How I Redesign an Agency Service Line That Was Improvised

Most agencies’ service lines are either improvised or designed. Here’s how to spot the difference, and what I do when I find one that was improvised.

An improvised service is when somebody sold it, somebody figured out how to deliver it, and every hire, client ask, and workaround since then got layered on top. Most agencies start here. A client needed something, the agency said yes, and the team worked out delivery while the client waited.

You can see an improvised line in the team’s week

The process lives in the heads of the people who built it. The project management system has a workflow in it, and the team works around it.

Nobody knows what a deliverable costs, because it hasn’t been measured. The team is busy all the time, and the utilization chart says they should have more time. When both of those are true at once, there’s usually a delivery problem sitting between them.

Then there’s the people side. It’s painful for anyone on the team to take time off, and extended PTO or a resignation notice is a nightmare waiting to happen. If you’ve ever replaced one key person on a team and had it turn into almost a full rebuild, you already know what I mean.

The account team feels it too, sometimes before delivery does. Your project manager or coordinator is managing the backside of every dropped ball, and your account manager is defending work that came out late, or came out weaker than it should have. That’s where you start to see rework, negative feedback from the client, questions about alignment, and eventually churn.

Designed is what the work feels like from inside the team

A designed service is when the people on the team have the time and space to do the best job they can.

They know who’s handing off to them and who they’re handing off to. They know where a deliverable stands, what it costs, where to go for the information they need, and how to deliver their piece of it.

The workload is realistic, and when it isn’t, somebody manages the overload. People end work at a reasonable hour and take time off without the pressure. There’s a clear line of feedback, so they can see the impact of their work, and the tools take care of the tedious stuff.

I describe it from the team’s side because that’s where the answer is. An owner looks at a service line and sees revenue and margin. The people doing the work can tell you whether it was designed, and they will if you ask them the right way.

Improvised works until you need to replace someone or sell more

Most agency service lines start improvised, out of immediate need, and that’s fine for a while. The people who built it are still there, and they carry it.

It stops being fine at two moments. The first is when you need to replace someone, because the process walks out the door with them. The second is when you need to sell more of it, because the only way an improvised line takes on more work is to hire more people, and every new person inherits the workarounds along with the work.

The first month is listening, measuring, and doing the work myself

When I redesign a service line, I don’t start with the workflow. I start with the people doing it.

I sit down with them and listen. There’s no judgment and no retribution. I take notes and get on their side, and I do it when they’re not between meetings or up against a deadline, because people only tell you what’s going on when they have room to.

The same week, I put time tracking on the team and gather two weeks to a month of data, even if they already track time. Unless an agency is dialed in on time tracking, I can’t take the existing data at face value, and the baseline has to be something I trust.

Then I log in to the tools and follow the digital footprint. I look at what’s blocked, where the comments pile up, where work flows well and where it doesn’t, and whether the conversation about the work lives in Slack or in the task thread.

Then I go where most agency leaders don’t, which is back into delivery. I onboard onto the team and take on a piece of the work myself for a few weeks. That’s where I reconcile what I heard with what’s happening, and it earns some buy-in, because the team sees me tracking my time and communicating in the places they’re supposed to.

I also read what the team was given, the SOPs, the training, and the templates, so I know whether part of the problem is a training problem. And I score the work blind. I pull a set of deliverables, strip the dates, and have the same reviewers score them the way a client would, which gives me a quality baseline to measure against later.

Then the workflow, and everything around it

With all of that in hand, I redesign the workflow itself, which means the stages, the owners, and a named artifact at every handoff. If there’s no artifact, there’s no handoff. I decide where AI comes in, what it’s allowed to touch, where a human signs off, and what gets retired.

Then I redesign what surrounds it. That means writing down what good looks like, deciding how the service gets scoped, priced, and reported on, working with the account team on how they talk about it and onboard a client into it, and setting targets for cost and output against the baseline from the first month.

I pilot it on one account before the whole team changes how they work. Once it’s proven there, it becomes the proof of concept for everyone else, and that’s when the SOPs, the recorded training, and the AI skills built inside the agency’s own stack come in.

Then I measure again, looking at cost per deliverable, output per head, and whether quality held. If it got faster and worse, it failed.

One agency cut its content cost by 60% on the same quality bar

At one agency, a piece of content went from 11.9 hours to 4.75. Content cost went from $88,000 a month to $35,000. It was the same team, held to the same quality bar.

A lot changed to get there. Strategy moved to the front of the process. The workflow, the systems, and the artifacts got rebuilt, and so did the way the SEO and content teams communicated and worked together. AI was one piece of it.

Bolting AI onto an improvised line makes the work worse

The obvious objection is that this is automation with extra steps. Put AI into a content workflow and of course the hours drop.

They do, and the quality usually drops with them. If you drop AI into a content workflow without redesigning anything around it, quality will go down by design. Most of the agencies I work with can’t afford that trade. The people in them work as craftsmen, they take the creative and strategic side of the job seriously, and using AI only to speed things up is not going to fly with them.

So the design has to answer the people questions too. The team needs to know how to use it, why we’re using it, and where we’re not using it. The version that works is the one where they see AI as a tool that helps their work make a bigger impact, and not a tool to get more work done faster.

That makes the whole thing a change management project, and it’s hard. Creatives have a hard time giving up control over the parts of the work where taste lives, and I understand why. Convincing them that this isn’t here to replace the heart and the brain and the spirit of the work takes more than one conversation. I overprepare, I overcommunicate, and I go in planning for the most resistance and the least belief. Unless you’ve gotten it wrong a few times, it’s hard to get it right the first time.

Somebody on the leadership team has to own it

You have to invest the time. Either you or someone on your leadership team has to own this with a big chunk of time and attention, for as long as it takes to get right. It doesn’t work as a side project squeezed in between client calls.

And if you’re not sure whether your services are improvised or designed, ask the people doing the work. I’d bet they already know the answer.