ChatGPT Ads: conversion bidding, geo exclusions, and bulk campaigns — the concrete impact
Summarize this article with AI
What I’m seeing across 15 ad accounts every week
I look at 15 ChatGPT Ads accounts per week. Fifteen e-commerce accounts, different sizes, different sectors. All the same core problems.
Bids too broad. Location targeting on autopilot. Campaigns launched one at a time like postage stamps.
Result: a budget leaking into ghost clicks. Regions that don’t convert. And no ability to scale without burning out.
One client, an online clothing shop — €400,000 annual revenue, 300 SKUs — came to me with this mix. Their account: 12 campaigns, all manual, zero reliable conversion tracking. In 3 months, €11,200 spent for 28 sales. That’s a CPA of €400. On a gross margin of €18 per item.
A disaster.
But not an inevitable one.
The three features ChatGPT Ads just rolled out — conversion bidding, geographic exclusions, bulk campaigns — fix these gaps. When you activate them at the right time.
Conversion bidding: when AI bets on the sale, not the click
Until now, ChatGPT Ads campaigns ran on cost-per-click bidding. The algorithm tried to maximize visits. Logical. Except visits don’t pay suppliers.
Conversion bidding changes that. The AI now optimizes for real conversions: recorded sale, cart addition, qualified lead. It shows ads to the right people, at the right moment, to trigger a purchase.
According to Search Engine Land, ChatGPT Ads now supports granular conversion tracking: you define the pivot event (transaction, « purchase complete » option, etc.), set a target CPA, and the algorithm adjusts bids continuously to hit it.
On my textile client’s account, we moved 8 campaigns to conversion bidding with a €65 target CPA — five times lower than their actual previous CPA. We connected the conversion pixel to the order confirmation page. Clean data flow, no dispersion.
Results after 6 weeks:
- Average CPA dropped to €87 (still above target, but 2.3× lower than the €400).
- Conversion volume tripled.
- Conversion rate improved: the AI automatically stops bidding on segments that don’t convert.
We’re not at target CPA yet. But the curve is plunging month after month.
Conversion bidding doesn’t work miracles. It does something more useful: it stops paying for clicks going nowhere.
On one e-commerce account, geographic exclusions alone slashed the cost per acquisition by more than half. Here’s the impact on a real budget.
Geo exclusions: CPA cut by 2.3×
From €400 to €174 per sale after excluding unprofitable zones
Geographic exclusions: money that was pouring away stays put
Let’s talk cash. France isn’t a monolithic block. A physical shop in Lille doesn’t ship to Guadeloupe under the same conditions. A beach equipment seller moves tons of units in July in Corsica, zero in February in the Cantal.
Without geographic exclusions, your ads display everywhere in the country, sometimes beyond. Your budget waters zones that will never convert, or worse, that convert at a loss.
ChatGPT Ads now integrates fine-grained geographic exclusions. Postal code, department, region. You no longer have to hope the algorithm « understands » your logistics constraints or your zone profitability analysis. You encode them. Directly. No more silent mortgages.
I saw a glaring case on a home decor account — 2,500 orders per year, €85 average cart. Their shipping cost to overseas territories was more than double the product price for some heavy items. Yet ads were running at full throttle in those regions. Result: loss-making sales, hidden in overall revenue.
We excluded 3 overseas regions and 2 mainland departments where average cart was 34% below the national median — historically unprofitable. In one click, ads stopped running there.
Results after a full month: total order volume down 14%. But net margin up 73%. Yes, you read that right. Fewer orders, more profit. Because loss-making sales stopped dragging down the result.
Geographic exclusions don’t limit your reach. They reallocate your budget to zones where every euro generates a profitable sale.
Bulk campaigns: the art of scaling without losing your advertising soul
I still see too many e-commerce teams creating every campaign manually. One per flagship SKU. Another for sales. A third for brand awareness traffic. Type the params, bids, audiences, creatives. Repeat. 4 hours for ten campaigns. Two days for back-to-school season.
This doesn’t scale.
ChatGPT Ads’ bulk campaign tool solves this bottleneck. You prepare a structured file. It contains campaign names, ad groups, keywords, initial bids, target audiences, matching creatives. You import it. The platform creates, verifies, activates. In batches.
On the textile account I mentioned earlier, we built a file with 47 campaigns targeting 8 product catégories, with 3 bid types (conversion, click, awareness) and 2 creative universes. Prep time: 1 hour of setup. Platform deployment time: 45 minutes.
Before June 2026, you’d budget 5 hours, with typing errors on every line.
What changes the game:
You no longer waste time duplicating. You focus your energy on structure, exclusions, optimization. The industrialization that matters.
ChatGPT Ads also handles bulk edits. Want to raise all bids by 15% on a product family? Edit one column, re-import. Not 47 individual changes.
Direct consequence: faster test cycles. You can launch 20 campaigns Monday, analyze data Wednesday, pivot Friday. Time is no longer the constraint.
The counterintuitive move: fine automation kills blind autopilot
Blind automation is the trap. Tooled automation is the force.
I see a trend among those adopting these tools: relief. Finally, no need to think about location targeting. Let the AI decide bids. Let the system create campaigns.
Bad idea.
Conversion bidding without a properly tuned target CPA generates unprofitable volume. Geographic exclusions without regular review become obsolete in 3 months (profitable zones shift). Bulk campaigns without strict naming become an unauditable mess.
ChatGPT Ads gives you a framework to exercise your judgment faster and wider. Not a replacement for your judgment.
On another account, a puericulture specialist, we almost broke everything: we activated conversion bidding with a CPA target too high, targeting « add to cart » as the main conversion event. The AI filled carts with €23 bottle warmers from low-cost segments. The cart-to-order conversion rate tanked. We went from 4.8% checkout validation to 1.9% in fifteen days.
We corrected course. Conversion event: « order completed. » Target CPA cut 28% versus previous average. Result: in three weeks, back to 4.4% validation, with order volume up 12%.
Fine automation demands more rigor, not less. That’s what separates a tool from a gimmick.
Here’s how to concretely deploy conversion bidding, geo exclusions, and bulk campaigns in sequence — no step skipped.
The 4-step deployment plan
From conversion tracking to bulk scaling in four moves
How to deploy this on your store — in 4 steps
Here’s how I guide an account through these three levers in practice. Don’t skip steps.
1. Certified conversion tracking
Place the ChatGPT Ads pixel on your order confirmation page. Verify the fired event is purchase with cart value. Without this, no credible conversion bidding.
2. Geographic profitability analysis
Export your orders from the last 12 months. Cross-reference region / average cart / margin after shipping. Identify zones with negative profitability or <5% net margin. Exclude them. Immediate first gain.
3. First conversion bidding tier
Set a target CPA equal to 80% of your historical CPA (for campaigns that perform). Activate on 20% of your most promising campaigns. Watch for 2 weeks. Iterate.
4. Bulk campaign template
Build a file template with required columns: name, group, keyword, bid, audience, destination URL. Pre-fill with 15 sample campaigns. Import on a test account if possible. Then deploy to production.
And from there, you trigger a positive cycle: cleaner conversion data, leading to better conversion bids, more profitable campaigns, and therefore more budget to create more bulk campaigns.
What endures once the noise fades
In 3 months, everyone will have activated conversion bidding. We’ll forget it was a feature launch. But accounts that built a strategy around the tool won’t run at the same level.
Those that identify their profitable zones — then exclude the rest — will see margin climb. Those that can industrialize without giving in to the « fully automatic » temptation will hold pace.
The rest will simply switch tools to repeat the same losses, faster.
I’m not selling you the method. I’m showing you the evidence.
And you? How many campaigns are you still launching by hand?
A live ad audit. 45 minutes. On your account.
We review your bids, geo exclusions, campaign structure. I show you what the tools don’t say.
Book a strategic call — 45 minFrequently Asked Questions
How do I set up conversion bidding on my ChatGPT Ads account?
Place the pixel on your order confirmation page. Declare the « purchase » event. In campaign settings, select the « Conversions » objective. Set a target CPA based on your history.
Can I combine geographic exclusions with conversion bidding?
Yes, I recommend it. Exclusions neutralize zones that deliver nothing before cost-per-conversion bidding starts. The algorithm works on already-cleaned ground.
Are bulk campaigns compatible with existing campaigns?
Import a file to create or update multiple campaigns at once. Changes apply to editable fields: bids, budget, status.
What target CPA should I choose to start with conversion bidding?
80% of your historical CPA over the last 30 days for campaigns that perform. Adjust weekly. Too low, you lose volume; too high, you degrade profitability.
Do geographic exclusions affect delivery in neighboring territories?
No. Geographic exclusions apply strictly: only the postal code or department you choose is excluded. Neighboring zones, even close ones, aren’t auto-deduped. Monitor your performance by zone with monthly analysis.
