Utilization should not be a monthly surprise

Most agency owners can tell you their broad utilization target. They might say 75%, 80%, or 85% for delivery staff. The harder question is whether they can see, right now, which projects are consuming unplanned hours and which people are about to run out of billable work.

For a marketing or creative agency doing $1M to $25M in annual revenue, that gap matters. A team can look busy all week while project margins quietly slip. A senior designer spends three hours revising a client deck that was sold with one revision round. An account manager is pulled into internal reporting, status updates, and last-minute client requests. A strategist sits between projects because the next approved brief has not landed.

None of that is visible in a utilization percentage reviewed 15 days after month end.

The best way to track utilization rate across projects is to connect three things in real time:

1. The work people are actually doing.
2. The billable status and budget of that work.
3. The future project pipeline and capacity required to deliver it.

That sounds obvious. In practice, most agencies have time data in one system, project budgets in another, staffing plans in a spreadsheet, and client context scattered across Slack, email, and account manager notes.

An AI operations layer can pull those signals together, calculate billable versus non-billable hours continuously, and alert the right person before a utilization dip becomes a payroll problem.

This is one of the operational areas we assess through the AI audit for marketing and creative agencies. The point is not to build another dashboard. The point is to give owners and delivery leaders a reliable view of where margin is being made or lost.

Start with the utilization rate that helps you make decisions

The basic calculation is straightforward:

Utilization rate = billable hours ÷ available working hours × 100

But agencies often get the denominator and the classification wrong.

Available working hours are not always 40 hours per person, per week. You need to account for public holidays, annual leave, internal meetings, training, business development support, leadership responsibilities, and expected administrative work. If your creative director is realistically available for 22 client-delivery hours each week, planning them as a 40-hour billable resource will create a false capacity picture.

The numerator needs even more care. Not every hour attached to a client is billable. A client call might be included in the retainer. A project manager might be fixing an internal handoff issue. An account manager could spend two hours pulling performance screenshots for a report that should have been automated.

Those hours may still be necessary. They are just not always recoverable against the project budget.

A useful operating model tracks four time categories:

Time category	What it includes	Why it matters
Billable client delivery	Work sold to a client and recoverable under the scope	This drives delivery revenue and shows productive capacity
Client-related non-billable	Rework, over-servicing, unscoped client requests, internal client prep	This is often where account margin disappears
Internal operations	Reporting, internal meetings, training, process work	Some is required, but it needs a planned allowance
Unallocated capacity	Available people without planned project work	This signals a sales gap, a resourcing issue, or poor scheduling

This is more useful than a single agency-wide number. If total utilization is 78%, you still need to know whether it is 78% because the team is delivering profitable work, or because people are spending unpaid time rescuing projects that were priced badly.

For most agencies, the management question is not “Are people busy?” It is “Are the right people spending enough time on work that is inside a profitable scope?”

Why manual tracking breaks as the agency grows

At a small agency, an owner can often see the workload directly. They know which client is demanding extra rounds. They know which account manager is overloaded. They can feel when the next month looks light.

That stops working as the agency reaches 15, 30, or 60 people.

At that point, utilization tracking usually turns into a monthly reporting exercise. Finance exports time data. Operations reconciles project codes. Account leads explain why certain jobs are over budget. Then someone produces a utilization report after the period is already closed.

The report might be accurate. It is still too late to change the result.

The common manual process looks like this:

• Team members log hours once or twice a week, often after the fact.
• Project managers compare submitted hours with project budget spreadsheets.
• Account managers ask specialists for status updates in Slack.
• Operations manually classify time as billable or non-billable.
• Finance calculates utilization by person, department, and month.
• Leadership reviews the result in a meeting.
• The same pattern repeats next month.

There are several failure points in this chain.

First, time entries are vague. “Client work” does not tell you whether the work was included in scope, a revision, or an internal effort caused by a missing brief.

Second, project codes become inconsistent. Someone books time to a client retainer instead of the specific campaign. Someone else logs work against an internal code because the project was not created in the time system yet.

Third, reporting is labor-intensive. In agencies with busy account teams, we commonly see account managers devote 30% to 50% of their week to reporting, decks, performance summaries, follow-ups, and client updates. That is not just an AM productivity problem. It distorts utilization because work that should be system-supported becomes invisible internal overhead.

Fourth, staffing decisions are based on lagging data. You hire because people seem overloaded, then find out that some of the pressure came from unplanned rework on two accounts. Or you delay hiring because the aggregate number looks healthy, while a specialist team is heading into a genuine capacity crunch.

For practical operating frameworks around this issue, our operations resources are a useful place to start. The key is to treat utilization as a live management signal, not an accounting report.

What real-time utilization tracking looks like

AI does not fix a weak project model by itself. You still need clean projects, service categories, budgets, and a clear definition of billable work.

What it can do is remove the manual reconciliation that prevents your team from acting early.

A real-time utilization workflow starts by connecting the systems you already use. That could include your time-tracking tool, project management platform, payroll or HR system, CRM, accounting system, Slack, and calendar. The exact stack varies, which is why an implementation should be designed around the agency’s actual operating rhythm rather than a generic template.

The system then builds a working utilization model.

1. It creates a current capacity baseline

Each person has a weekly available capacity based on their role, contracted hours, leave, internal commitments, and utilization target.

For example, a paid media specialist may have 32 delivery hours available in a 40-hour week after meetings, learning time, and operational work. A department lead may have only 18 to 24 client-delivery hours available because management is part of their role.

The model should not punish people for doing necessary internal work. It should make that work visible and planned.

2. It classifies time against the right project context

Instead of relying only on a staff member to decide whether every entry is billable, an AI agent can use project metadata, task names, the statement of work, retainer terms, and recent activity to suggest or apply a classification.

For example:

• A designer logs time against “Website refresh, revision 4.”
• The project scope allows two revision rounds.
• The agent identifies that the work is client-related but likely outside the planned delivery allowance.
• It marks the time as a margin-risk category and asks the project lead to confirm it.

This is not about silently changing financial records. It is about reducing the number of hours that disappear into broad client codes with no explanation.

3. It calculates utilization at several levels

A useful view updates by:

• Individual
• Role or discipline
• Client account
• Project
• Service line
• Office or delivery team
• Current week, next four weeks, and quarter outlook

That lets you spot different problems quickly. A client project may be over-consuming creative hours while the agency overall still appears on target. A content team may be fully booked for three weeks while paid media has a 15% capacity gap.

Those require different actions.

4. It predicts gaps before they hit payroll

This is where the work becomes more valuable than a normal dashboard.

The AI model looks at contracted retainer capacity, confirmed projects, proposal probability from the CRM, planned start dates, delivery estimates, and team availability. It then forecasts likely utilization by role and week.

If your motion team is projected to fall below its profitability threshold three weeks from now, the system can flag it while there is still time to move work, accelerate a proposal, schedule internal capability work, or reduce contractor commitments.

If your strategy team is projected above capacity for the next month, it can surface the risk before delivery dates slip or your people start absorbing unpaid hours.

Set thresholds that reflect profitability, not wishful thinking

A utilization alert is only useful if the threshold means something.

Many agencies choose a blanket target such as 80%. It is a reasonable starting point, but it may not fit every role. Account management, creative leadership, new business, and delivery specialists have different required levels of non-billable time.

Your threshold should reflect:

• Fully loaded cost by role
• Average effective billable rate
• Expected contribution margin
• Delivery mix across retainers and projects
• The amount of internal work each role is expected to carry
• Contractor versus employee capacity

A simple practical approach is to use three alert bands.

Alert band	Example use	Leadership response
Watch	Utilization is trending below target for two weeks	Check pipeline, work allocation, and pending project starts
Action	A role or team is likely to miss its minimum profitable level	Reallocate work, bring forward sales activity, or adjust contractor plans
Risk	A project is consuming hours beyond budget or a team is over capacity	Escalate with the account lead and protect scope or delivery timing

The numbers should be tailored to your economics. A senior creative lead might have a lower billable target than a production designer. A specialist contractor may need a higher booked level to be viable. The system should support those differences rather than force a single percentage across the business.

For an agency with annual leakage in the $60K to $180K range, the issue is rarely one dramatic mistake. It is usually hundreds of small losses. Two unscoped rounds here. Underused specialists there. Account managers building reports manually every month. A hire made because workload felt high, when the real issue was poor project visibility.

The agent workflow from time entry to leadership alert

An effective utilization agent should work in the background and bring exceptions to people, rather than demanding another daily admin task.

Here is what that can look like across a normal week.

On Monday morning, the agent checks planned capacity against confirmed project work. It sees that the content team has 18 unallocated hours this week, while the design team is forecast at 94% capacity. It sends a concise note to operations and the creative lead with the specific projects and skills involved.

On Tuesday, it detects that a campaign project has used 72% of its design budget with half the planned milestones still open. It reviews task activity and finds three additional revision tasks added after the initial approval. It drafts a summary for the account manager, including the likely remaining hours and a recommendation to discuss scope.

On Wednesday, it sees that two account managers have spent a large portion of their available hours compiling monthly reports. The Reporting Agent can pull performance data from connected platforms, prepare the monthly report, and draft the AM’s email summary for review. That changes reporting from repetitive manual production into an approval step.

On Thursday, it predicts that the paid media team will drop below its target billable capacity in three weeks if three open proposals do not close. It provides a list of accounts with renewal discussions due, open expansion opportunities, and potential internal work that can be scheduled without masking the commercial gap.

On Friday, agency leadership receives a short weekly utilisation brief. It does not need 40 charts. It should answer four questions:

1. Where is billable capacity below target next week and over the next month?
2. Which projects are consuming unplanned non-billable hours?
3. Which teams are heading toward overload?
4. What action has an owner, deadline, and expected commercial outcome?

That is the difference between tracking and managing.

The Account Health Agent supports this process from the client side. It watches accounts for risks and opportunities, then drafts the next-step message before the account manager has to chase the information. If an account is using more delivery time than planned while engagement or results are weakening, that is not just a utilization issue. It is a renewal and margin issue.

The Content Production Agent can also reduce the non-billable drag that often sits behind poor utilization. It produces a first-pass draft from a brief, on-brand and in the required format, so your team edits rather than starting from a blank document. That does not eliminate creative judgment. It reduces the time spent on repeatable setup work.

If you want to map this against your agency’s systems and delivery model, Book a 60-min Omni Audit.

Avoid the mistakes that make utilization data useless

There are a few predictable ways agencies undermine this effort.

Treating time tracking as employee surveillance

People will resist a system that appears designed to police every six-minute increment. Make the purpose clear. You are trying to protect project margins, reduce unnecessary reporting work, plan staffing properly, and stop over-servicing clients without visibility.

Use the data to improve the operating system, not to blame someone for being assigned impossible work.

Measuring hours without project budgets

Hours alone cannot tell you if a project is healthy. Twenty billable hours might be profitable on one engagement and disastrous on another. The agent needs access to budgeted hours, commercial terms, delivery stage, and scope assumptions.

Ignoring non-billable client work

This is one of the most expensive blind spots in creative agencies. If an account manager spends five unpaid hours smoothing over a delivery issue, the client may be happy. The margin still fell. Track the cause, not just the category.

Running alerts without ownership

An alert that goes to a generic inbox changes nothing. Each alert type needs an owner. Project margin risk should go to the project or account lead. Team capacity issues belong with operations and department leadership. Pipeline-driven gaps need commercial ownership.

What to review every week

A weekly utilization meeting does not need to last an hour. For many agencies, 20 to 30 minutes is enough if the data has already been prepared.

Review these areas:

• Current-week billable utilization by role and team
• Forward capacity for the next four to six weeks
• Projects above budgeted hours or with rising revision activity
• Non-billable time by cause, especially reporting and client over-servicing
• Contractor commitments against confirmed demand
• Accounts where delivery effort is increasing without a corresponding scope change
• Actions from the prior week and whether they were completed

The goal is not to explain every variance. The goal is to identify the few decisions that protect the next month’s margin.

If your team is still building this information manually, the first win may be much simpler than a full forecasting model. Automate the monthly reporting workload. Standardize project codes. Connect budget data to time records. Then add capacity forecasts once your core definitions are trusted.

You can find more examples of practical AI operating use cases in our guides library and insights for agency leaders.

Build the system around your actual delivery model

A content-heavy retainer agency needs a different utilization model from a brand studio that works in fixed-scope projects. A performance agency needs to account for recurring campaign management and reporting cycles. A creative production business may have more complex freelancer dependencies.

That is why generic benchmarks only get you so far. The best model starts with your own service lines, pricing, project structures, roles, and systems.

An Omni Audit is a 60-minute working session built around that reality. You leave with three outputs:

1. The manual operational work currently creating leakage.
2. The highest-value agent opportunities across reporting, account health, production, and capacity.
3. A practical first implementation path, without a deck full of generic recommendations.

For marketing and creative agencies, utilization is not a back-office metric. It is one of the clearest signals of whether growth is creating profit or simply creating more work.

See Omni for marketing and creative agencies to understand the approach, or Book my Omni Audit to work through your capacity, margin, and reporting model in detail.