Field notes · Manu
Your edge isn't campaign management anymore
I know that's uncomfortable. I built a career on it. But the part we charged for is the part the platform just took.
The biggest mistake I see media buyers make right now is thinking their edge is still in the campaign build. The tight structure. The bid tweaks. The Thursday budget shift.
It isn't. Not anymore.
The platform ate that part. Smart bidding. Advantage+. Auto-everything. Meta and Google are getting closer to the buying decision itself, and they're doing the mechanical work you used to charge for.
So where did the edge go?
So where did the edge go?
The edge went to judgment. Three jobs the platform did not take:
- Reading what actually happened, across two platforms that never agree with each other.
- Holding the client’s context, which no dashboard stores and no model is given.
- Deciding what to do about it, and being the one who answers for the decision.
Judgment is the part a platform can’t take, because it’s the part that was never mechanical. And it is where being right more often than the algorithm still pays.
Eight years and €1.3M+ in managed ad spend taught me the same lesson twice: the buyers who last aren't the best at building campaigns. They're the best at reading them and making the call.
Reading is the one job nobody built for
Which is a problem. Because reading is the one job with no owner. Launching has an owner. Reporting has a deadline. Looking back to check whether last month's call was actually right? Nobody's Monday has room for it. So we skip it, and we decide from memory - which is a worse record than we think.
The missing habit of looking back is the actual gap. Not campaign skill - we have plenty. The gap is a cheap, honest way to look back, so the judgment has something real to stand on.
That gap is the whole reason I started building a console that reads my Meta and Google data and never changes anything I haven't approved. Not to replace the judgment. To feed it. The human still makes the call - the tool just makes sure the call isn't based on a Thursday you half-remember.
Your edge is judgment now. Protect the inputs to it.
Questions people actually ask
Are you saying campaign skill does not matter?
No. It says it stopped being the scarce part. You still have to build the thing properly. You just cannot charge for that alone any more: the platform does most of it, and never sends an invoice for it.
The platforms are not that good yet.
In places they are not, and those places are exactly where your judgement earns its money. The argument is not that the machine is better than you. It is that the mechanical half is the half it took.
So what do I do differently on Monday?
Protect the inputs to the judgement. In practice that is one honest look back per account per week, written down, so the next call stands on a record instead of a Thursday you half-remember.
Is this a pitch for your console?
Partly, and it would be dishonest to pretend otherwise - I built one because I wanted it and could not buy it. The argument holds without it: a notebook and a repeated hour does the same job right up until the number of accounts makes it impossible.
Related reading. Autonomous vs approved AI ad tools. How to connect your AI to your Meta & Google ad data (2026). You're deciding from memory.
— Manu
Eight years buying media, still doing it. If the "looking back" is the part you keep skipping too, that's what I'm building Climent Ads Assistant for. Open-source, and it never touches a campaign you haven't signed off. Runnable demo now.
July 2026