Growth creates a deceptively difficult operating problem for agencies. More retained revenue can mean more profit, but it can also mean rushed delivery, senior staff covering specialist gaps, excess contractor spend and clients receiving less than the scope they bought.
An agency capacity planning model turns this problem into a manageable set of decisions. It connects contracted client work, likely new work, team availability, role-level delivery capacity, hiring lead times and gross margin in one view. The objective is not perfect prediction. It is to see pressure early enough to change the plan.
In my view, the useful model is usually simpler than agencies expect. It does not need a complex resource-management platform on day one. A well-structured spreadsheet or business-intelligence dashboard can work, provided the definitions are consistent and someone owns the weekly update.
Start with the decisions the model must support
Do not begin by collecting every operational metric available. Begin with the decisions leaders repeatedly need to make:
- Can we accept this client, and on what start date?
- Which role becomes constrained first over the next 13 weeks?
- Should we hire, use a contractor, delay a start or re-scope delivery?
- Which accounts are consuming more delivery time than their commercial model supports?
- What happens to gross margin if pipeline converts at a low, expected or high rate?
A planning model should answer these questions by role and by week or month. Agency-wide capacity is misleading. You may have spare generalist account-management time while technical SEO, paid-media analysis or senior strategy capacity is already exhausted.
For service businesses selling SEO, paid media, analytics or AI automation, delivery is rarely interchangeable. Treating every person as one generic full-time equivalent hides the exact bottleneck that later causes missed deadlines.
The five connected layers of an agency capacity planning model
Build the model as five linked layers rather than one large worksheet. That separation makes assumptions visible and easier to challenge.
1. Client scope and revenue schedule
Create one record for every active client and signed future engagement. Capture the contract value, start and end dates, service line, billing cadence, renewal probability where relevant, and—most importantly—the planned monthly hours by role.
A £6,000 monthly retainer is not a capacity input until it becomes a delivery plan. For example, it may require 10 strategist hours, 18 technical SEO hours, 12 content hours and six account-management hours each month. The total is useful, but the role split is what enables staffing decisions.
Include distinct rows for projects, migrations, launches and one-off audits. These create concentrated workload that retainers alone do not reveal. If your scope documents are inconsistent, standardise them first. The principles in productising SEO services without sacrificing strategic quality are especially helpful: define repeatable deliverables while preserving room for judgment.
2. Team supply and productive capacity
Start with contracted working hours, then subtract time that is genuinely unavailable for client delivery: annual leave, public holidays, training, internal meetings, management responsibilities, sales support and agreed business-development work.
The result is available hours, not billable capacity. Apply a planned utilisation target to arrive at productive client-delivery capacity:
Planned delivery capacity = available working hours × target utilisation
Target utilisation is a planning assumption, not a universal benchmark. A delivery specialist may sustain a higher target than a practice lead who manages people and develops proposals. Set targets by role, then compare them against your own historic evidence. Sustained actual utilisation above target is a warning sign, not automatically a success.
3. Demand from active work and pipeline
Confirmed demand comes from signed scopes. Pipeline demand requires a separate treatment because a promising opportunity does not justify a permanent hire on its own.
Use three scenarios:
- Committed: signed work, renewals with firm agreement and work already scheduled.
- Expected: committed work plus weighted late-stage opportunities.
- Upside: expected work plus credible early-stage opportunities.
Weight pipeline using your own stage conversion history where possible. For each opportunity, estimate its delivery hours by role, expected start month and probability of reaching that start date. Avoid using revenue probability alone; a £20,000 project can have a very different staffing profile from a £20,000 retainer.
4. Cost and gross margin
Capacity planning without cost becomes a scheduling exercise. Add each employee’s loaded hourly cost, including salary, employer costs and any benefits you consistently include in management accounts. Add contractor rates and directly attributable software or delivery expenses where material.
At account level, use:
Gross margin = (recognised revenue − direct labour cost − direct delivery costs) ÷ recognised revenue
Be explicit about your accounting convention. Some agencies include account management in direct labour; others classify it differently. Either can be useful internally if it is applied consistently. The greater risk is comparing accounts using different logic.
For a stronger account view, pair capacity reporting with a regular SEO client profitability framework. It shows whether a capacity shortfall is temporary demand pressure or a fundamentally underpriced scope.
5. Exceptions and management actions
The final layer is a short action log. A model should not merely show red cells. For each gap, assign an owner, action and deadline: approve a contractor, move an onboarding date, reduce a non-essential deliverable, recruit, or protect time for a key hire interview.
This is where planning becomes an operating rhythm rather than a retrospective report.
Build the core calculations by role and time period
Use weekly periods for the next 8–13 weeks when projects and onboarding are volatile. Monthly periods are usually adequate for the following six to 12 months. Keep one consistent role taxonomy across scopes, time tracking and payroll cost data.
| Metric | Calculation | What it reveals |
|---|---|---|
| Demand hours | Active scope hours + weighted pipeline hours | Required delivery by role and period |
| Productive capacity | Available hours × target utilisation | Safe planned delivery supply |
| Capacity gap | Demand hours − productive capacity | Shortfall or usable headroom |
| Forecast utilisation | Demand hours ÷ available hours | Operational pressure before work begins |
| Forecast gross margin | (Revenue − direct costs) ÷ revenue | Whether planned delivery supports economics |
Do not force all headroom to zero. Some buffer is valuable because client approvals slip, urgent requests arrive and team members take leave. The appropriate buffer depends on the volatility of your work. A high-volume content production team can plan differently from a specialist technical consultancy where one person holds critical expertise.
Use a concrete data-quality audit before trusting the forecast
Pipeline assumptions are only as reliable as the records underneath them. This is where many agencies build an impressive-looking model on incomplete CRM data.
Illustrative anonymised audit example: a 90-day agency new-business data review compared 286 captured enquiry records from form and calendar-booking exports against CRM, sales and invoicing data. Of those records, 254 matched a CRM contact or opportunity (88.8%). A delivery-service assignment was present for 218 matched records (85.8%). Sales marked 146 of those assigned records as accepted for active pursuit (67.0%), and 91 accepted records had a traceable invoice or recognised-revenue match (62.3%). The source figures are a deliberately anonymised operating-audit example, not an industry benchmark.
The important lesson is not the percentages. It is the break in the chain. Without service assignment, expected close timing and reliable revenue matching, the agency cannot translate demand into future specialist hours. Fix the fields before refining probability formulas.
This is closely related to building a clean marketing measurement taxonomy: define the event, owner, source and status once, then use those definitions everywhere.
Make hiring decisions from the duration of the gap
A role gap in one busy week does not necessarily warrant a hire. A persistent shortfall across committed and expected scenarios may. Look at three dimensions together:
- Duration: Is the gap present for one month or several?
- Confidence: Does it exist in committed work, or only the upside scenario?
- Economic case: Does anticipated gross margin support the fully loaded cost and recruitment risk?
Contractors are often sensible for short, specialised or uncertain peaks. They cost more per hour, but that premium can be preferable to carrying fixed cost ahead of proven demand. Employees generally become the stronger option when the role is persistently constrained, strategically important and supported by stable demand.
Model hiring lead time honestly. Add recruitment time, notice periods, onboarding and the period before a new employee reaches expected productivity. Hiring after the model shows a crisis often means the team is already paying for the delay through lower quality, overtime or margin leakage.
Protect the model from the common sources of margin leakage
Time tracking is not perfect, but no time data is worse when assessing whether scopes are workable. Keep categories simple enough that people will use them: client, project or workstream, role, and a small set of activity types. Review exceptions, not every timesheet line.
Pay particular attention to unplanned senior time. It often appears as “quick reviews,” rescue calls or client communication, then becomes invisible because it is spread across the month. If senior involvement is routinely required, it belongs in the scope and the model.
Also separate paid discovery from delivery commitments. For search and AI-search work, an initial technical audit can reveal a much larger effort than sales expected. Reference implementation guidance from sources such as Google Search Central or Bing Webmaster Tools when scoping technical requirements, but do not turn documentation into a fixed-hours promise. Complexity still depends on the client’s stack, approvals and development resource.
Run a weekly operating cadence, not a monthly post-mortem
Update active scope changes, leave, project dates, pipeline stage movement and contractor commitments every week. Hold a short capacity review with delivery, client-service and commercial leads. Focus on movements since the prior review and decisions needed this week.
Then conduct a monthly finance check: compare planned versus actual hours, revenue and direct cost by client and service line. Investigate material variance. The point is not to punish teams for imperfect estimates; it is to improve future scope assumptions and pricing.
Automation can reduce manual work by pulling time, CRM and finance data into a shared view. However, automate collection before automating judgment. The guidance on automating agency reporting without losing judgment applies here: exceptions, decisions and commercial context still need accountable human review.
FAQ and conclusion
What is an agency capacity planning model?
It is an operating model that compares required client-delivery hours with available team hours by role and time period. A useful version also links those hours to client revenue, labour cost, pipeline confidence and gross margin.
How far ahead should an agency forecast capacity?
Use weekly planning for the next 8–13 weeks and monthly planning for the next six to 12 months. Near-term precision matters for staffing and onboarding; longer-range scenarios are more useful than false certainty.
What utilisation target should we use?
Use a role-specific target based on your own delivery model and historic results. Account for management, sales support, training and leave rather than treating every contracted hour as client capacity.
When should we hire instead of using contractors?
Hire when a role-level gap is sustained, commercially supported and strategically important. Use contractors when demand is short-lived, highly specialised or uncertain. Test both choices against forecast gross margin.
Conclusion: A reliable agency capacity planning model makes growth decisions visible before they become delivery problems. Start with clean role-level scopes, realistic availability and three demand scenarios. Review it weekly, connect it to actual client profitability and act on persistent gaps early. The model will not eliminate uncertainty, but it will replace reactive staffing with clearer commercial judgment.