The billable-hours problem is usually hiding in plain sight

Most agency owners don’t need to be told that admin is expensive. They see it every Friday afternoon when timesheets are incomplete, account managers are building client update decks, creative leads are chasing approvals, and project managers are trying to work out who can take the next urgent brief.

The harder part is seeing the full cost.

A team member might spend 20 minutes collecting campaign data, another 30 minutes formatting it into a report, and an account manager another hour translating the result into a useful client email. None of those tasks looks alarming in isolation. Across 20 accounts, every week, it becomes a serious block of non-billable time.

For marketing and creative agencies in the USD 1M to USD 25M range, we commonly see annual operational leakage in the $60K to $180K band. That doesn’t mean every dollar of admin time can become billable tomorrow. It means the agency has capacity tied up in repeatable work that should not require skilled people to start from scratch.

The pattern usually looks like this:

• Account managers spend 30% to 50% of their week reporting, updating clients, chasing information, and preparing for internal meetings.
• Creative teams lose time interpreting briefs, looking for past work, and waiting for a decision that should have been routed earlier.
• Project managers maintain schedules manually because delivery data sits across email, Slack, spreadsheets, and project tools.
• Leadership hires another account manager or coordinator because the current team has reached an account load of six to 10 clients.

That last point matters. If headcount is the only lever for growth, margin gets squeezed every time you win a new account.

The better question is not, “How do we make people work faster?” It’s, “Which work should stop landing on their desks in the first place?”

See Omni for marketing and creative agencies to understand how we map those workflows before recommending tools.

Where agencies lose 10 to 15 hours per person each week

The 10 to 15 hour opportunity is rarely one task. It is the total of small operational loops that pull people out of revenue work. AI automation works best when it removes the handoffs, copying, follow-up, and first-draft effort around those loops.

Here are four places to start.

1. Timesheets become a Friday chase

Timesheets are essential for agency margin management. Yet the process is usually reactive.

People work across client calls, Slack threads, Figma files, campaign platforms, project boards, and email. Then, at the end of the week, they are asked to reconstruct where their time went. Project managers chase missing entries. Finance closes the period late. Account leaders discover after the fact that a retainer has been over-serviced.

An AI-supported timesheet workflow can collect work signals from connected systems, group them by client and project, and propose time entries for the team member to review. It should not silently submit time on someone’s behalf. The person needs to confirm the entry and correct the context.

The practical result is different from a basic timer. Rather than asking a strategist to remember six client tasks from Tuesday, the system presents a draft such as:

• 45 minutes reviewing paid social performance for Client A
• 30 minutes preparing a creative feedback summary for Client B
• 60 minutes in the weekly delivery meeting for Client C
• 20 minutes updating the campaign brief

The strategist reviews it in a few minutes. The project manager sees exceptions early instead of chasing 12 people on Friday. The owner gets cleaner utilisation and profitability information.

That is operational control, not just a better timesheet reminder.

2. Status updates are rebuilt from fragments

Client communication often creates a strange form of double handling. The delivery team updates tasks in the project platform. The account manager checks Slack for blockers. Someone exports performance data. Then the account manager turns all of it into an email, a deck, or a meeting agenda.

The Reporting Agent in Omni ops handles this kind of work. It pulls performance data from connected platforms, drafts the monthly report, and drafts the account manager’s email summary ready for review and sending.

The key word is drafts. A useful Reporting Agent does not send generic commentary like, “Engagement increased this month.” It follows the agency’s reporting structure, applies client context, identifies meaningful movement, and shows the account manager where it found each number.

For a paid media account, it might compile spend, leads, cost per lead, conversion rate, and pacing against the monthly target. For a content client, it could assemble production volume, top-performing posts, draft approvals, and upcoming publishing risks. For a brand project, it can summarise the work completed, decisions required, and what the client will see next week.

The account manager still owns the relationship. They add judgment, tone, and the conversation that data alone can’t provide. They no longer need to spend two hours building the first version of the update.

You can see how this sort of workflow fits within Omni ops, where the focus is on connecting recurring business processes rather than adding another isolated AI tool.

3. Internal approvals delay delivery and consume senior time

Creative agencies often describe approvals as a client problem. In many cases, the internal approval process is just as costly.

A copywriter finishes a landing page draft. A designer needs sign-off on a concept direction. An account manager needs confirmation that extra work is in scope before responding to a client. The request gets posted in Slack, then disappears under other messages. Someone follows up. A creative director reviews it late at night. The delivery date moves, or the team works around the delay.

AI won’t replace the judgment of a creative director or agency partner. It can remove the administration that gets in the way of that judgment.

A properly designed approval workflow does five things:

1. It identifies that a decision is required from the project stage, task status, or submission form.
2. It assembles the relevant context, including the brief, asset link, client requirements, budget status, and deadline.
3. It routes the request to the right approver based on work type and spend authority.
4. It sends reminders only when the decision remains outstanding.
5. It records the decision back in the project system and alerts the people who need to act.

This is particularly useful for scope control. If a client asks for a third round of revisions outside the agreed allowance, the account manager should not need to search email and project notes before replying. The workflow can surface the relevant contract language, previous feedback rounds, and estimated impact. It then drafts a clear response for the account manager to approve.

That protects margin without turning every client conversation into a dispute.

4. Resource scheduling depends on memory

Resource scheduling looks simple when the agency has eight people and a short project list. It gets difficult when a team has multiple disciplines, part-time contractors, changing priorities, and clients who expect quick turnarounds.

A project manager often becomes the central memory system. They know that the senior designer is on a brand sprint until Thursday, the paid media specialist has a campaign launch coming, and the video editor is waiting on client footage. If that person is away, the schedule loses reliability.

An AI scheduling workflow can combine open tasks, estimated effort, role requirements, due dates, leave, confirmed client priorities, and current capacity. It can then flag problems before they become emergencies.

For example, it may show that a client launch is at risk because 26 hours of design work is due next week but only 14 hours are available from the right people. It can suggest options: reassign work, move a lower-priority task, use an approved contractor, or ask the account manager to reset the date.

That does not mean allowing an algorithm to allocate people without oversight. Agency work needs human judgment. It does mean the project manager starts with an informed recommendation instead of rebuilding the schedule from a spreadsheet every morning.

For related thinking on where AI belongs in business systems, our insights library is a useful place to continue the research.

What an AI agent workflow looks like end to end

The agency owner should be able to trace the workflow from trigger to outcome. If the process is vague, it will create more review work than it saves.

Take a standard monthly client reporting cycle.

The trigger might be three business days before the client’s reporting deadline. The Reporting Agent pulls data from connected ad platforms, analytics tools, CRM records, the project management system, and previous report templates. It checks for missing data, unusual movements, and incomplete delivery items.

It then prepares four outputs:

• A client-ready report using the agency’s approved format
• A concise account manager email with key outcomes and proposed next steps
• An internal account health summary
• A list of questions where human input is needed

The account manager receives the draft with source links and a short explanation of what changed. They edit the narrative, add commercial context, and approve it. If results indicate risk, the Account Health Agent can flag the account before the reporting deadline and draft a next-step message.

The Account Health Agent watches connected client accounts daily. It can identify warning signs such as a sudden drop in lead volume, an overdue approval, a project consuming more hours than planned, or a long period without a meaningful client interaction. It then brings the issue to the account manager with context, not just another notification.

A useful alert might say:

Client B has used 82% of its monthly delivery hours with 11 days remaining. Two revision requests are pending. Draft response prepared with scope options.

That is a better use of an account manager’s time than asking them to manually inspect every account’s status.

The same pattern applies to production. The Content Production Agent takes an approved brief, brand guidelines, source material, client preferences, and output format. It produces a first-pass draft that the team edits. For recurring content programs, this reduces the blank-page problem and shortens the path from brief to review.

The agent should not publish unreviewed client content or make strategic claims without controls. It should be configured around the agency’s standards, approval stages, and brand rules. Human review remains part of the delivery model.

If you want to map these workflows against your own systems, Book a 60-min Omni Audit. It is a working session, not a software demonstration.

Turn recovered capacity into a margin decision

Reclaiming admin time only creates value if you decide where that capacity goes.

For some agencies, the right choice is more client delivery without immediate hiring. For others, it is reducing overtime, improving strategic planning, or giving senior people time to retain and expand accounts. The answer depends on your current utilisation, sales pipeline, and delivery quality.

Start with a simple capacity model.

Say you have 12 client-facing people. If your process review finds an average of eight recoverable hours per person each week, that is 96 hours of potential capacity. You will not recover every hour in month one. Systems need tuning, and people need to trust the workflow. But even recovering a portion can change the need for the next hire.

Translate the time into the measure that matters in your agency:

• Gross margin protected on fixed-fee retainers
• Additional project capacity at your normal blended rate
• Fewer contractor hours used to cover avoidable internal work
• Better client retention because account managers have time for proactive conversations
• Faster billing and cleaner visibility into account profitability

Don’t treat the model as a promise from a vendor. Treat it as an operating hypothesis to test. Baseline the current process for four weeks, measure the time involved, automate one repeatable workflow, and compare the result.

This is often where owners find that their systems are not the whole issue. The agency may have unclear service boundaries, inconsistent templates, or too many exceptions. AI exposes that operational debt quickly. That is useful, because automating a messy process only makes the mess travel faster.

You can also review Omni advisory if the operational question is bigger than a single workflow and includes decisions about priorities, team structure, and implementation sequencing.

A sensible first 90 days

Don’t try to automate every department at once. Pick one high-frequency workflow with a clear owner, available data, and an obvious review step.

For most agencies, monthly reporting or timesheet completion is a good starting point. Both have repeatable inputs, visible cost, and an easy way to measure improvement.

In the first 30 days, document the current workflow. Track each handoff, tool, approval, and exception. Identify which data sources are reliable and which need cleanup. Agree on what the agent can draft, what it can update, and what always requires human approval.

In days 31 to 60, build the first workflow around a defined group of accounts or one delivery team. Keep it narrow. Compare draft quality, turnaround time, time spent by role, and error rates. Ask the people doing the work where the agent creates friction.

In days 61 to 90, refine the prompts, decision rules, templates, and routing. Then consider the next linked workflow. A reporting process may naturally lead to account health monitoring. A better timesheet process may reveal where resource scheduling needs attention.

The aim is not to make your agency feel more automated. The aim is to create more time for the work clients pay you to do.

There are more practical examples and operating ideas in our guides for agency leaders. Use them to pressure-test where your own team spends its week.

Find the work that should leave your team’s desk

The real opportunity is not replacing talented agency people. It is removing the administrative load that prevents them from doing the work only they can do.

Your account managers should be building client confidence, spotting growth opportunities, and managing commercial conversations. Your creatives should be improving ideas and output quality. Your project team should be protecting delivery, not acting as a manual integration layer between disconnected systems.

An Omni Audit gives you three practical outputs in 60 minutes: the workflows creating the biggest leakage, the AI agent opportunities worth prioritising, and a clear next-step plan. There is no slide deck designed to impress you. You leave with a view of what to fix first.

See the AI audit for marketing and creative agencies for the agency-specific approach, or Book my Omni Audit when you’re ready to put real numbers against the time your team is losing.