Agency capacity planning comes down to one translation most agencies never make: converting your weighted sales pipeline into projected delivery hours by role, then comparing that number to the hours your team actually has available at 75–85% utilization. Do that comparison every week and you stop overpromising when the pipeline is hot and stop sitting on idle payroll when it’s not. Everything else in this article is the mechanics of making that translation reliable.

Why sales and delivery end up in separate worlds

In most agencies, the pipeline lives in a CRM owned by whoever sells, and the delivery schedule lives in a spreadsheet, a PM tool, or someone’s head. Sales forecasts revenue. Delivery forecasts… usually nothing. They meet for the first time when a deal closes and someone asks who’s free to start Monday.

The answer is nobody, or everybody, and both are expensive. Nobody free means you burn out your best people, slip deadlines in the first month of a new relationship, or scramble for freelancers at premium rates. Everybody free means you’ve been carrying payroll for weeks against work that didn’t land. We see agencies swing between these two states on a quarterly cycle and call it “feast or famine,” as if it were weather rather than a fixable process gap.

The gap is fixable because your CRM already contains a demand forecast. It’s just denominated in the wrong currency — dollars instead of hours.

Step 1: Convert weighted pipeline into hours by role

Start with the weighted pipeline you already have. If a $60,000 project sits at a stage you close 50% of the time, that’s $30,000 of expected revenue. Nothing new there — this is sales pipeline forecasting, and your CRM does it out of the box.

The move that changes everything is converting that expected revenue into expected hours by role. For each service you sell, build a simple delivery recipe: what a typical engagement of that type consumes. A brand-and-website project might be roughly 40 hours of design, 80 hours of development, 25 hours of project management, and 15 hours of strategy. A monthly retainer might be 20 design hours and 10 PM hours, every month, indefinitely.

Now every deal in your pipeline becomes: probability × recipe hours, spread across the likely delivery window. That $60K project at 50% probability, expected to start in six weeks and run for ten, contributes 20 expected design hours and 40 expected dev hours per… you get the idea. You don’t need decimal precision. You need to know whether September looks like 400 development hours of demand against 320 hours of supply.

Two practical notes from doing this with clients:

Use start-date estimates, not close-date estimates. A deal that closes March 30 doesn’t consume delivery hours on March 30. Kickoffs typically lag signatures by two to four weeks, and ignoring that lag makes your near-term demand look scarier than it is.
Keep recipes coarse. Three to six roles, round numbers. If your account manager can’t maintain the recipe in ninety seconds when scoping a deal, it won’t get maintained at all.

Step 2: Plan supply at 75–85% utilization, not 100%

The supply side is where agencies lie to themselves. A full-time employee does not have 40 billable hours a week. After internal meetings, admin, pitching, learning, and the normal friction of working with other humans, a healthy agency utilization rate for delivery staff lands somewhere in the 75–85% range — and that’s utilization of capacity you planned, not of a mythical 40-hour billable week.

So the math looks like: a designer with 40 contracted hours, minus roughly 8 hours of non-billable overhead, gives you about 32 plannable hours. Then plan against 80% of that — around 25–26 hours — and treat the rest as buffer.

Why leave 20% on the table? Because the buffer isn’t slack, it’s shock absorption. Scope creep, sick days, the client who takes eleven days to send feedback and then wants everything in two — all of it lands in that margin. Agencies that schedule people at 95–100% don’t deliver more; they just convert every small surprise into a fire drill, and every fire drill into overtime, and eventually overtime into resignation letters. Chronic over-scheduling is the single most common root cause we find when an agency tells us “our delivery quality is slipping.”

The opposite failure is quieter but just as real: if your planned utilization keeps landing at 55–60%, you’re either underpricing recipes or carrying more team than your pipeline supports. The number tells you which conversation to have. That’s the actual job of resource management for agencies — not policing timesheets, but surfacing the staffing decision early enough that you have options: ramp marketing, pull a deal forward, delay a hire, or line up freelance bench before you need it.

Step 3: Run the weekly capacity ritual

A forecast you build once is a snapshot; a forecast you review weekly is a steering wheel. The ritual we recommend takes 30 minutes with three people in the room: whoever owns sales, whoever owns delivery, and whoever owns money. The agenda never changes:

Committed work vs. capacity, next 8–12 weeks. Signed projects and retainers, laid against available hours by role. Any role over 85%? Under 60%?
Weighted pipeline layered on top. Where does expected demand collide with a role that’s already tight? Development is fully booked through October but the pipeline says two more builds are likely to land — that’s a decision, and it’s due today, not at kickoff.
One staffing decision. Every week, the meeting should end with a concrete call: shift the proposed start date on the Meridian deal, brief the freelance developer, pause the hire, greenlight the hire. If the meeting ends with “let’s keep watching it,” it wasn’t a capacity meeting.

The ritual only works if the data is one system, not three exports. When the pipeline, project plans, and time tracking live in disconnected tools, someone spends half a day each week rebuilding the picture, the picture is stale by Thursday, and the ritual dies within a quarter. This is why we built pipeline and delivery into one connected workspace at OpenAva — the same logic that carries a client from lead to invoice should carry their hours from forecast to schedule.

The mistakes that quietly break agency resource planning

Forecasting in revenue only. $50K of branding work and $50K of development work are identical in the CRM and completely different on the schedule. Hours by role, always.
Sandbagged or inflated stage probabilities. If sales keeps deals at 80% that close 40% of the time, your hours forecast inherits the fiction. Check stage probabilities against actual close rates a couple of times a year.
Ignoring retainers. Retainer hours feel invisible because they’re recurring, but they’re the floor under every capacity calculation. Model them first, then layer projects on top.
Planning individuals instead of roles. At the 8–12 week horizon, plan role-level buckets. Assigning named people to unclosed deals is precision theater.
Treating the buffer as sellable. The moment leadership starts “borrowing” the 15–25% buffer to squeeze in one more project, you’re back to 100% scheduling with extra steps.

Where to start this week

You don’t need new software to run the first pass. Pull your current pipeline, apply rough recipes, and compare the next eight weeks of expected hours to your team’s plannable hours at 80%. The first version will be crude and it will still tell you something you didn’t know — usually that one role is the bottleneck for everything and another is quietly underused. Put the 30-minute ritual on the calendar, run it four weeks in a row, and then decide whether your current toolset can keep the picture live or whether it’s time to connect pipeline and delivery in one place. If it’s the latter, that’s the problem OpenAva was built around, and we’re happy to show you how other agencies run it.