ChatGPT's Ad System Update Hides an Ambition

A while ago, I was chatting with a media-buyer friend of mine.

He was venting: there are too many new channels. Each one claims huge traffic and great conversion rates. The moment you actually try it — nothing. Budget burned, and all you got back was a pile of clicks. The real orders? Nowhere to be found.

I asked him: what do you want most?

He didn't even have to think: something I can optimize. Clicks that actually drive sales, budget that paces itself, data that tells me which dollar was worth spending.

I stayed quiet. Because I'd just seen a piece of news that, in one stroke, filled in every single thing he was asking for.

A New Player Starts Catching Up

ChatGPT — yes, that same chat box you use to draft copy and look things up — is getting into advertising.

And not in a small, experimental way. It just rolled out a batch of updates in one go, all pointed in a single direction: making the people who run ads feel that this is no longer a "let's give it a try" toy, but a proper workbench where you can actually settle the books.

Let me start with the biggest move.

oCPC: From "Pay Per Click" to "Pay Per Conversion"

What is oCPC?

Simply put: conversion-optimized cost-per-click billing (oCPC).

Before, when you ran ads on ChatGPT, you were charged per click. Someone clicks, you pay. Whether that person, after clicking, actually buys anything, leaves a phone number, or places an order — the platform didn't care, and you couldn't either.

It's like going to a farmer's market where the stall owner charges you a stall fee based on how many people walk past. Lots of foot traffic? You pay more. But of all those people walking by, how many actually buy vegetables? No idea.

Now it's different. You pick a "conversion" objective, and the system judges for itself: which clicks are most likely to turn into a real sale. It puts its weight behind those clicks.

You still pay per click. But the quality of the click has changed.

This is what advertisers actually want. Not cheap. Worth it.

Budget: It Spends Itself Evenly Now

The second change is about how the money gets spent.

In the past, you'd set a daily budget and the system might burn through it in the first half of the day. By the afternoon, when users are most active, you had no quota left. Sound stressful?

The new rule is: a 7-day rolling average daily budget. Spend too much today, it gets made up tomorrow; spend too little today, you get a bit more tomorrow. As long as you stay within the total budget envelope, you're fine.

And the system automatically gives you paced delivery — within a single day, instead of letting you go all-in on one time slot, it spreads the money more evenly across the day.

In plain English: it stops you from waking up at 3 a.m. to check the dashboard.

Data: You Can Finally See It

Anyone who runs ads knows the saying: no data, no optimization.

You spent the money — you need to know what that money brought back. Otherwise you're just gambling.

This update does two things on the question of "is the measurement actually accurate":

First, it integrated AppsFlyer and Adjust. These two are veterans of app data attribution. What that means: your app installs and in-app events can now be measured.

Second, auto advanced matching. Information users leave on web pages — email, phone number, that kind of thing — the system hashes it and uses it to more precisely reconcile whether a given person was actually brought in by this ad.

Before, when you ran ads on ChatGPT, performance was a black box.

Now, the black box is starting to crack.

The Features People Kept Asking For

There's another batch of updates — the "advertisers yelled for ages and finally got it" kind.

Geo-exclusion. Before, you could only choose "who to target," not "who not to target." Now you can — a certain city, a certain province: if you say don't run there, it won't run there.

Bulk operation API. Before, if you wanted to build dozens of ad sets, you had to click through them one by one. Now you can batch-create and batch-edit through an API. A godsend for the heavy spenders.

Product card upgrade. Product cards with prices, with star ratings — they've started showing up in the feed.

None of these, taken on its own, is a big deal. But put them together and you notice a common thread.

What Is This Update Really Saying?

Did you notice that this wave of updates didn't ship any flashy new ad placements?

No full-screen immersive ads, no AR interactive ads. Just very "basic" stuff: how to optimize conversion, how to attribute data, how to manage budget, how to do bulk operations.

Why?

Because what ChatGPT wants right now is not to attract more small advertisers to come play. It wants that crowd of big clients who make a living off media buying — the teams that drop hundreds of millions on other platforms every year and need automation, third-party measurement, and tools to manage things at scale.

This crowd doesn't pay up for a "novel" ad placement. They only recognize one thing: can this system, like the two I already know, let me spend money efficiently, see the books clearly, and scale delivery steadily.

So this update, translated into one sentence:

ChatGPT is catching up on the basics. The few foundational classes it was still missing compared with mature platforms.

What Does This Mean?

My personal read: this matters more than it looks on the surface.

It's not that oCPC itself is some revolutionary feature — it's been validated long ago, the ad industry has been running on it for years. The key is who's doing it.

ChatGPT is one of the fastest-growing traffic entry points right now. Every day it sees that many conversations, that many questions, that many people in there looking for answers and making decisions. This traffic used to be an "ad blind spot" — advertisers had their eyes on it but couldn't reach it.

Now, it's within reach. And the way you reach it is one advertisers already know — same logic, same metrics, same workflow.

New traffic, caught with the old playbook. That's where the real weight of this thing lies.

Of course, this is just the start. Once the basics are in place, the real show begins — when, in its own conversational way, it grows something nobody else has.

That thing — nobody has seen it yet.

I don't know what it'll look like either. But I'll be watching.

Because every time a new channel pivots from "catch-up" to "original," a few people emerge who capture the whole dividend of being early. I hope this time, you're in the game.