Doesn't the System Do It for Me?

Where Is the Human Element?

Since the late 2010s huge advancements have been made towards automation taking more and more control of where and to whom our ads are delivered. At this point the systems we work on are doing our job for us! Or are they?

To understand what role the human marketing expert has today in digital paid advertisements, we need to better understand what exactly AI, AKA automations, the system, cutting edge tools, or the all powerful algorithm, is doing for us. To get into that, first, let's align on the basic elements of paid advertising online.

• Audience
• Ads (Creative and copy)

That's really it. Of course there are other vital elements, including SEO, GEO, tracking infrastructure, and more, but in the end it all boils down to two questions: Who are your potential customers, and what are you showing them to get them to convert. I'm not going to be touching too much on ad creative and copy today, which leaves the real meat of digital marketing. Audience.

Your potential customers. This is — at absolute best — everyone who has a chance to purchase your product. Of course, we can't be perfect; some potential customers will take more time, some need more convincing, some will go with a competitor, and some just won't be interested no matter what you do. A media buyer's job is to spend the least amount of time and money possible while reaching the largest amount of converting customers. To do this, we need to put some thought into who we are targeting; minimize showing our ads to people who won't convert, and build our audience based on real statistics, probability, and experience.

In other words, digital marketing is about finding the right people.

Here is something to keep in mind: this is one of AI's best strengths in marketing.

The goal isn't to understand how to outperform AI, the goal is to understand what key factors to monitor and what adjustments to make when considering modern, algorithmic, audiences.

"So, What Is the AI Doing?"

AI can use a customer list to find similar people, adjust your targeting while your campaign is running based on performance with specific audiences, match specific creatives and ad copy with the people who respond best, and provide estimated results for your campaigns based on what audience you are reaching. That is quite the list! So what is left for the marketer to do?

Past understanding how to use the tools — which is an undervalued skill for anyone who is accustomed to sitting on a computer all day, and a magical world of wizardry, even in 2026, for many business owners who spend their time fulfilling orders and making cold calls — what's left is understanding what the AI is getting right and what isn't working, what ground rules or limits to set when developing these audiences, and when and where to make the necessary adjustments to maximize your product's potential.

Let's Start with Why the AI Gets Things Wrong

Data is only perfect to an extent that the math is all right, but when it comes to making actions based on that information, now we are dealing with averages. As long as your business is totally average, then all of the results that usually work the best are going to be great for you! I haven't heard anyone describe their business as average. I actually don't think that it is possible for any individual business to have a totally middle of the road customer base.

AI builds your audiences based on averages. Let's start with what it will do first when building your audience, assuming that our example business has a good tracking system and a good data stream pointing back at their accounts.

If we give it a customer list to work with, then it will take the average of your list and work with that to start (great!). If you let it go with no restrictions, it is going to start to find the average person that engages with your ad, and eventually the average person who purchases from your ad.

Eventually, it's going to end up finding you a decent audience, assuming that during its initial experimentation you earned enough sales (or other goals) for it to learn enough about your audience.

This is a great starting point if you have absolutely no customer data.

Now that we have a better understanding of how AI builds audiences, let's see how to best use these tools without just letting them run loose!

Human Impact #0: Run Ads with Quality and Clear Messaging

You have heard this, your kid has heard this, your dog has probably been experimenting with it — creatives are the new targeting!

Really, in modern digital marketing, creatives are a targeting decision just as much as a positioning, comparison, or selling tool. Every modern ad system — like the media favorite example of Meta's Andromeda system — analyzes your creatives for signals that it can match to your specific audience members, even toggles on many platforms to allow the system to make changes to copy and creative on an auction by auction basis depending on what customer you are bidding on. In its current stage, it's difficult for me to recommend the AI creative optimizations for every creative, as I still see a lot of "hallucinations" for the on the fly edits… but it is constantly improving and I have high expectations for the near future.

The key to taking advantage of this is finding the right balance between budget and creatives. In modern digital marketing, it is almost always better to run more creatives than in the past for a single campaign to give the system the most to work with during testing and selecting which ad to deliver to which user, but if you don't have the budget for your ads, then you're wasting potential training for the system by forcing it to make either less educated decisions — ads don't get to run enough individually to truly test them — or some of your ads simply won't get any delivery because the system has already determined a "winner" using its limited budget.

Human Impact #1: Quality Customer Data from the Start Can Make or Break Your Audience

Build your audiences based off of your highest value customers

Having a clear understanding of what makes a high value customer to your business and ensuring that each platform you use is aligned with that goal is one of the most important steps of ensuring you maintain your audience quality as you grow.

If there is little variation between your customers or you don't rely on upgrades or higher order values, then running with minimal restrictions makes the most sense, but many businesses are not in this situation.

You might have a section of your audience who is only purchasing a specific low value package, doesn't include any addons, or in general has a low lifetime value. It is important to distinguish between these customers and your high value customers when selecting a customer list or specific web event(s) to build your audience off of.

Human Impact #2: Ensure That Your Business' Goals Match with What the System Thinks Your Goals Are

Even if you started with a strong audience to build off of, not having the right goals in place may decrease the value of your audience gradually and may take some work to break out of.

This comes in two forms: setting a target CPA/ROAS/CPC/whatever it might be for this specific campaign, and sending the right message back to the system when it does get an action. If you don't have proper product prices, order values, add to cart statistics, or are missing tracking on any of your website's pages, lead forms, or other important factors that could be used to distinguish a high value customer from the rest as quickly as possible, then your performance is all but guaranteed to be impacted.

The modern digital marketing space is all about the ad auction, and the less information you have on your customers, the more guesses and "averages" the system needs to use when determining which auctions to enter. So even if a strong foundation paints a great picture of the goal, if the system is running blind, there is a much smaller chance that it is going to be able to replicate it.

If you are setting target CPAs, ROAS, or just reach, paying close attention to how these goals interact with your performance, audience, and vary from platform to platform will prevent your targeting from scaling far past your target audience, or restricting itself into an ultra conservative bidding strategy halting your growth in its tracks.

Human Impact #3: Watch What It Learns and Optimize to Match!

When you review performance results, your only goal shouldn't be to send an accurate report to a client or understanding how profitable your ads are, although those are both important; you should still thoroughly inspect your audience and where/how you are successful.

Pay attention to what demographics are responding best to your product, specific ads, or have any other unifying factor. If I am diving into an account with a goal of understanding their audience, I make a couple of reports. Primarily I'm looking for age, placement, day, and location for any significant outliers. If my 65+ audience is hogging all of the reach because they are watching the video all the way through but actually are low value conversions, I might start looking in to either limiting delivery to that age group, or set up some re-targeting campaigns with creatives optimized for that age group.

Only after getting a baseline understanding of who and where our audience is will I start to dive into the overall KPIs like CPC, CPA, CPM… whatever makes the most sense for that business. When you have a baseline understanding of your audience's demographics, location, and web browsing/scrolling habits, the rest of your metrics are a lot easier to understand and define.

This is also extremely important to understand when you consider the higher amount of creativity recommended in modern days compared to even a few years ago. Your creatives need to be made in response to actual results from your campaigns and audience, not just based on what you expect your average customer to be like.

Human Impact #4: Keep Your Separate Audiences Separate, and Keep Your Similar Audiences Together

This is an important distinction to make when we optimize for the fastest learning possible.

When you keep all of your audience segments in one audience and put all of your ads together, that means the system needs to figure out which ads work with which part of your audience, and you pay for that learning from the time you hit publish. In much the same way that it is important to start your audience with a good foundation, you should be matching your audience to your creative right off the bat.

Let's assume you have been keeping up with steps 1-3, you have a great customer list to start with, you have good tracking sending data back, and the system understands what a high value customer is to you. Now, your goal is to continue to convert these high value customers and grow your business with this customer base.

Let's use a sunscreen brand for example. Our sunscreen is high quality, all natural ingredients, wildlife safe, and high SPF. Some of our audience respond great to the all natural wildlife safe aspect, while others are going to be more convinced by the high SPF and quality personal safety aspect of it. The ads that we run for these two audiences are going to look very different. Although there may be minor overlap between the two, these two audiences are likely dissimilar enough that any overlap concerns are outweighed by the time saved learning about your "creative targeting" from this quality foundation.

Another important element of using AI audiences is finding the right balance between keeping it open enough for testing and giving it the right guidelines to learn as quickly as possible.

Human Impact #5: Minimize Restrictions, but Be Quick to Stop a Leak

Throughout this audience building process (which really never stops) AI might start seeing something that no one else does, and something no one is looking for for good reason. If you spot a leak somewhere, maybe Advantage+ placements going too heavy on Messenger or Reddit's automated targeting is putting your ads in communities that don't make sense for your brand, be sure to stop it once you see it through the specific platform's brand guidelines/brand settings or in your individual ads or campaigns.

Conclusion

The better you understand how the system works the better you can understand how to actually get better at it. It all comes down to this: pay close attention to your progress, make adjustments when necessary, and trust the system to do its job.