An agency’s biggest cost is its people. Salaries, freelancer fees, and the overhead that comes with keeping a team productive typically account for 55-70% of revenue. Get resource planning wrong and you either burn out your team or bleed money on underutilised capacity.
Most agencies manage resourcing with spreadsheets, gut feel, and Monday morning standups. It works until it does not. The project that quietly consumes 40% more hours than scoped. The senior designer who is at 120% capacity while a mid-weight sits at 60%. The freelancer brought on three weeks too late because nobody saw the crunch coming.
AI does not replace the judgement calls. But it replaces the guesswork that leads to bad ones.
Forecasting project demand
Historical data is the foundation. Every completed project in your agency contains resourcing lessons: how many hours it actually took versus the estimate, which roles were needed at which stages, where scope crept, and when the team was stretched.
AI analyses this data and produces demand forecasts that are significantly more accurate than human estimates alone. Feed it your pipeline (confirmed projects, probable wins, and proposals out) and it models the resource requirements across the next 4-12 weeks.
What this looks like in practice:
- Week-by-week capacity forecast showing projected utilisation per team member
- Skill demand curves highlighting when you will need more design capacity versus development versus strategy
- Confidence intervals based on pipeline probability, so you can plan for best and worst case scenarios
Agencies using AI-powered forecasting report 20-30% improvement in estimate accuracy. That translates directly to fewer scope overruns and better margin protection.
Identifying overallocation before burnout
Burnout does not announce itself. It accumulates silently until someone resigns or makes a costly mistake. Traditional resource planning catches overallocation after the fact, when the timesheet shows 55 hours logged for the third week running.
AI catches it in advance by monitoring:
- Current allocation versus capacity. Not just this week, but projected forward based on project timelines and upcoming deadlines.
- Allocation trends. A team member whose utilisation has crept from 85% to 95% to 105% over three months is heading for trouble.
- Task concentration. Someone allocated to five projects at 20% each is context-switching constantly, which is more draining than one project at 100%. AI flags this pattern.
- Leave and availability. Cross-referencing holiday bookings, sick days, and training commitments with project schedules to identify clashes before they become crises.
The output is an alert system. Not a daily barrage of notifications, but a weekly summary: these three people are at risk of overallocation in the next two weeks. Here are the projects causing it. Here are the options to redistribute.
The impact is real. One agency we work with reduced unplanned staff absences by 35% in the first six months after implementing AI-driven capacity monitoring. The connection between overwork and sick days is direct, and catching it early makes a measurable difference.
Matching skills to projects
Not all hours are equal. A senior strategist and a junior account executive both have available capacity, but they are not interchangeable. AI maps skills and experience to project requirements, producing better resource matches.
This works in two ways:
Project staffing recommendations. When a new project comes in, AI analyses the brief against your team’s skills, experience with similar projects, and current availability. It recommends the optimal team composition.
Development opportunity matching. AI identifies projects where a junior team member could be paired with a senior one for skills development, without compromising delivery quality. This is resource planning that doubles as professional development.
The benefit compounds over time. As the AI learns which team compositions produce the best results (measured by client satisfaction, margin, and on-time delivery), its recommendations improve.
Predicting when to hire versus freelance
The hire-or-freelance decision is one of the most consequential an agency owner makes. Hire too early and you carry cost before revenue justifies it. Hire too late and your team burns out while quality drops. Use freelancers too much and you lose institutional knowledge and margin. When it is time to hire, AI can accelerate the recruitment process significantly.
AI helps by modelling scenarios:
- Sustained demand analysis. Is the increased workload a spike or a trend? AI analyses your pipeline depth, repeat client patterns, and seasonal trends to distinguish temporary surges from genuine growth.
- Cost comparison modelling. What does the next 12 months look like if you hire a full-time designer at £38,000 versus using freelancers at £350/day? Factor in utilisation projections, bench time, and the overhead of management.
- Break-even calculations. At what utilisation rate does the hire pay for itself? AI models this against your projected pipeline and shows you the risk.
The output is not a binary “hire” or “don’t hire” answer. It is a data-informed view that sharpens the decision. You still apply the judgement (culture fit, growth ambition, risk appetite), but you do it with better information.
Scenario planning
Resource planning is full of “what ifs.” What if we win that pitch? What if the retainer client doubles their spend? What if two team members go on parental leave in the same quarter?
AI handles scenario modelling quickly. Define the variables, and it shows you the resource implications of each scenario and combination of scenarios. This is particularly valuable for:
- Pipeline planning. Model the resource impact of winning 2, 3, or 4 of your current proposals
- Growth planning. What does your team need to look like at £1.5M revenue versus £2M?
- Risk planning. What happens if your largest client leaves? How quickly can you redeploy the team?
Running these scenarios manually takes hours. AI produces them in minutes, which means you actually do the planning instead of putting it off.
The limitations
AI resource planning is powerful but not omniscient. Be aware of:
Context it cannot see. AI does not know that Sarah works best in the mornings, that James and the client’s marketing director have a difficult relationship, or that the new project requires on-site presence three days a week. Human context matters, and the resource planner should account for it.
Data quality dependency. If your team does not track time accurately, or your project scoping data is inconsistent, the AI’s forecasts will be unreliable. Garbage in, garbage out. Implementing AI resource planning often requires improving your data practices first, starting with better time tracking.
Over-optimisation risk. AI will naturally push towards maximum utilisation, because that is mathematically optimal. But agencies are not factories. Creative people need breathing room. Sustained 90%+ utilisation kills innovation, morale, and eventually, quality. Set a utilisation ceiling (75-80% is healthy for creative roles) and let the AI optimise within that constraint.
Getting started
You do not need a dedicated platform to begin. Start with what you have:
- Export your project data. Hours logged, estimates versus actuals, team allocations. Even six months of data is enough to start.
- Build a simple forecasting model. Use AI to analyse your historical projects and produce average resourcing patterns by project type. This alone improves your future estimates.
- Add pipeline modelling. Map your current pipeline against the forecasting model to see where capacity gaps will appear.
- Layer in alerts. Set thresholds for utilisation and have AI flag when team members are approaching them.
For agencies ready for a more robust solution, tools like Runn, Forecast, and Float are integrating AI features that handle much of this natively. The investment is modest relative to the cost of getting resourcing wrong.
Your people are your product. Planning their time well is not just an operational task. It is one of the most direct ways to protect your agency profit margins, your culture, and your ability to do great work.
This is part of Margin Watch, a series on how AI is reshaping the business of running an agency. Subscribe to the newsletter to get new articles weekly.
