Why clients actually fire agencies (and it’s rarely the results)
Most clients don’t leave over one bad month. They leave because they feel like nobody’s paying attention. Here’s how to make sure the client is never the first to notice a problem.
I’ll say up front that this is a prediction. I could be wrong about the timing. I don’t think I’m wrong about the direction.
Look at what’s already happened. Google’s Performance Max decides where your ads show, across Search, YouTube, Display and Gmail. Meta’s Advantage+ campaigns handle audiences, placements and budget on their own. A few years ago, a good media buyer spent most of the day doing exactly those things by hand.
The platforms want this. The less you touch, the more they control, and they keep making the automated options better and the manual ones harder to find.
What’s left for the human
If the machine does the buying, what does the media buyer do?
Quite a lot, actually. It’s just different work.
Someone has to feed it good creative, because the AI can only pick from what you give it. Someone has to set it up properly: conversion tracking, the right goals, clean data going in. Bad inputs give you a confident machine optimizing for the wrong thing. Someone has to watch it, notice when it drifts, and catch it spending on junk placements at 2 a.m. And someone has to understand your strategy well enough to know when the AI’s “win” isn’t a win for the client.
That job needs an operator.
The part that excites me, and the part that doesn’t
Here’s what gets me. If the platforms do the hands-on work, one person who really owns the process could oversee far more accounts than anyone could build by hand. Potentially hundreds of clients, if the system is set up right and they work closely with your strategist.
For an agency, that changes the math on a media buying hire completely.
I’m also a little uneasy about it. Handing more control to Google and Meta means trusting systems that are built to spend your client’s money. They’re good, but their goals and your client’s goals aren’t always the same. The agencies that do well here will be the ones with someone sharp enough to push back on the machine.
What I’d do now
Put someone in that seat before the shift fully lands. Skip the person who wants to click around in Ads Manager all day. Look for someone curious about the AI tools, who’ll document the process, build the system out for you, and get better at it every month.
If you wait until everyone’s doing it, you’ll be hiring for it at the same time as everyone else.
I don’t know if this takes two years or five. Nobody does. I’d rather have the person in place early.
