Can You Replace Your Email Marketing Agency With AI?

Can AI Replace an Email Marketing Agency?

I own an email marketing agency, and honestly, AI is getting dangerously good at doing our job. It can write emails, design them, analyze customer data, build strategies, and it's actually starting to execute the work, too.

So that raises a question that probably isn't great for my business: do e-commerce brands even need email marketing agencies anymore?

Instead of giving you a biased answer that you probably expect from an agency owner, I'm going to break down everything we actually do for our clients, show you what AI can already replace, where it still falls short, and whether I'd actually trust it to run email marketing for e-commerce brands today.

What an Email Marketing Agency Actually Does

First, we need to define what an email marketing agency actually does. Because if you think you're paying an agency to write four emails a week and make them look pretty, then honestly, AI is already getting pretty damn close to replacing that.

But good email marketing management should go a lot deeper for our clients. I would break the job down into five major categories.

Number one is strategy, which is figuring out what we should actually be doing.

Number two is creative, which covers everything from copywriting all the way to design.

Number three is data and analytics, which is understanding what's happening and why.

Fourth is optimization. This is where we improve the customer life cycle of our clients' customers over time.

And finally, execution, which is actually building everything, launching it, and making sure it works.

So let's go through each one and see how close AI actually is to replacing us.

Strategy: What Could Be Done Versus What Should Be Done

Let's start with strategy, which is figuring out what we should actually be doing right now.

AI is very good at giving strategies. I can ask ChatGPT and give it information about a brand: what they sell, their average order value, customer behavior, existing email program, etc. Within seconds, it can give me a pretty impressive retention strategy.

It might recommend improving the welcome flow, building a better post-purchase experience, creating a win-back automation, whatever. And most of these recommendations might be completely reasonable.

However, the problem is that AI tends to be very good at telling you what could be done, but it's much harder to know what should actually be done. Because in a real e-commerce business, you have limited time, limited resources, and a million different things that are more of a priority.

Maybe the biggest opportunity isn't necessarily another Klaviyo flow, for example. It could be that customers aren't coming back because the product just naturally has a six-month purchase cycle. Or maybe your acquisition quality has dropped. Maybe you're discounting so aggressively that you've trained customers to never buy at full price.

AI doesn't have these nuances. And that's where human intervention really gives you an advantage.

To simplify, AI can give you 20 or 50 or a million different things you could be doing, but a good strategist should be able to tell you what are the two or three that actually matter. And that's where I think humans have a significant advantage.

AI has access to an unbelievable amount of knowledge, but knowing something and understanding the nuance of a specific business are two very different things.

So for strategy, AI is an incredible tool I use myself, but I wouldn't blindly let it make final decisions.

Creative: Copywriting and Brand Context

Next, we have creative, which for email marketing mostly means copy and design. And this is where AI has made the most obvious progress.

Let's start with the copy. I could ask ChatGPT to write me, for instance, a Labor Day email for an apparel company offering 25% off, and it's going to give me something usable in 5 to 10 seconds.

But if you do that for 10, 15 different apparel brands, you'll start to notice that a lot of it sounds exactly the same. Things like, "Your wardrobe deserves an upgrade. Fall favorites are here. Don't miss out. Shop now before it's gone."

AI is really good at producing acceptable copy, but acceptable copy is becoming pretty much worthless now because everybody is seeing it everywhere. It's super saturated.

Where it becomes much more powerful is when you actually give it context about the brand. You have to feed it information about the products, customers, previous campaigns, reviews—everything nuanced and specific to that brand—in order to get the best performance out of it.

We've been incorporating more AI into this process ourselves, but we're not just opening ChatGPT or Claude and saying, "Write an email." We give it enough information that it actually understands the business before it writes anything. And you can actually create unique GPTs for each of your individual clients, so that nothing gets repeated across the board.

So can AI replace an email copywriter? Honestly, for generic e-commerce copy, the answer is 100% yes.

Creative: Email Design and the Designer's Changing Role

But creative isn't just copy. There's also the design component. And this is probably the area where AI has surprised me the most.

As of filming this video, GPT Astra has made remarkable progress with the design components of email marketing. You can give it a brand, creative direction, product images, an idea for a campaign, and it can generate a complete email design in minutes.

Some of the stuff it produces is genuinely impressive. We're not just talking about generating random images anymore. It can actually put together the structure of the email, typography, product sections, imagery, calls to action, and the overall creative direction.

What's made this even more useful is that now we have features to export all of this into Figma. So instead of expecting AI to create the perfect finished design, you can get 80 or even 90% of the way there, export it to Figma, change the spacing, swap an image, adjust the typography, and suddenly the workflow looks completely different.

And if you actually want to know how you can do this yourself, I included a link below that's going to give you all the information you need to know.

That said, is it better than a good graphic designer? No, at least not yet. A great designer still has something that AI struggles with, which is ultimately taste. They can look at something and immediately know if something feels off or if there's an obvious rule being broken.

Good designers understand hierarchy, consistency, restraint, and brand identity better than any AI can ever know. And all of those little details are what separates emails that look pretty good from something that actually looks great.

AI still makes weird decisions, too. Sometimes the spacing is slightly off. Sometimes the hierarchy doesn't make sense. Sometimes something looks amazing at first glance, but the longer you look at it, the more problems you notice.

But here's where I think things get uncomfortable. The comparison shouldn't be AI versus the best graphic designer in the world. It should be AI versus the average production designer building relatively straightforward e-commerce emails every day. And that gap is getting smaller and smaller.

So I don't think great designers disappear. I think their role ultimately is just going to change. Instead of spending hours building every email from scratch, AI generates the starting point and the designer becomes more of a creative director: actually choosing the direction, refining the work, making the decisions AI still isn't very good at making.

And copywriting is already moving in this direction.

Data and Analytics: Powerful Insights, Confident Mistakes

Next, we have data and analytics. This is where AI is the strongest today.

AI is really good at gathering information, processing huge amounts of data, and presenting it in a way that's incredibly easy to understand and digest.

Instead of spending hours going through spreadsheets, reports, and dashboards, you can now start asking questions. What changed this month? What campaigns underperformed? Which products are driving the most repeat purchases?

AI can take enormous amounts of information and turn that into something that is easy to understand incredibly quickly.

This also opens up some really interesting possibilities with segmentation. Most e-commerce brands have an enormous amount of customer data sitting inside Shopify and their email platform, which could include things like what people bought, when they bought, how much they spent, what categories they're interested in, how frequently they purchase, when they've clicked, what they've clicked, and how long it's been since their last order.

Imagine being able to ask which customers are most likely to make a second purchase in the next 30 days, or what customers usually buy after purchasing a certain product. That's where AI is incredibly powerful.

But there's a pretty major problem. AI can also be wrong. And worse, it can be wrong with an incredible amount of confidence. Shopify Sidekick is a prime example of this.

AI Data Analysis Still Needs Human Judgment

As of this video, I've asked Sidekick questions about store data, customer LTV, and gathering average order value. It has a hard time differentiating between, for instance, wholesale versus online store data, time frames when customers bought, and maybe two numbers are being compared incorrectly. Sometimes the AI just makes a genuine mistake.

This isn't unique to Sidekick, though. ChatGPT has done it, Gemini has done it, Claude can do it. That's why there's a really important distinction between AI being incredibly good at working with data and AI being the final authority on what that data means. Those aren't the same thing.

You still need someone who understands the business well enough to look at the output and say, “Hold on a second, this doesn't make sense.” You need someone who understands the nuance behind the numbers, knows which questions to ask, knows when a metric might be misleading, and can validate the conclusion before you start making business decisions around it.

Attribution is a perfect example. Klaviyo might tell you that email generated 30% of your revenue, but that doesn't mean email caused 30% of your revenue. AI can't magically solve this problem. In fact, if you give AI misleading data, it can give you an incredibly sophisticated explanation of the wrong conclusion.

I think AI is going to become one of the most powerful analytical tools we've ever had. But right now, I still want a knowledgeable human between the AI and the final business decision.

Optimization: Turning Opportunities Into Action

Next is optimization. This is where I think the line between AI and humans gets a lot more interesting, because once we have the strategy, creative, and data, somebody needs to figure out how to actually improve the customer life cycle.

Let's say AI identifies that a huge percentage of customers buy once and never come back again. That's super useful, but what do we actually do about it? Maybe we need to build a better post-purchase experience. Or maybe customers need more education after the first product order or whatever.

This is where optimization becomes much more than just changing the subject line or A/B testing a button. You're trying to understand why customers behave the way they do and then change that behavior. That's really what retention marketing is all about.

The same thing applies to flows. AI can absolutely tell you that you should have a welcome flow, abandoned checkout, browse abandonment, post-purchase, whatever. That's the easy part. But the harder part is figuring out how these systems should actually work for this particular business.

Should a first-time customer receive the same abandoned checkout experience as someone who has purchased five times? When should you introduce a second purchase? What should that second product be? And more importantly, how should your brand operate differently from any other ecommerce brands in your industry?

Of course, AI can help answer all those questions. Over time, I think it's going to get extremely good at it. But right now, this is another area where I want AI surfacing opportunities and humans deciding what we actually do about them.

Execution: Actually Building and Launching

Finally, we have execution. This is actually building everything, launching it, and making sure it works. Right now, I think this is one of the biggest things preventing AI from completely replacing an agency. There's a huge difference between telling you what to do and actually doing it.

AI can tell me that a client needs a new post-purchase flow. Great. But someone still needs to build the segments, create the logic, write and improve the emails, finalize the designs, load everything into the platform, make sure the discount codes work, etc.

This is an area I expect to change fast, because AI is rapidly moving from something that you talk to to something that can actually take actions. But as of right now, humans are still far better.

I think in the future, though, we will have AI agents that can reliably do everything that we ask them to do, and they will understand the nuance between brands. But as of right now, that just isn't the case.

How Much of an Email Agency Can AI Replace?

So the question: can AI replace an email marketing agency? After looking at strategy, creative, data, optimization, and execution today, I don't think it can completely. But I also think it's the wrong question.

A better question is, how much of an email agency can AI replace? And the answer is already quite a bit.

It can replace generic copy, first drafts, basic campaign ideas, data processing, reporting, research, early-stage design work, and basic strategy recommendations. Increasingly, more of the actual production work can already be accelerated or completely handled by AI.

If you're an agency whose entire value prop is, “We'll write your emails, design them, and schedule them in Klaviyo,” you should definitely be worried, because the value of that work has already gone down significantly.

That doesn't mean that good agencies will just disappear, because the real value of an email agency moves up the chain. The email agencies that understand the customer the best, understand the business, and can decide what actually matters—these are the agencies that will succeed.

Using AI to Make the Agency Better

Personally, I'm not trying to protect my agency from AI. I'm just trying to figure out how much of my agency AI can make better.

If something we're currently spending five hours doing can be done in 30 minutes with AI, then I don't want someone on my team spending five hours doing it. That's just not valuable anymore. I'd rather have them spending time actually moving the client's business forward and doing things that AI can't do as well. Could be coming up with better ideas, analyzing customer behavior, improving the strategy, etc.

It doesn't necessarily eliminate the agency, but it eliminates a lot of the work that agencies used to get paid to do, which means the agencies that survive will probably look very different. Smaller teams, significantly more leverage, much more automation, and humans spending more of their time making decisions rather than manually producing things. Potentially, those agencies become significantly better because of it.

That's what I'm going to be experimenting with inside my own agency over the next year. I'm not particularly interested in using AI just to write more subject lines. I'm actually interested in what happens when AI can understand customer behavior, purchase data, campaign performance, and so forth.

Should Your Brand Hire an Email Agency?

Whether you're using AI or not, there's still another question you need to answer. Should your ecommerce brand even hire an email agency in the first place?

Because I own one, you might obviously think I'm going to say yes. That's not necessarily the case. I actually think there's a lot of ecommerce brands out there that have absolutely no reason to pay an agency thousands of dollars every month.

In my last video, I broke down exactly when I think hiring an email marketing agency makes sense, when it doesn't, and what you should look for if you're considering one. It also includes some of the biggest red flags that I'd stay away from.

So click the video on your screen right now, and I'll see you over—