Testing ads in ChatGPT
so… you didn’t get an email from me last week, aand the reason is that I got married 🎉
so… you didn’t get an email from me last week
aand the reason is that I got married 🎉
ok.
back in the trenches now
and before I get into the new stuff - most of you are STILL over-complicating your ad accounts
more on that in a sec. first, two things that have been living rent free in my head all week 👇
openai ads manager

yep. OpenAI now has an ads manager
I’m already testing it with one of my clients. no results yet, way too early… I’ll report back the second I see real signal (good or bad)
honest take on the UI so far - it’s SIMPLE. like, almost too simple. you don’t get much control over it (for now)
reminds me of AppLovin in the early days. clunky, bare-bones, but you could feel something underneath it
so don’t expect a Meta-level dashboard yet.
but here’s the part that actually surprised me
while I was poking around, I found something the SEO crowd should probably already know…
when you do AEO, ChatGPT silently rewrites your query before it answers. it doesn’t show you. it just… does it
real example from me:
- I typed: “best dtc consultants”

- what ChatGPT actually searched: “best DTC consultants ecommerce growth consultants top names direct-to-consumer brands” WHAT!??!?!?!?!?

I never typed “ecommerce.” never typed “top names” either
it added a lot!!!
now think about what that does to your SEO/AEO spend…
you’re not optimizing for what people type anymore - you’re paying to rank for what the model decides they meant
so if you’re still buying keywords off what shows up in search bars, you’re optimizing for a query that’s already gone by the time the model answers
that one’s gonna sit with me for a while
now the boring one that still works
re-sharing this because a TON of you are still over-complicating your account setup 😂

I’ve said it a hundred times - the scaling logic is stupid simple:
- launch it
- wait at least 48h
- performing? scale it. I usually just double the budget
- not performing? analyze it
- in line with benchmark? leave it alone
- below benchmark? kill it, then share the learnings with the team
that’s it
let it cook 48h, then decide
Talk soon,
Kris
P.S. still owe you that loom on the google agent. didn’t get to record it past week, you’ll most likely get it in next week’s email 🚀
P.P.S. I redid my website. again. a bunch of you told me the old one was a wall of text / way too much reading (thank you for the honesty!), so I rebuilt it. go take a look and hit reply - tell me what’s working and what isn’t. I read every response 💡