There was a time when Meta's ad targeting was genuinely extraordinary.

If you wanted to reach BMW drivers who drink chai lattes with no water, listen to Slipknot, and enjoy romantic comedies on Friday nights, you could build that audience. Precisely. Cheaply. And it worked.

Then a privacy lawsuit happened. A significant one. I will not name it directly because the relevant parties have demonstrated a robust appetite for litigation and I am a small man on Vancouver Island with a newsletter and a coffee habit. You know the one.

The targeting system broke. Not partially. Substantially.

Campaigns that had been running beautifully for years started behaving like toddlers who had gotten into the Red Bull. Audiences that should have been tight were suddenly broad. CPAs climbed. Relevance scores dropped. Nobody could explain why.

I called our Meta rep.

I said: "I don't understand why the campaigns failed. I have the targeting set perfectly."

She said: "Unfortunately our targeting system doesn't work anymore. We are working on internal solutions."

I thought about that for a moment.

Then I said: "So you mean the targeting in the back of the Facebook ads manager is now as useful as the close button in a New York elevator?"

She said: "That is an excellent way of putting it. Yes. We have other ways to direct your audience now. Just give us a little more time."

This was early 2022.

Here is what "a little more time" has looked like in practice.

Meta moved aggressively toward Advantage+ campaigns, which are their answer to the broken targeting problem. The pitch is that their AI can find your audience better than you can by manually building one. Hand over the controls and trust the machine.

For some accounts this works well. For others it is a polite way of saying "we no longer know who your customer is and neither do you, so let's spend broadly and see what sticks."

The honest state of Meta targeting in 2026 is this: the granular interest targeting that made Facebook ads extraordinary between roughly 2014 and 2021 is largely gone and not coming back. What replaced it is a probabilistic system that works best when you give it three things.

Volume. The algorithm needs conversion data to learn. If you are running campaigns with fewer than fifty conversions per week, the machine is guessing. Your manual targeting instincts were probably better.

Creative. This is the part most advertisers are slow to accept. Creative is now the targeting. The right image or video attracts the right person more reliably than any interest category you can select. Your audience targeting box is increasingly decorative. Your creative brief is increasingly everything.

Patience. Meta's learning phase is real and it is longer than it used to be. Campaigns that look broken in week one often stabilize by week three. The instinct to intervene kills more campaigns than bad targeting ever did.

What the rep could not say out loud in 2022 is that the product had fundamentally changed and nobody was sure what it was anymore.

What she was allowed to say was: give us a little more time.

Three years later the system is better. The elevator button still does not close the door.

But if you give the machine good creative, enough budget to learn, and the discipline not to touch it every time the CPA twitches, Meta is still one of the most powerful acquisition channels available.

Just not for the reasons it used to be.

Rob

P.S. If you are still manually building detailed interest audiences in 2026 the way you did in 2019, you are optimizing a system that no longer exists. Start with broad targeting and let the creative do the work. I know it feels wrong. Do it anyway.