Digital Marketing Is Getting Easier. Being a Good Marketer Isn't.

By James Arnold
Rooster Chief Digital Officer

There was a time when much of digital marketing's value came from simply knowing how to do the work.

You knew how to build the campaign. You understood the platform. You knew which targeting options to select, how to structure an audience, where to place the pixels, how to create enough variations to test, and which levers to pull when performance started moving in the wrong direction. There was strategy involved, certainly, but there was also a significant advantage in knowing how the machinery worked.

That advantage is changing quickly.

Google, Meta, LinkedIn and other platforms are automating more of the decisions marketers used to make ourselves. AI is making analysis that once took days possible in minutes. Creative variations that once required significant time and budget can increasingly be produced at scale. Research is faster. Optimization is faster. And the connections between media, analytics, CRM systems and business data are beginning to create possibilities marketers have talked about for years.

That all sounds like digital marketing should be getting easier. In many ways, it is. The interesting part is that being a good marketer may be getting harder.

We are handing over the keys

Bethany Andrew, Executive Marketing Consultant at State of Marketing, sees the same trend across the dominant digital platforms.

“All of them are going more into this black box of AI,” Andrew said. “They really want marketers to hand over the keys to them to determine where and what your ad is placed.”

There are good reasons to hand over some of those keys. Andrew has seen the bidding technologies in Google and Meta become very good at optimizing toward the conversion events marketers select. Google's Performance Max can make decisions across a massive ecosystem. Meta's Advantage+ can identify and pursue potential converters without requiring marketers to manually define every audience.

That's an enormous amount of computing power working on our behalf. But there is a tradeoff.

“It makes it really hard for marketers to be able to pull any optimization levers because we really don't have the visibility into where, what or why things are converting,” Andrew said.

Darcy Schultz, Associate Director, Business Strategy & Development at Goodway Group, sees the same loss of transparency.

“We've lost some transparency in a lot of those platforms where AI has been implemented,” Schultz said. “A lot of those algorithms that optimize campaigns behind the scenes, we don't have as much insight into as we used to.”

That isn't necessarily an argument for returning to manual marketing. We aren't going back there, nor should we. The bigger question is what we do with all of this new power.

The marketer can't beat the machine at being a machine

There is something a little strange about fighting automation simply because we used to do the work ourselves.

A platform can evaluate more signals than we can. AI can analyze larger datasets than we can. Automated bidding can make decisions faster than someone sitting at a desk checking campaigns every morning. There are things machines should do.

Schultz points to analytics as one example. Historically, marketers might have needed a month of data before they could confidently identify a performance trend. Smaller campaigns might not have generated enough information to evaluate until a quarterly report. AI changes the speed of that process.

“That data can be analyzed so much quicker,” Schultz said. “So you can make some of those optimizations from a human touch point a lot faster.”

The objective shouldn't be human or machine. It should be machine, then human. Let the technology process the enormous dataset. Let it identify the anomaly. Let it recognize the pattern. Let it make the thousands of routine decisions that don't require our attention. Then give the marketer better information, sooner, so we can make the decisions that do. Because the machine still doesn't know everything we know.

The weather still matters

Agriculture provides a particularly good example of where human context remains critical. An advertising platform can know that conversion rates have dropped in a geographic area. It can adjust bids, audiences or delivery based on that performance. What it may not understand is why.

Maybe there was a major weather event. Maybe planting was delayed. Maybe harvest moved earlier. Maybe commodity prices changed the economics of a purchasing decision. Maybe a dealer in that market changed ownership. Maybe an agronomist with tremendous local influence left a retailer. Those aren't necessarily advertising signals. They're business signals.

“The one thing AI can't do is it can't analyze regional or local factors, external factors that are happening,” Schultz said. “That's where you need that human touch.”

The technology will undoubtedly get better at incorporating external information. But Schultz's larger point remains. Performance data doesn't exist in isolation. Someone has to understand the business well enough to know what the numbers actually mean. That may become one of the defining skills of the next generation of digital marketers. Knowing Google Ads matters. Knowing agriculture matters more.

The bigger opportunity isn't inside the advertising platform

For years, we have talked about digital marketing platforms as though they were individual destinations. Google is over here. Meta is over there. The website has its analytics. The sales team has a CRM. Dealers have another system. Sales information may exist somewhere else entirely. And somebody eventually pulls all of those numbers into a report and tries to explain what happened. The bigger opportunity ahead isn't another new advertising feature. It's connecting the system.

Schultz sees new technology beginning to address something that has been particularly difficult in agriculture: Connecting what happens in marketing to what happens much farther down the sales chain. She points to emerging platforms that can help brands understand dealer or cooperative sales activity that historically existed in silos. If that information can be tied more effectively into CRM and marketing data, we get closer to understanding what marketing actually contributed.

“Now you're getting more sales data information that can be tracked back at a corporate level,” Schultz said. “Using that data and really building out your CRM at the same time is just going to help close that attribution gap that we've seen for so long, especially in longer buying cycles across agriculture.”

We've wanted that for a long time. The difference is that the technology is finally making some of those connections more practical. And AI adds another layer. It gives us a way to make sense of data that may technically have been available before but was too fragmented, too cumbersome or simply too time-consuming to analyze.

The future of digital marketing isn't another dashboard. It's getting the pieces to work together.

CRM isn't particularly exciting. It is increasingly essential.

It's easy to get distracted by whatever new AI product or platform capability launched this month. CRM doesn't have quite the same novelty. But the more automated the marketing ecosystem becomes, the more valuable good first-party business data becomes.

“If CRM isn't something that's at the top of your list to work on or you haven't implemented yet, you definitely need to start there,” Schultz said.

Andrew makes a similar point from the media side. If marketers are going to allow platforms to make more decisions, clean conversion tracking and a reliable source of truth become essential.

“You have to make sure your conversion tracking is set up correctly,” she said.

That's because a platform can only optimize toward what we tell it matters. Give the machine a weak signal and it can become extraordinarily efficient at finding more weak signals. Tell Google that a low-value website action is success, and Google will find people likely to complete that action. Tell Meta to optimize toward the wrong conversion, and Meta can get very good at producing it.

The sophistication of the algorithm doesn't compensate for a bad definition of success. It magnifies it.

Creative is about to become a very different game

Not all of this loss of friction is scary. Some of it is fantastic.

Schultz points to creative production as an area where marketers now have capabilities that would have been prohibitively difficult or expensive only a few years ago. The idea of serving different creative based on audiences or moving consumers through sequential messaging isn't particularly new. We've had versions of those capabilities for years. Actually producing everything was the problem. You could envision 20 different creative variations. Then someone had to make 20 different creative variations.

Today, platforms and AI tools can use existing photography, video, messaging and product information to develop variations at a scale that dramatically changes the economics of testing.

Andrew is seeing improvements there as well. While she remains cautious about blindly opting into every automated creative enhancement, she has been pleasantly surprised by some of the variations platforms can now produce. This matters because one of the biggest constraints on digital marketing experimentation has never been our ability to imagine a test. It has been our ability to afford one. And that barrier is falling.

New doesn't automatically mean useful

Of course, there is another side to all this innovation. Schultz remembers a time when a major platform might introduce something meaningful every six months or once a year.

“Now they're launching new things monthly,” she said.

Anyone working hands-on in digital platforms knows the feeling. You log in one morning and something has moved. There's a new setting. An old control is gone. There's another AI recommendation. A new campaign type is in beta. A platform wants permission to rewrite something, resize something, retarget something or optimize something you didn't ask it to optimize.

Keeping up is becoming a job in itself. But keeping up doesn't mean using everything. Schultz argues marketers should know what's coming, understand what it does and maintain budget for experimentation. But every new capability still has to pass a much older test: Does this solve a business problem?

“There are a lot of new things that are popping up on these platforms that are not the best way to be spending your money, even in a test,” Schultz said.

That's a particularly important reminder in an AI environment. Novelty is not strategy.

Even the definition of a platform is changing There's another complication coming. We're getting more places to market.

ChatGPT and other AI environments are developing advertising and commercial opportunities. Reddit has become increasingly relevant to how people research products and how AI systems find information. TikTok is a discovery and search environment. YouTube is video, search, entertainment and education simultaneously. LinkedIn continues to offer unusually precise B2B targeting. X remains relevant to parts of agriculture despite the challenges marketers may have with the advertising environment.

And those platforms themselves keep changing.

So the question isn't simply, “Which platforms should we use?” It's becoming, “What role does each environment play in the customer's decision?” It gets us away from buying media because a platform exists and back toward understanding the person we're trying to influence.

So what is the marketer's job now? If platforms increasingly choose audiences, placements and bids; AI helps analyze the results; automated tools create variations of our ads; and interconnected systems become better at finding patterns across all of it, what exactly are we supposed to do? More than ever. But probably less of the stuff that used to consume our days.

We need to decide what the business is trying to accomplish. We need to understand the customer. We need to know which signals actually indicate progress. We need to build better first-party data. We need to connect marketing activity with CRM and sales information. We need to determine which technologies deserve experimentation and which are distractions. We need to understand the external realities an algorithm can't see. And we need to give increasingly powerful machines better instructions about what success actually looks like.

The marketer's value is moving away from operating the machinery and toward making better decisions about what the machinery should accomplish.

For decades, digital marketing has promised us better targeting, better measurement, better personalization and better optimization. We're finally reaching a point where the technology can deliver many of those things at a scale humans can't match. The question is whether we're prepared to do something more valuable with the time and capability that gives us.

Because digital marketing is going to keep getting easier. Knowing what marketing should do next won't.