How to Manage Campaigns, Clients, and Billing for Ad Agencies

The thing nobody tells you when you start an agency is that the work is not the problem. You can ship beautiful creative. You can run a paid media account that beats benchmark. You can write copy that the client's CEO forwards to their board. None of that saves you from the operational mess of running a 12-person shop where four projects are over budget, two retainers are silently leaking hours, and the bookkeeper just sent a Slack asking why three invoices from October are still unpaid.

Agency project management is not Asana with a brand kit attached. It is the discipline of tracking four things at once — hours by person, by client, by project, billable vs. non — while running a portfolio of revenue models (retainer, project, media commission, hourly with a cap) where the same client might be on three of them simultaneously. Most PSA tools handle two of those four axes well and treat the rest as someone else's problem.

This post is a working operator's guide. The math, the workflow, the stack, and the specific places where small agencies leak margin. If you run a 8-25 person agency and you are bleeding profit you can't explain, the answer is almost certainly in the operating system, not the work.

The shape of the business has shifted in the last three years. The classic full-service agency model is rarer at the SMB end. What we see most often:

A shop of 8-25 people. Some mix of strategy, creative, paid media, content, SEO, sometimes a PR or social specialty. A book of 10-25 active clients. Average revenue per client somewhere between $5k and $25k per month for retainers, with project work layered on top. Net margin in the 15-25% range when the agency is healthy, 5-10% when scope creep and hour overruns are unaddressed, and negative when a single bad client is allowed to drag the team for a quarter.

The pressure has come from three directions. In-housing — the brand has hired a head of growth and a junior media buyer, and now wants you to be the strategy layer only. AI-driven creative production — the cost floor on a competent landing page has dropped, and clients know it. Procurement-led RFPs — even at SMB size, clients are getting more sophisticated about how they buy agency services. Margin protection used to be a finance concern. In 2026 it is the operational concern.

Most operators we talk to think of their agency as primarily one revenue model. "We're a retainer shop." "We do project work." In practice, when you look at the book, almost every agency above 5 people is running all four at once across different clients — and often on the same client.

The four models, and what each one demands operationally:

Notice that every model demands a different operational discipline. A retainer shop that runs project work without change-order discipline will eat the project margin. A project shop that picks up a retainer client and tries to manage it like a project will under-deliver on the relationship. The agencies we see make the most money are not the ones who pick a model — they are the ones who explicitly track which model each engagement is on, and apply the right operating rhythm to each.

Every agency needs to answer four questions at the end of every week:

1. Who worked on what? Hours by person. 2. What client did the work serve? Hours by client. 3. What project or deliverable did the work feed? Hours by project. 4. Was the time billable? And if billable, against which revenue model?

Those four axes have to roll up cleanly. A senior designer's Tuesday afternoon needs to attribute to her name, to Client X, to the Q3 brand refresh project, and to billable retainer hours (not project hours, not internal admin).

Most project management tools handle the project axis fine. Most time-tracking tools handle the person axis fine. Almost none of them handle all four with the rollups an agency principal actually needs at the end of the month — which is why so many agencies end up with a spreadsheet pulling exports from three systems and a part-time ops person whose job is to reconcile them. That spreadsheet is where margin goes to die.

There is a data model that works for agencies and one that almost works and breaks at scale. The working version:

Account is the client company. "Patagonia." One record, owned by an account director, that contains the business relationship — primary contacts, billing details, contract on file, the high-level strategy notes.

Engagement is the commercial agreement. Patagonia's monthly retainer is one engagement. The brand refresh project that started in Q2 is a second engagement under the same account. The TV spot they hired you for in November is a third. Each engagement has its own revenue model, scope document, start/end date, and budget.

Projects are the deliverables under the engagement. Under the monthly retainer engagement you might have a recurring "April execution" project, an "April reporting" project, and a one-off "summer campaign brief" project. Each has tasks, assignees, dates, and tracked hours.

Time gets logged at the project level. It rolls up to engagement profitability. Engagements roll up to account profitability. Account profitability rolls up to the agency P&L. When this structure is in place, the answer to "how much did we actually make on Patagonia this quarter?" is one query, not a three-hour bookkeeping exercise.

Every agency principal we've talked to can name a moment when they realized the books didn't make sense — the year the agency grew 30% in revenue and somehow lost money. The forensic answer is almost always in one of five places.

The retainer scope says four blog posts a month and two paid social variants. The client asks if you can also redesign the landing page "since we're already in there." The account manager says yes because the relationship matters and the lift seems small. Six weeks later, the design team has burned 40 hours on a landing page that was never in scope, and the retainer P&L is upside down for the quarter.

The fix is not less generosity. The fix is process: every out-of-scope request triggers a change order. The change order is a one-page document that names the new deliverable, the estimated hours, and either an additional fee or an explicit reduction in other scope to absorb it. Some are unbilled (you decide it's worth eating to keep the relationship warm). Some are billed (a one-time addition to the next invoice). The point is that the decision is conscious, documented, and tracked against the engagement.

Agencies that run a change-order process consistently report 8-15% margin improvement on retainer business within two quarters. The math is not exotic — they are just stopping the leak.

You quoted the rebrand at 180 hours. The team has logged 240 hours and you are 70% through the work. Nobody flagged it at 100 hours. Nobody flagged it at 150. By the time it surfaces, the project is already underwater and the client expects what they paid for.

The operational fix is a weekly variance report: estimated hours versus actual hours, by project, with a flag at 75% and 100% of the estimate. A senior on the team owns the conversation when the flag fires — usually that means a check-in with the client, sometimes a change order, occasionally an honest write-off. The signal you want to avoid is finding out about the overrun in the next month's P&L.

A retainer is a commitment. The client paid for X, and at the end of the month you should be able to demonstrate that you delivered X. Most SMB agencies skip this step. The client is happy enough, the work shipped, the invoice cleared. Then the renewal conversation comes up and the client says, "What exactly are we paying for again?" and nobody on the agency side has the answer ready.

The fix is a monthly retainer reconciliation document. Two pages, sent to the client by the 5th of the following month. What we committed to. What we delivered. Variances either way. Time spent (in hour bands, not exact minutes — clients don't need the surveillance, they need the accounting). This document costs an hour to produce and saves the renewal conversation. It is also a gift to the agency team — when the client downscopes, you have a paper trail. When they upscope, you have evidence of the additional ask.

The math is brutal and most agency founders learn it the hard way. You do the work in March. You invoice on the last day of March. Client terms are net-30, so the invoice is due end of April. The client pays late, so the cash arrives mid-May. You started spending the money on payroll in early March. That is a 75-90 day gap between expense and revenue collection, and it is funded entirely out of the agency's working capital.

There are three operational levers that close the gap. First, bill in advance where possible — retainers are billed for the upcoming month, not the past one. Second, milestone billing on projects — 50% on kickoff, 25% at mid-point, 25% on delivery. Third, automated reminder cadence — client portal access where the invoice is visible, automated emails at day 7, 15, and 30 past due, and a polite-but-firm cadence after that. Agencies that systematize collections, rather than treating it as the principal's awkward Friday afternoon task, cut their average days-sales-outstanding by 20-30%.

Strategy briefs creative. Creative briefs production. Production hands to media. Media hands to reporting. Each handoff is a place where context drops. A campaign that ships late because the production team didn't have the brand assets, that performs badly because the strategy intent didn't reach the media buyer, or that gets reported on with the wrong metrics because reporting wasn't briefed on the success criteria — all of these are handoff failures, not skill failures.

The fix is a campaign brief that travels with the work. Brand guidelines, strategic objective, success metrics, key constraints, contact owner at each stage. Not a Notion doc that goes stale — a structured record attached to the project that every team member opens before they start their step. The agencies who do this well rarely have to redo work.

Below is what a working agency operating system looks like, broken down by function. The point is not the specific tool — it is the function. Some agencies cobble this together from 8-10 separate tools and run a Zapier graph between them. Some consolidate. Both can work. The cost equation is different.

Most SMB agencies have some version of all eight, but in disconnected tools. The classic stack is something like HubSpot for CRM, Asana or Monday for projects, Harvest or Toggl for time, Float for resource planning, a Google Sheets media plan, Dropbox for assets, QuickBooks for invoicing, and a one-off client portal that the developer cousin built in 2023 and nobody maintains. The integration tax on that stack is real, and the cost of "did the time entry in Harvest sync to the project in Asana?" gets paid by the ops lead, every week.

Utilization — the percentage of available work hours that are billable — is the single most predictive number for agency health. Industry benchmarks have been remarkably stable for two decades: target 70-75% utilization for client-facing billable staff (designers, copywriters, media buyers, account leads). Below 60% sustained, the agency is overstaffed for current revenue. Above 85% sustained, the team is burning out and quality is starting to slip, even if nobody is admitting it yet.

A few things to know about the metric. First, leadership and ops roles should not be measured against it — a creative director who is 70% billable is not doing strategic leadership work. Second, the denominator matters. "Billable percentage of 40 hours" gives a different answer than "billable percentage of capacity after vacation, holidays, training, and pitch work." Pick one definition and stick with it.

The practical use of utilization is not as a stick to hit the team with. It is as a leading indicator of the agency's commercial state. Utilization trending down for three weeks running is a sign that either you've over-hired, demand has softened, or projects have stalled on the client side. Each of those calls for a different action.

Below is how the eight operational functions map to the Deelo platform for an SMB agency. The point is not that Deelo is the only way — it is to show what a consolidated stack actually looks like compared to the 8-tool patchwork. Where you actually land is a function of your tooling history, your team's preferences, and your switching cost. The exercise of mapping function-to-tool is the valuable part, regardless of which tools end up filling the boxes.

The two things that change materially when the stack is consolidated are the integration tax and the data layer. The integration tax goes to roughly zero because there is nothing to integrate — the CRM and Projects and Time share the same account record. The data layer means that when a sales rep wins a new account in CRM, the corresponding engagement and project shells exist before the kickoff meeting. The account director does not have to recreate the relationship in three systems.

Six rituals that the best-run small agencies have in common, and that any agency can adopt in a quarter:

There is a transition that happens around the 20-25 person mark, and the agencies who navigate it well do something specific. They stop running the operating rhythm out of the principal's head, and they document it.

Under 15 people, the agency works because the principal knows every client, every project, every team member's current load. The operating system is largely the principal's brain, supported by a shared project tool and a Slack channel. It works, but it is fragile — if the principal is heads-down on new business for two weeks, the operating rhythm stalls.

Above 20 people, the principal can no longer hold it all. The agencies who scale past this point either hire a head of operations whose job is the operating rhythm itself, or they implement a tooling stack rigorous enough that the rhythm is the system, not the person. The ones who try to scale without doing one or the other tend to plateau and start losing senior people to better-run shops.

The practical signal is when you stop being able to answer "what is everyone working on this week" in your head. When that happens, the rhythm has to be in the tool.

Deelo's pitch to agencies is straightforward: one platform, one customer database, one login, for the eight functions above. CRM, Projects, Time, Marketing, Docs, Customer Portal, Invoicing, and the AI Assistant for proposal and report drafting all sit on the same account record. The integration tax that eats a small agency's ops capacity is replaced by a shared data layer.

At $19 per seat per month, a 12-person agency runs the full platform for $228/month. The same workflow stitched together across HubSpot, Asana, Harvest, Float, Dropbox, a custom portal, and QuickBooks typically lands in the $1,800-3,200/month range before the integration platform fees and the ops time to maintain the integrations. The trade-off — and we are clear-eyed about it — is that no single Deelo app has the deepest feature set in its category. Dedicated category leaders will out-spec the individual apps. The operator's question is whether you need feature depth in every category, or whether you need a stack that gets your team out of coordination overhead and back to client work.

For a 12-person agency where the principal is currently spending half their Friday on operational reconciliation, the answer tends to be the second.

The agencies that survive 2026 are not the ones with the best creative reel or the cleverest positioning deck. They are the ones whose operating system protects the work. That means tracking the four axes that matter, structuring clients in a way that lets you see profitability at the engagement level, and running the six monthly rituals that catch leaks before they compound. The work — the creative, the strategy, the media — is the reason you started. The operating system is the reason you get to keep doing it.