“We know our personas”: why that certainty costs you on Meta

In short

Ever since Meta’s algorithm started choosing on its own who sees your ads, the content of your creatives has been doing the targeting. A persona described in a few demographic words produces creatives that speak to everyone, and the algorithm brings you everyone. Each of those visitors costs money, whether or not they belong to your core target. A useful persona meets three conditions: it exists independently of your product, with a problem it already lives with; it is precise enough to be told as a story; and you know which stage of the funnel it sits in, which decides the argument to make to it. The audience of each creative gets smaller; you cover the market by multiplying personas, with a better conversion rate on each one.

Why the idea is appealing

The two reflexes that make teams neglect personas are understandable.

The first comes from the culture of paid media. The job is to invest one dollar and get back four, five or six. In that logic, working on personas looks like theory, an exercise for large marketing departments that can afford research. When you run campaigns day to day, you have more urgent things to do.

The second comes from experience. The vast majority of marketers feel they know their personas by heart, and their short description seems enough to them.

I have worked on this before: before my current company, I founded a software company that used data to define the personas of e-commerce brands. What I took from it is that both reflexes are very common. They don’t hold up against the way Meta delivers ads today.

Reason 1: on Meta, the creative does the targeting

“Creative is the new targeting”: the phrase has been going around the Meta ecosystem for several years. It describes a concrete change. With Advantage+, Meta automates budget allocation, audience targeting and bidding; the delivery system decides on its own who sees each ad, based on what the ad contains.

This mechanism relies in particular on Andromeda, which Meta presented in December 2024 on its engineering blog. It is the first stage of its ad recommendation system: for each person, it keeps a few thousand relevant ads out of tens of millions of candidates. Meta says it is designed to take advantage of the growing number of creatives in circulation. What your creative shows therefore largely decides who it is shown to.

In practice, the chain runs like this:

1. 1you define a persona;
2. 2you come up with ad concepts that speak to it;
3. 3you brief the production of the creatives;
4. 4Meta receives those creatives and shows them to the people they seem to match;
5. 5those people land on your site.

What it changes for you: if you want to reach a very particular target, that target has to appear in your creatives. The persona becomes the raw material of targeting. So it has a direct effect on your campaigns, starting with the ads you are preparing for next week.

Reason 2: a five-word persona describes your product

Here are three real answers, from brands I know well.

1. 1A lingerie brand, about €12 million in revenue: “women aged 25 to 60 who want more comfort.”
2. 2A large mattress brand, €60 million in revenue: well-off urban couples living downtown.
3. 3The founder of a software product for paid media teams, in other words me two or three years ago: “the media buyer, at an agency, freelance or in-house.”

Look at them closely. Each one describes the company more than the customer. The lingerie brand is talking about its comfort, meaning its cotton and how it is made. The mattress brand is talking about its price positioning: an expensive product, made in France, so a customer base assumed to be well-off. And I was talking about the features of my tool, because I was absorbed in building it.

I call these product-centered personas: a feature of the offer, turned around into a customer description. They have a fundamental flaw. They say nothing about what the person goes through before meeting you, so nothing that could stop them in their feed.

Reason 3: a broad target brings in visitors who cost money without buying

The objection comes naturally: “If the algorithm brings me a broad audience, my ideal customer is bound to be in it.” That is true. The problem comes from everyone else, who costs you money too.

Let’s go back to the mechanism. Give Meta’s delivery system creatives that speak to “all women who love comfort,” and it will show them to a lot of people, since the description fits almost everyone. Every person who lands on your site has cost you money; that is what cost per click measures, and it quickly reaches several dollars. If a large share of those visitors doesn’t match your ideal customer, you are paying for a large share of your traffic for nothing.

This trap is especially dangerous for accounts with budget. With $5,000, $20,000 or $100,000 a month to invest, a precise persona feels too narrow, and you want to speak to your whole market in every ad. Across the mass, a generic ad will find a few buyers; most people, though, don’t have time to wonder whether a “comfortable” product is for them.

What to do instead: three criteria

A stand-alone persona

A good persona lives independently of your solution. It has a problem, a pain, what marketers call a “pain point,” before it even knows your brand. Here are the three earlier examples, rewritten.

1. 1Lingerie. A woman aged 30 to 40 who, when she gets home in the evening, takes off her bra first and puts on her pajamas, because her underwear leaves marks and has bothered her all day.
2. 2Mattress. A man over 35 who wakes up every morning with his back in bad shape. He just had a child and struggles to carry his 22-pound daughter. At the office, he can’t stay seated for thirty minutes anymore; he has bought a walking pad and a standing desk.
3. 3Paid media software. A media buyer who finds himself in a meeting with his client, with a dashboard that shows wrong numbers and takes ten minutes to load. Ten minutes face to face, with nothing solid to show, is a long time.

Each one describes a situation the person will recognize immediately, without mentioning the product.

This criterion also works as a market test. If your personas only discover their problem when they come across your solution, it is a sign that the market doesn’t really exist yet.

A specific persona

The second criterion follows from the first: a good persona is precise enough to be told as a story. The ad that speaks to the woman who puts on her pajamas as soon as she gets home gets her attention. The one that announces “comfortable clothes for all women” blends into the feed.

A persona placed in the funnel

The third criterion is the one people think of least, and it matters as much as the first two: a good persona sits at a precise point in the awareness funnel. At the very top are the people who aren’t even aware they have a problem (unaware); at the very bottom, those who are going to buy, today or tomorrow (most aware). In between, three stages: awareness of the problem (problem aware), awareness that solutions exist (solution aware), then knowledge of your product (product aware). This scale follows the levels of awareness described by copywriter Eugene Schwartz in Breakthrough Advertising (1966).

The whole funnel makes up your target market. The stage your persona is at completely changes how you talk to it. To show this, I walk through each stage with the same persona: a media buyer, the typical customer of the software I am building today.

1. 1He doesn’t know he has a problem (unaware). This media buyer blocks four hours every Monday morning for reporting on his four clients, one hour per client. He copies and pastes, takes screenshots of Shopify, Meta and Google Ads, hands them to Claude, and puts it all in a Google Sheet. By noon, everything is ready, and he feels efficient. He is not aware he has a problem. To reach him, I use a contrarian truth, a statement that shakes what he believes: “Your client judges you less on your reporting than on the quality and precision of your answer when they ask why their cost per acquisition is going up.” The goal is to wake something up, so that he becomes aware of the problem.
2. 2He knows his problem (problem aware). The same media buyer has dropped his Monday morning slot to fit in back-to-back client meetings. He now does his reporting on Sunday evening, in front of the TV, with a plate of pasta next to him. These are low-value tasks, the kind you can do without paying attention, but he can see this pace won’t last. Here, I agitate the problem: “You’d like to earn more, but every new client means one more hour on Sunday evening. Isn’t it time to find a solution?”
3. 3He knows solutions exist (solution aware). This media buyer has hooked Supermetrics up to Looker Studio so that his data updates on its own. He has become a data analyst without choosing to, and spends a good part of his time on maintenance. He has Claude or Gemini write his SQL queries, which don’t always give him what he wants; he sometimes mixes up figures before and after tax. In one meeting out of two, his dashboard takes ten minutes to load in front of the client. Yet he believes he has found the best compromise. For him, I build a comparison table: on one side, two hours of maintenance a week and embarrassing errors in meetings; on the other, an automated solution, maintained by a team of experts. It costs a little more. What, against that, are his two hours a week worth, his boredom, and the client who leaves, tired of reports they can’t get anything out of?
4. 4He knows your product (product aware). He knows automated solutions exist, but he hesitates: it is one more subscription, and most of his clients are used to Looker Studio, “the Excel of data,” where you can present the numbers however you like. Two levers are available. Social proof: the number of users already on board. And a concrete fear: “Picture a pitch. You show up with a makeshift dashboard; across the table, a competitor presents clear dashboards that load fast and filter in one click. Who do you think wins?”
5. 5He is ready (most aware). He knows the solution is good: it will help him win pitches and save time, improve his client relationships, and leave him more room for analysis than for producing numbers. There is nothing left to prove. The ad explains how it will go: three steps, from connecting the account to the first report, and the price. It is a very bottom-of-funnel ad, focused on how it works: the “no-brainer.”

The starting profile is the same, and yet the five ads have almost nothing in common: each stage of the funnel has its own persona, with its own arguments. Knowing which stage your persona is at therefore decides the argument, the tone and the format of every ad.

Covering the market: more personas, each one narrower

One caveat is in order: a specific persona reaches fewer people than the broad target it comes from.

The answer is to develop as many as it takes to cover your target market. The young father’s aching back is only one of the reasons to buy a mattress; there are others, each one carried by a different person and a different situation. Each persona feeds its own creative concepts, and the funnel multiplies them further: the same profile, at five stages of awareness, calls for five different ads. Each creative then carries a precise situation, and each ad attracts a smaller audience that converts much better.

To be precise: this method takes more work upfront and more creatives to produce. That is consistent with the way Meta designed Andromeda to take advantage of a growing number of creatives. The effort moves from tuning campaigns to understanding customers.

The questions to ask yourself

1. 1Can your personas be described without mentioning your product, your price or your features?
2. 2Does each of them have a concrete problem they already live with, in a situation your creatives could stage?
3. 3For each creative, do you know which stage of the funnel the person it speaks to is at?

If you answer no to any of them, start there, before you touch your campaigns.

On Meta, what your creatives show decides who comes to your site.

Sources: Meta Engineering, “Meta Andromeda: Supercharging Advantage+ automation with the next-gen personalized ads retrieval engine,” December 2, 2024; Eugene Schwartz, Breakthrough Advertising, 1966.