up making forecasts from precise-looking fiction. Decision-grade and rough beats audit-grade and false, which is the same argument behind the per-client profitability method — and capacity planning can borrow that method's machinery wholesale.

Step 1 — Supply: deliverable hours, honestly

For each delivery person, one number: hours per week genuinely available for client work. It is never 40. After internal meetings, email and Slack triage, admin, and the tax of context-switching, a full-timer reliably delivers 25–30 client-facing hours; team leads who run standups and reviews are closer to 18–22; a selling founder might honestly be a 10.

Write the number down per person, revisit twice a year, resist the urge to flatter it. An inflated supply number doesn't make the team bigger — it just moves the overload somewhere the board can't see.

Step 2 — Demand: weight classes, not estimates

The instinct is to estimate hours per task and sum them. Don't — task-level estimation is the timesheet problem wearing a different hat. Classify clients into weight classes instead:

• Small — maintenance retainer, one department, light cadence: ~3–5 hours/week
• Medium — active work in one or two departments, monthly reporting: ~8–12 hours/week
• Large — multi-department, weekly client contact, heavy deliverables: ~15–25 hours/week

Calibrate the bands the same way the profitability method does: one honest sampling week per quarter, cross-checked against the per-client activity you already have in your system of record — tasks completed, messages handled, meetings held. Two refinements matter. Weight classes are per department, because a client can be Large for PPC and Small for SEO, and the load lands on different people — the reason department workflows shouldn't share one undifferentiated queue. And onboarding is its own weight: a client's first month runs roughly double their steady-state class, which is exactly the spike that sinks agencies signing three logos in one quarter.

Step 3 — The board: committed over deliverable

Now it's division. Per person, per department: committed hours from the weight classes, over deliverable hours from Step 1, as a percentage. Review it weekly — it's a ten-minute agenda item inside the operating rhythm you already run, not a new meeting.

The reading that matters: 70–85% committed is healthy. The slack isn't laziness — it's where reactive work actually lives. Client fires, platform changes, the "quick call" that eats an afternoon: 15–30% of every delivery week, every week, unplannable in specifics and utterly predictable in aggregate. A board showing 100% planned means reactive work is being paid for with nights, weekends, or quality — usually in that order.

The four over-capacity signals (no timesheet required)

Between weekly reviews, the system of record leaks the truth. If tasks, client communication, and deadlines live in one workspace, these are queries — the same cross-client visibility the cross-client reporting view exists for:

1. Overdue tasks trending up across consecutive weeks — the earliest quantitative signal.
2. Client response latency creeping — same-day replies becoming two-day replies, visible in the message threads.
3. Senior people doing junior work, because delegating now costs more than doing.
4. Internal work perpetually bumped — the SOP updates, the agency's own marketing, the process fixes that compound. First thing sacrificed, least visibly missed, most expensive over a year.

Two or more of these for two consecutive weeks outranks whatever the board says: you're over capacity, and the board's calibration is stale.

Hiring triggers and the freelance valve

The board turns the hardest staffing call into a rule. Sustained above 85% in a department for six-plus weeks, with sold pipeline behind it — open the role now, because a good hire takes six to ten weeks to land and longer to ramp; waiting for 100% means the new person arrives mid-burnout. A spike — one heavy onboarding, a seasonal push — gets the freelance valve, which only works if the bench is built before the spike. Sustained below 60% isn't a staffing decision at all; it's a sales problem, and the answer lives in scaling without losing margin, not in the hiring plan.

The quiet payoff: saying "when," not "whether"

The compounding return shows up in new business. An agency without a capacity board answers every prospect with optimism; an agency with one answers with a date — "we can onboard properly on the first of next month" — which protects the existing book, makes the onboarding spike plannable, and reads as operational maturity to exactly the kind of client worth signing. Prospects who walk over a three-week wait were going to be the clients who treat every request as urgent. The board filters them out before they cost you a coverage ratio.

Capacity planning isn't a spreadsheet hobby. It's the difference between an agency that absorbs growth and one that's permanently surprised by it — and it costs three rough numbers and ten minutes a week.