I use AI almost every working day. And I use it a lot.
Not because I want to say that I “work with AI,” but because it has genuinely changed the way I work.
Tasks that used to take hours can now take a fraction of the time. I can analyze data faster, review campaigns, conduct research, create content, draft emails and reports, review websites, and develop ideas and variations at a speed that simply wasn’t possible before.
We are currently in the middle of the holiday season, and that provides a very simple example.
In the past, when a client needed a holiday greeting, the process would usually involve a designer: preparing a brief, creating the design, making corrections, adjusting the copy, and probably making another small correction before approval.
Today, I can create a holiday banner with AI in minutes, sometimes even seconds, and the quality can be remarkably high.
And I’m not talking about a generic image that looks like everyone else’s holiday greeting. With the right input and direction, we can create visuals that reflect the brand language, colors, audience and message.
That is a huge saving in time. But it is also a fundamental change in the way the work is done.
And creative work is only one example.
Campaign management is another.
Today, it is possible to connect an AI system such as Claude to an advertising account through a connector and give it access to campaign data. It can then perform a thorough review, process large amounts of information, identify patterns and anomalies, and flag areas that deserve attention.
Work that once required a considerable amount of manual data collection and analysis can now be completed much faster.
That means I can spend less of my time gathering information and more of it answering the question that actually matters:
What does this data mean, and what should we do about it?
There are already more and more systems and AI agents capable of performing tasks that previously required manual work: preparing reports, summarizing information, sending emails, monitoring activities and connecting different systems.
For me, as a digital marketing agency owner, all of this raises a much more interesting question than “What can AI do?”
If I can accomplish much more in much less time, should my clients pay less?
What Happened to the 25-Hour Retainer?
Let’s take a very simple example from my own work.
If a client used to purchase a 25-hour monthly retainer, there was a certain amount of work I could complete within those 25 hours.
Today, I can accomplish much more.
Sometimes I don’t even need all 25 hours to produce the same amount of work. The hours simply shrink.
If my pricing is based entirely on time, the logical conclusion is that my fee should shrink too.
And this is where I start having a problem with the model.
If years of experience, deep professional knowledge and investment in advanced tools allow me to produce better work faster, does it really make sense that I should earn less because of it?
In other words, should we reward professionals according to how long a task takes them, or according to how well they can do it?
We Were Having This Conversation Long Before ChatGPT
Long before I started my own agency, I worked at a digital marketing agency.
Even then, we used to discuss the problems with hourly billing.
There was one sentence that came up repeatedly:
Clients pay for our knowledge, not our time.
Back then, it was mainly a discussion about business models.
In the age of AI, that sentence has taken on a completely different meaning.
Suppose an analysis once took me three hours. Today, with the tools available to me, I can perform it in one hour and potentially reach a better conclusion.
Has the value of that work fallen by two thirds?
I don’t think so.
AI Makes Execution Cheaper. It Doesn’t Necessarily Make Expertise Cheaper.
The holiday banner and campaign review are two very different examples, but together they illustrate the point.
AI can produce a beautiful banner for me at incredible speed.
But it didn’t decide that the client needed that banner in the first place. It doesn’t automatically understand the relationship between that brand and its customers, the right tone, or the fine line between a meaningful holiday greeting and yet another generic post people scroll past.
The same applies to campaign analysis.
An AI system can process an enormous amount of campaign data very quickly.
But when it identifies an unusual number, someone still has to decide whether that number actually represents a problem.
When it recommends a change, someone has to decide whether that recommendation makes sense in the context of the client’s business objectives.
AI dramatically shortens the journey to information and execution.
It does not eliminate the need to know what to do with them.
Execution is becoming faster and cheaper. Judgment is not.
We Are Selling Less Execution and More Decision-Making
For many years, a large part of digital marketing was execution.
Write. Design. Upload. Check. Prepare a report. Summarize the data. Build a presentation. Send an email.
AI can now shorten many of these processes dramatically.
But it doesn’t remove the important questions.
What should we do next?
Where should we invest the budget?
Which audience isn’t working?
Why are people visiting the website but not contacting the company?
Is the problem with the campaign, the message, the landing page or perhaps the offer itself?
Do we really need more content right now, or should we stop and fix something more fundamental first?
And perhaps one of the most important decisions in an era when we can produce almost anything quickly:
What should we not do?
These decisions come from experience, understanding the client’s business, knowing the market and being able to connect many different pieces of information.
AI can help me reach a decision faster. It can collect information, present possibilities and highlight things worth investigating.
But the responsibility for the decision is still mine.
Is an Hourly Retainer Still the Right Model?
This is a question I find myself asking more and more.
If one hour of my work today can produce significantly more than it did five years ago, what exactly does a 20-hour or 25-hour retainer mean?
Should we continue selling hours?
Or should we start pricing our services according to responsibility, scope, deliverables, availability, expertise and the decisions we are expected to make?
I don’t have a definitive answer yet.
I also don’t think hourly retainers are going to disappear tomorrow. They are simple, familiar and relatively easy for both the agency and the client to understand.
But I do think the model is starting to creak.
Because under a purely hourly model, the more efficient I become, the less I am supposedly worth.
And that makes very little sense.
Clients Should Benefit From AI Efficiency Too
There is another side to this discussion, and I don’t think we should ignore it.
If technology allows me to perform work faster and at a lower cost, it would be difficult to argue that all of the benefit should remain with the agency.
Our clients should benefit from this change too.
But that benefit doesn’t necessarily have to come only in the form of lower fees.
Clients can receive much more within the same working framework.
More testing.
More variations.
Faster response times.
Deeper analysis.
The ability to test an idea before investing significant money in it.
And more time spent on strategy instead of hours spent on repetitive technical work.
So perhaps instead of asking only, “How many hours did we save?” we should ask:
What can we now do with the time we saved?
There Is Another Risk: We Can Produce Much More of What Nobody Needs
This may be one of the biggest problems I see right now.
AI enables us to produce enormous quantities of content and creative assets.
If we once created five social media posts, today we can create 30.
If we once produced three versions of an ad, today we can produce 20.
If writing an article once required a full day, today it is technically possible to produce several articles in a day.
But who says we should?
More content does not necessarily mean better marketing.
More ads do not necessarily mean a better campaign.
And more activity certainly does not guarantee more leads.
Perhaps in an era when we can do almost everything faster, one of our most important responsibilities as marketers will be knowing when to stop.
So What Are Clients Actually Paying For?
I think AI is forcing us to return to this fundamental question.
If a client is paying me only for the time it takes to perform a task, then yes, AI should significantly reduce the cost of my services.
But if the client is paying for my ability to understand their business, identify a problem, select the right tools, make decisions, use AI intelligently and understand what to do with its output, the picture looks very different.
Artificial intelligence is certainly reducing the cost of executing digital marketing.
I’m not convinced that it is reducing the value of digital marketing expertise.
In some ways, it may be doing exactly the opposite.
Because when the ability to produce becomes available to almost everyone, the ability to know what to produce, why, for whom and what to do next becomes far more valuable.
Perhaps it is time for us, as marketers, to stop measuring our value only by the number of hours we work.
And perhaps our clients should start asking less:
“How many hours am I getting?”
And more:
“What can you actually do for my business with those hours?”
