Rethinking media strategy in the age of algorithms
For a long time, building a digital media strategy meant exercising a certain amount of control over how it was delivered: we defined audiences, chose placements, tested two versions of an ad, isolated a single variable, and eventually found a winner. A good part of our expertise as strategists was built around this ability to understand the different delivery levers, adjust them, and search for the best-performing combination.
But over the past few years, I've watched these levers slip away from us, little by little. Many of the decisions we used to make are now entrusted to advertising platforms' algorithms: who to target, where, when, and with which combination of messages?
Naturally, this reality leads marketing strategists like me to ask a somewhat uncomfortable question: if the algorithm is making more of the decisions, what does that make me?
You could say our role is shrinking; I'd say it's shifting instead, and here's why!
In brief
- Advertising platforms now handle a large share of media optimization decisions, but optimizing those decisions isn't the same as doing strategy.
- Strategists are no longer there to control every variable of a campaign, but rather to define the problem to solve and the business framework within which the algorithm must optimize.
- Targeting isn't disappearing, but part of its role is shifting toward the creative itself: by running several messages built around different purchase motivations, each message is left to naturally find the right audience.
- This new role for creative only works if the messages are genuinely distinct, not just variations that say the same thing. And with generative AI, the risk of producing more without truly diversifying is growing.
- Answer engines and AI assistants could also give brand awareness more weight again, since part of the discovery process can now happen before someone even clicks through to a site.
- The more execution becomes automated, the more strategists' value lies in their ability to ask the right questions and make sense of the answers.
When platforms change the game
In my early years optimizing digital campaigns, the logic was fairly intuitive: test, compare, then select. We pitted two audiences, two messages, or two landing pages against each other, isolating a single variable at a time as much as possible. The goal was to understand what explained the performance, then gradually shift budgets toward whatever worked best.
This way of working still has its place; it hasn't become bad or outdated. What's actually changed is the environment. Major platforms can now make thousands of decisions simultaneously, drawing on a volume of signals that no human team could process the same way. At that point, trying to keep control over every variable becomes less realistic. But more importantly, I'm no longer convinced that's where the real value of strategy lies.
In the constant search for the best combination of audiences, messages, and placements, it's easy to end up confusing optimization with strategy. Optimizing means choosing the best option among those that already exist. Doing strategy also means deciding which options deserve to exist in the first place: what problem are we really trying to solve, and what will count as success?
This distinction becomes especially important as platforms take on a larger share of optimization. They work from the framework we give them, without knowing the full business context or the real value of each outcome to the company. Our role as strategists is therefore as much about defining what we're trying to move forward as it is about interpreting performance in light of the business's stakes.
When creative becomes the new targeting
Recently, with a client, we tested three different creative angles for the same ad campaign: one built around price, one around speed, one around trust. And unlike what a classic A/B test might have produced, no single option came out as the "winner." All three approaches performed well, just not with the same people.
That's where it gets interesting: it's no longer necessarily about finding the right formula and eliminating the rest, but about letting different messages reach different motivations.
This shift isn't limited to one client or one industry. Advertising platforms now work with multiple audience types and creative variations at the same time. Targeting is still there, but it now rests on both the audiences we choose and the diversity of the messages we run.
Given this, the trap would be to think it's now just a matter of producing more creative. But having ten variations of the same message doesn't get us very far. What matters is putting genuinely differentiated angles into circulation, ones that speak to several different motivations. That's far more than simple micro-variations in color or calls to action!
That said, this remarkable optimization capability comes with a caveat: platforms can very quickly identify what performs best and adjust delivery accordingly, but they can't say why a given message worked better than another. Their job is to optimize, not to explain! And that's exactly where strategists remain relevant, because someone still needs to understand what that performance reveals about the market and carry that learning forward beyond the current campaign.
The creative limits of generative AI
Producing large numbers of variations in a matter of seconds has become child's play with generative AI. But more variations doesn't necessarily mean more ideas!!
Early research on AI-assisted creativity actually points to a risk of homogenization: trained on similar content, these tools tend to converge toward the same formulas. Their output can be good individually, but it all ends up looking alike when viewed as a whole. It's therefore up to us to bring more diversity to the table, not just more variations.
When the decision happens before the click
In addition to changing how advertising creative is made, artificial intelligence is also transforming the way people discover a brand. In the past, digital marketing focused on what could be measured quickly, like the number of clicks, the conversion rate, or the cost of acquisition. Now, with answer engines and AI assistants, people can form an opinion about a brand or product before ever clicking through to a website.
The firm McKinsey also reports that more than 70% of people conducting an AI-assisted search do so right at the very start of their journey, to learn about a category, brand, product, or service. Meanwhile, the Pew Research Center found that internet users click noticeably less on traditional search results when an AI summary appears on Google. In other words, a good portion of the decision can now happen before the click.
Does this trend spell the end of performance marketing? I don't think so! That said, it does push us to pay closer attention to what happens before conversion: awareness, brand credibility, and the content that helps people discover and understand it.
For strategists, the question is no longer how to capture existing demand, but rather how to ensure the brand is already present at the very first signs of that demand.
Controlling less, questioning better: the new reality
So, what's left for digital strategists like me if algorithms are making more of the decisions in our place? I think the answer lies less in the platforms themselves than in what we manage to draw from them! Of course, algorithms can decide which creative to run, to whom, and when. But it's still up to the humans behind the scenes to make sure every optimization cycle leaves us with a better understanding of the market as a whole, not just a better result within the platform.
If algorithms become better optimizers, all the better! It's a wonderful opportunity to bring strategy back to its original essence: asking the right questions before looking for the right answers.
