Why Your Best Campaign Can’t Fix a Targeting Problem

Let me start with something I genuinely believe: most agency media buyers are very good at what they do.

I've watched media buyers take a struggling campaign and turn it around in 48 hours. I've seen them read a data set and spot the signal that everyone else missed. I've worked with people who can look at a Meta dashboard and know, almost instinctively, which lever to pull next. That skill is real. It's hard-earned. And in the right conditions, it produces exceptional results.

But here's what I've also watched happen, more times than I can count: a genuinely talented media buyer, running technically excellent campaigns, producing mediocre results. Not because they're doing anything wrong. Because the foundation is wrong. And no amount of optimization can fix a flawed foundation.

This was always true. But something has changed in the last two years that makes it more true, and more urgent, than ever.

The shift most agencies haven't fully absorbed

If you're running Meta ads in 2025, you've noticed the shift toward Advantage+ and broad targeting. Less manual audience selection. More reliance on the algorithm. A lot of agencies have interpreted this as "targeting matters less now." The platform handles it. Just go broad and let the AI find your buyers.

That interpretation is dangerously wrong. Targeting hasn't become less important. It's moved.

In the old world, targeting lived in the ad set. You picked interests, demographics, lookalikes, custom audiences. You told Meta who to show the ad to. The creative's job was to convert whoever saw it.

In the current world, the creative IS the targeting. Meta's ad delivery engine, Andromeda, reads the language in your ad copy, the imagery, the tone, the hooks, and uses those signals to decide who sees the ad. Your words tell the algorithm who to find. The phrases you use, the fears you name, the desires you speak to, the specific language that makes one person stop scrolling while another keeps going: that is now the primary mechanism that determines who your ad reaches.

This isn't a subtle shift. It's a fundamental change in how advertising works on the platform. And it means that knowing your customer's language isn't just helpful for writing better copy. It's the input that tells Meta who your customer is.

The language ceiling

There's a concept I think about a lot that I've never heard anyone in the agency world talk about. I call it the language ceiling. It's the maximum performance you can extract from a campaign when you don't know how your customer actually talks.

Here's what I mean. Say you're running ads for a supplement brand. You don't have deep research on the target customer, so your ad copy uses general benefit language. "Support your gut health naturally." "Clean ingredients for a healthier you." "Feel the difference in 30 days." This is competent copy. It's clear. It communicates a benefit. It looks like it should work.

And it does work. At a 3x ROAS. Maybe 4x on a good week. Your media buyer tests different hooks, different creative formats, different calls to action. They iterate. They optimize. After a month of solid work, performance has improved, but it's plateaued. The campaign isn't failing. It's just not breaking through.

Here's what's happening under the surface. Your ad copy is speaking in general wellness language. Meta's algorithm reads that language and shows the ad to people who engage with general wellness content. That's a massive, undifferentiated audience. The algorithm is doing its job. It's finding people who respond to what you wrote. The problem is that what you wrote could apply to 50 million people, and you need to reach the 50,000 who will actually buy this specific product.

Now imagine you know the customer. Not the demographic. The actual person. You know that the woman who buys this specific gut health supplement isn't searching for "gut health." She's searching for "why am I bloated every afternoon." You know she's tried two other supplements that didn't work and she's skeptical. You know she trusts clinical studies more than influencer recommendations. You know the phrase that stops her mid-scroll isn't "support your gut health" but "I tried everything for the 3pm bloat. Here's what finally worked."

When your copy uses that language, Meta's algorithm reads a completely different signal. Instead of "show this to general wellness people," the signal says "find the person who has this specific problem and is looking for this specific type of solution." The algorithm gets a precise address instead of a zip code. And it finds her.

That's the language ceiling. Your media buyer hasn't hit the limit of their ability. They've hit the limit of the language. And the language is limited because nobody did the research to find out how the actual customer talks, thinks, and searches.

Why this matters more than it did two years ago

In the old targeting world, a media buyer could compensate for vague copy with precise audience settings. You could write generic ad copy and still reach the right person if your lookalike audience was dialed in or your interest stacking was sharp. The ad set did the work of finding the person. The creative just had to convert them.

That safety net is gone. With Advantage+ and broad targeting as the default, the audience settings are wide open. The algorithm is looking at your creative to decide who should see it. If your copy speaks in broad, generic language, the algorithm targets broadly and generically. If your copy speaks to a specific person with a specific problem in their specific language, the algorithm finds that person.

This is why I see agencies that are technically excellent at campaign management hitting a wall they can't explain. They've mastered the platform. They know every lever in Ads Manager. But their creative is working with vague customer language because nobody did the upstream research to find the precise language. And in 2025, vague language doesn't just mean weaker ads. It means vague targeting. Because the language and the targeting are now the same thing.

The agencies that understand this shift are the ones writing ad copy that sounds like their customer's internal monologue. Not "improve your energy levels." But "I'm tired of being tired by 2pm and I've tried everything." Not "premium joint support for active adults." But "I stopped running six months ago because of my knees and I miss it every day." That specificity isn't just better copywriting. It's a targeting instruction. It tells the algorithm exactly who to find.

When the language changes, the ceiling disappears

I worked with a sports memorabilia brand that was stuck at exactly this point. They were running ads to "collectors and sports fans." Their copy used general collector language: "authentic memorabilia," "limited edition," "for the true fan." ROAS was hovering around 3x. The agency was good. The campaigns were well-built. Everyone had settled into the idea that 3x was just what the market would bear.

When we did the upstream research, the customer who emerged wasn't "collectors and sports fans." It was a much more specific person with a much more specific motivation for buying. The language that person used to describe why they collected was different from what the ads had been saying. The emotional trigger wasn't "authenticity" or "limited edition." It was something more personal and more specific.

The ad copy shifted to reflect that language. And because the language changed, the signal to the algorithm changed. Meta stopped showing the ads to general sports memorabilia browsers and started finding the specific person the research had identified.

Same product. Same platform. Same agency. Same budget. ROAS went from 3x to 12x.

The agency hadn't been doing anything wrong at 3x. Their media buying was excellent. Their creative process was professional. What was missing was the language. They were sending the algorithm a vague signal, and the algorithm was doing exactly what they asked: finding a vague audience. Once the signal got specific, the algorithm found the specific person. And that person bought.

What this means for how agencies think about research

Here's what I want every agency owner to sit with for a minute.

If creative is now the targeting, and the algorithm uses the language in your ads to find the right person, then knowing your customer's exact language isn't a copywriting exercise. It is the most important strategic input in the entire campaign.

Not the audience settings. Not the budget allocation. Not the bid strategy. The language. The specific words your customer uses to describe their problem. The phrases they type into Google at 11pm. The way they talk about their frustration to their spouse. The internal monologue that's running when they see your ad and either stop or keep scrolling.

Most agencies get this language from one of three places: the client's intake form, competitive research, or their own experience in the category. All three produce general language. Useful language. But general. And general language produces a general signal, which produces general targeting, which produces a general audience, which produces a 3x ROAS when the product should be doing 8x or 12x.

The research required to find the specific language, the real words your customer uses, the emotional triggers that actually drive purchase behavior, the fears and desires that are specific enough to function as a targeting instruction, doesn't exist in most agency workflows. It's not a step in the process. The process jumps from "sign the client" to "write the ads" with an intake form and a competitive scan bridging the gap.

That gap was always expensive. In the Advantage+ era, it's the single biggest constraint on campaign performance. Because the algorithm can only work with what you give it. And what you give it is the language.

The uncomfortable question

If you've been running an agency for a while, there's an uncomfortable question in everything I've just described: How many of your current clients are sitting at their language ceiling right now?

How many of them are getting decent results, 3x, maybe 4x ROAS, that could be significantly better if your ad copy used the specific language of a researched customer instead of the general language of a category? How many campaigns are plateaued not because your media buying hit a wall, but because the creative is sending the algorithm a signal that's too broad to find the right person?

This isn't comfortable to sit with. Because it means some of the campaigns that plateaued, some of the clients who churned, some of the results that felt stuck despite your best efforts, weren't execution failures. They were language failures. And the language failed because the research to find the right language was never done.

The media buyer didn't fail. The creative team didn't fail. The process failed. The step that should have produced a specific, researched customer profile, complete with their actual language, fears, desires, and buying triggers, simply didn't exist. And in 2025, that missing step doesn't just mean weaker copy. It means weaker targeting. Because they are now the same thing.

The new foundation

I'm not suggesting that agencies stop optimizing. Optimization is essential. The skill of reading data, finding patterns, and making smart adjustments is the core of what makes a good agency valuable. None of that changes.

What I'm suggesting is that optimization has a prerequisite that most agencies skip. The prerequisite is knowing, with real research and real confidence, who the customer is and how they talk before the first ad is written.

When that research exists, everything downstream changes. Your copywriter doesn't write general benefit language. They write in the customer's voice, using the customer's words, naming the customer's specific fear. Your media buyer doesn't just launch and iterate. They launch with a creative signal precise enough that the algorithm finds the right person from day one. Your creative testing isn't about finding the customer. It's about finding the best way to speak to a customer you've already identified.

The difference between 3x and 12x isn't usually better creative skills or smarter bid management. It's better language. And better language comes from better research. From knowing the one customer so specifically that every ad reads like it was written for them.

Because in 2025, it should be.