Operations

The Next Hire Is a Workflow. Not a Headcount.

Before you write the job description, count the hours. Most agencies at capacity are not short of people — they are short of the hours they are already paying for.

Four colleagues gathered around a desktop screen in an agency studio, one pointing at the display
The same team, more clients

The Tab You Have Not Closed

There is a half-written job description open somewhere on your machine. It has been there three weeks. You keep not posting it, because the arithmetic does not quite work and because some part of you suspects the problem is not headcount.

That instinct is usually right, and it is worth taking seriously before you commit to a salary finding out. A hire adds capacity and cost on the same day. It also adds onboarding, management, and a fixed obligation that outlives the busy quarter that justified it. Those are real costs, and they are the reason the tab is still open.

The alternative is not “automate everything.” It is to find out where the hours you already pay for are going, and notice how many of them are being spent on things no client has ever been billed for. That number is almost always larger than people expect, and it is recoverable in a way that a hire is not.

Where the Week Goes

Take one real person — an account manager carrying ten clients — and account for forty hours. Not the idealized version. The version with the Tuesday afternoon in it.

An illustrative forty-hour week for an account manager carrying ten clients
Where the hours goHoursBillable?
Client calls and meetings6Usually
Actual delivery work11Yes
Building and checking monthly reports6No
Chasing approvals and sign-off4No
Moving data between tools by hand4No
Answering “where are we on…” messages3.5No
Scheduling, rescheduling, reminding2.5No
Invoicing, chasing payment, admin2No
Internal standups and planning1No

Eleven hours of delivery in a forty-hour week. Twenty-three hours that no client is billed for and that no client would notice if they vanished — provided the outcome still arrived.

Run this for your own team before believing the shape of it. The specific numbers will be wrong for you; the ratio rarely is. The exercise takes one week of honest time-noting and it is the cheapest diagnostic available to an agency owner.

A hire buys you forty new hours at full price. The audit usually finds fifteen you are already paying for and not receiving.
The reframe that changes the hiring decision

Three Kinds of Hours

Not all non-billable time is waste, and treating it that way is how agencies automate the wrong things and damage the client relationship in the process. Sort every row into one of three kinds.

Billable hours

The work the client is buying. Strategy, design, build, campaign management, the judgment calls. Protect this ruthlessly — every hour recovered elsewhere should land here.

Necessary invisible hours

Work the client never sees but genuinely benefits from: quality checks, the thinking before a recommendation, the internal conversation that stops a bad idea. This looks like overhead on a timesheet and is not. Automating it produces cheaper work, not more capacity.

Pure friction

Work that exists only because of how your systems are arranged. Re-typing a lead. Building the same report layout for the eleventh month. Chasing a signature through email. Sending a reminder about a meeting that a calendar already knows about. Nobody chose this work and no client values it.

The Triage

Three buckets, and the discipline is being honest about which is which rather than automating whatever is easiest to automate.

  1. Automate. Repetitive, rule-shaped, and low-judgment. The output is checkable at a glance, and being wrong is cheap and visible.
  2. Delegate. Needs a person but not that person. Often the real answer for work that is judgment-light but relationship-heavy.
  3. Keep. Judgment, taste, relationship, or accountability. This is what you sell.
Work through your hours table, row by row
  • Does this follow the same steps every time? → automate candidate.
  • Does being wrong here cost more than an apology? → not automate; gate it or keep it.
  • Would a client be annoyed to learn a machine did it? → keep, or draft-and-review.
  • Does it exist only because two tools do not talk? → automate, or remove the seam entirely.
  • Does it need context only one person holds? → keep, and write the context down.
  • Is it the same output with different numbers each month? → automate first.
  • Has someone asked to stop doing it twice? → they are right; act on it.

The sixth line is where most agencies should start. Monthly reporting is the single largest block of recoverable friction in almost every hours audit, and it is the one with the clearest automated form: connect the data once, and the report is current whenever the client opens it rather than assembled on the fifth.

What a Hire Still Wins

Automation is not a substitute for people, and an article that pretended otherwise would be selling you something. There are things a hire does that no workflow will, and recognizing them is how you avoid automating your way into a worse agency.

  • Holding a relationship. Clients renew with people. A named human who knows their business is the product, not the overhead.
  • Judgment under ambiguity. When the data is contradictory and the client is anxious, somebody has to decide. That is the job.
  • Taste. Knowing that a design is technically correct and still wrong. No amount of tooling supplies this.
  • Capacity you can redirect. A person can be pointed at a new problem on Monday. A workflow does exactly what it was built to do.
  • Accountability. Someone whose name is on the outcome, who can apologize and mean it. Clients can tell the difference.

If your bottleneck is on this list, hire. Automation will not fix a shortage of judgment, and stretching a team thinner in pursuit of efficiency is how agencies lose the people who had the judgment.

The Capacity Math

Return to the forty-hour week and recover the friction rows — reporting, data re-entry, scheduling, status chasing. Not all of it; assume half, because automation always leaves a remainder.

The same account manager week, before and after recovering half the friction hours
RowBeforeAfterChange
Delivery work11 hrs18 hrs+7
Report building6 hrs2 hrs−4
Manual data entry4 hrs1 hr−3
Status chasing3.5 hrs2 hrs−1.5
Scheduling2.5 hrs1 hr−1.5
Clients comfortably carried1014–16+4–6

Seven hours a week back on one person, and delivery capacity up by roughly sixty percent without a single new contract of employment. That is not a hire replaced — it is a hire deferred, which is the more honest framing. At sixteen clients you will need one anyway, and you will be hiring from a position of profit rather than panic.

One caution, because the arithmetic flatters itself: recovered hours only become capacity if somebody redirects them. Hours freed with no plan attached are absorbed by whatever shouts loudest, usually more status chasing. Decide in advance where the seven hours go.

What to Automate First

Order matters more than ambition. Take the rows in this sequence and each one funds the next.

  1. Reporting. Largest block, clearest automated form, and the client experience improves rather than degrades — a live page beats a monthly attachment.
  2. Lead intake and routing. Every re-typed lead is a chance to lose one. Scoring and ownership within a minute pays for itself on the first deal that does not go cold.
  3. Scheduling and reminders. Small, unglamorous, and entirely mechanical. Reclaiming the no-show is pure margin.
  4. Approvals and status. Give clients somewhere to look instead of somewhere to ask. Most “where are we on…” messages are a visibility problem.
  5. First drafts. Last, and always as a draft a person releases — see what to automate and what to gate for where that line sits.

Items one and four are the same surface in practice: a portal where clients see projects, approvals, files and analytics removes the status question and the report attachment together. Items two and three live in the same record — leads, pipeline and scheduling on one lead history — and item five is where drafted outreach that waits for a human belongs. The full module list is the map if you would rather see how the pieces sit together first.

Then close the loop on cost. This audit priced the hours going out; the stack audit prices the money — and the two usually point at the same handful of seams.

Keep the job description tab open. Just do not post it until you know which of the twenty-three hours you were about to buy back at full price.

ShareXLinkedIn