Running ChatGPT Ads and Google AI Mode Ads at the Same Time
By Ansh Khandelwal 9 min read
In brief
Run ChatGPT Ads and Google AI Mode Ads together without double-counting conversions by aligning UTMs, windows, CRM rules, and test budgets.
Last verified: 12 September 2026 | Version: 1.0 | Next scheduled review: 12 October 2026
Running both is operationally straightforward and analytically messy. The split is easy, the creative mostly reuses, and the problem is measurement: two pixels, two attribution models, two reporting windows and one conversion that both will count.
This is the operating guide. The head-to-head comparison lives in our ChatGPT Ads versus Google AI Mode Ads piece and is not repeated.
The structural asymmetry to understand first
These are not two versions of the same buy.
Google AI Mode serves Conversational Discovery ads inside the existing Google Ads auction. Practically that means it is an inventory extension of an account you already run, reconciled by a finance team that already reconciles it, measured by conversion tracking you already trust, inside a bidding strategy that already exists.
ChatGPT Ads is a separate platform with its own account, its own verification requirements, its own pixel, its own postpay invoice and its own reporting surface. Standing it up is a build. Turning on AI Mode inventory is closer to a setting.
Almost every operational difference below follows from that one asymmetry.
How to split budget between them
The honest starting position is that there is no defensible ratio, because no cross-platform benchmark exists to derive one from. What there is instead is a set of constraints that narrow the range.
| Constraint | Effect on the split |
|---|---|
| ChatGPT minimum daily budget is 25 USD with no lifetime commitment | Sets a floor, not a target. Below roughly 40 to 50 USD a day the data is too thin to read within a quarter |
| Google AI Mode spend is drawn from existing search campaigns | Often not separately fundable at all. The money is already in the account |
| ChatGPT serves only to Free and Go plan users, in reported markets US, UK, Canada, Australia, New Zealand, Japan, South Korea | Caps addressable reach in a way AI Mode is not capped |
| ChatGPT reporting stops at ad group level with no per-hint view | Argues for fewer, larger ad groups and therefore a concentrated budget |
| Category restrictions: financial, health and legal services are US-only and case-by-case as of 10 September 2026 | May remove ChatGPT from the split entirely for some advertisers |
A workable shape for a first quarter: hold Google AI Mode inside the search budget it already sits in, and fund ChatGPT separately as a bounded test on an experimental or brand line. Do not create a single "AI ads" budget pool split by percentage. It invites exactly the comparison that will kill the smaller of the two.
The measurement conflict, concretely
This is where the trouble is. Two systems, both claiming credit, neither aware of the other.
Two pixels, two definitions. ChatGPT Ads measures through the OpenAI pixel plus a Conversions API. Google measures through its own conversion tracking. If both are firing on the same thank-you page, both will record the conversion when either can claim the visit within its window.
Two attribution windows. ChatGPT Ads shipped configurable click-through attribution windows in 2026. Google Ads has its own, set separately, often at a different length. Two different windows on the same journey produce two different answers about the same conversion, and neither is wrong inside its own system.
No view-through parity. ChatGPT Ads reports impressions, clicks, spend, CTR, average CPC, average CPM and conversions. There is no adjacency reporting and no conversation data. You cannot reconstruct the pathway, only the endpoint.
No shared identity layer. There is no deterministic way to tell that the person who saw a ChatGPT ad on Tuesday is the person who clicked a Google AI Mode ad on Thursday. Custom audiences on ChatGPT Ads are first-party upload only, which is a targeting input, not a cross-platform identity resolution.
What double counting looks like in practice
The symptom is arithmetic that does not close.
Add the conversions in the ChatGPT Ads interface to the conversions in the Google Ads interface. Compare that total to the conversions your CRM recorded for the same period with the same definition. If platform total exceeds CRM total, you are double counting, and the gap is roughly the size of the overlap.
A worked illustration, invented for shape and carrying no account data. ChatGPT Ads reports 18 conversions. Google Ads reports 47. CRM records 52 new inbound opportunities from paid sources in the period. Platform total is 65, CRM total is 52, so around 13 conversions are claimed twice. On a quarterly report presented as 65, you have overstated by 25 percent and every downstream cost per acquisition figure is wrong in the same direction.
Three fixes, in order of how much work they are.
Make the CRM the reporting source of record. Platform numbers are for in-platform optimisation only. Nothing leaves the marketing team as a result without passing through the CRM count first. This is the only fix that actually resolves the conflict rather than reducing it.
Enforce one UTM convention across both. Same parameter names, same casing, same source values, distinct enough that a CRM report can attribute a record to exactly one platform on first touch. This is what makes the first fix possible.
Align the attribution windows. Set the ChatGPT Ads click-through window to the same length as your Google Ads window. It will not eliminate overlap, but it removes one of the two reasons the numbers disagree, so what remains is genuine overlap rather than a windowing artefact.
Creative reuse: what transfers and what does not
More transfers than people expect, with two hard stops.
The proposition, the offer, the disqualifying condition and the landing page all reuse directly. A price floor or a headcount band works identically on both surfaces, and both surfaces sit below or beside an answer the user already has, which means both reward specificity over appeal.
What does not transfer is the asset shape. ChatGPT Ads takes one headline, one description, one image that is PNG or JPG, square, at least 256 by 256, one destination URL, plus brand name and logo. Character limits are genuinely contested across published sources, from 16 and 32 characters up to 50 and 100, and truncation is documented. A responsive search ad with fifteen headlines has no equivalent, and you cannot rely on the platform assembling combinations for you.
The second hard stop is crawler access. OAI-AdsBot must be allowed to reach your landing page or ads can be rejected. If the landing page you are reusing from Google sits behind a bot filter, it works on one platform and fails review on the other.
What we cannot tell you
- What the right budget split is. No cross-platform benchmark exists. Neither OpenAI nor Google publishes comparable cost or conversion data, and no third party has published a controlled parallel test.
- How much genuine overlap there is between the two audiences. Not measurable. There is no shared identity signal and ChatGPT Ads provides no adjacency or conversation data.
- Whether one surface assists the other. No incrementality study exists on either platform for this pairing.
- Whether Google AI Mode spend can be isolated from search spend in reporting. Varies by account and rollout, and we have no first-party account to verify it in. InPromptAds runs no client campaigns.
Quick answers
Can I run ChatGPT Ads and Google AI Mode ads at the same time? Yes. They are separate systems with no shared auction, so there is no platform conflict. The difficulty is measurement, not delivery.
How should I split the budget? Do not create a shared pool split by percentage. Keep Google AI Mode inside the search budget it already draws from, and fund ChatGPT separately as a bounded test with the 25 USD daily minimum as a floor rather than a target.
Will both platforms count the same conversion? Yes, whenever both can claim the visit inside their own attribution window. Add both platform totals and compare to your CRM count for the same period. The gap is the overlap.
How do I stop double counting? Make the CRM the source of record for anything reported outside the marketing team, enforce one UTM convention across both platforms, and set the ChatGPT click-through attribution window to match your Google window.
Can I reuse my Google creative? The proposition and landing page transfer. The asset shape does not: ChatGPT Ads takes one headline, one description and one square image of at least 256 by 256, with contested character limits and documented truncation.
Is there anything that breaks a reused landing page? Yes. OAI-AdsBot must be allowed to crawl it. A page behind a bot filter can pass on Google and be rejected on ChatGPT.
Sources
| Claim | Source | Tier |
|---|---|---|
| Measurement is via the OpenAI pixel and Conversions API; reporting covers impressions, clicks, spend, CTR, avg CPC, avg CPM, conversions at campaign, ad group and ad level | OpenAI Ads Manager documentation, 2026 | Confirmed, primary |
| Configurable click-through attribution windows shipped in 2026 | OpenAI Ads Manager documentation, 2026 | Confirmed, primary |
| No adjacency reporting, no conversation data, no cross-advertiser benchmarks, no per-hint reporting | OpenAI documentation, by absence | Absent |
| Custom audiences are first-party upload only | OpenAI Ads Manager documentation, 2026 | Confirmed, primary |
| Creative is brand name, logo, one headline, one description, one square PNG or JPG at least 256x256, one destination URL | OpenAI Ads Manager documentation, 2026 | Confirmed, primary |
| Character limits conflict across five published sources, from 16/32 to 50/100; truncation is documented | Multiple published sources, 2026 | Reported |
| OAI-AdsBot must be allowed or ads can be rejected | OpenAI Help Center, advertiser crawler guidance, 2026 | Confirmed, primary |
| Google AI Mode serves Conversational Discovery ads inside the existing Google Ads auction | Google Ads announcements and trade press, 2026 | Reported |
| Minimum daily budget 25 USD, no lifetime commitment | OpenAI Ads Manager documentation, 2026 | Confirmed, primary |
| Financial, health and legal services US-only, case by case | OpenAI Ad Policies, updated 10 September 2026, v1.6 | Confirmed, primary |
| The CRM should be the reporting source of record when both platforms run | InPromptAds | Inference, ours |
Related reading
- ChatGPT Ads vs Google AI Mode Ads
- UTM Conventions for ChatGPT Ads
- Setting Up the OpenAI Pixel
- The ChatGPT Ads Conversions API
- Where the ChatGPT Ads Budget Should Come From
Changelog
12 September 2026, v1.0. First publication. Sets out the build-versus-setting asymmetry, the four sources of measurement conflict, and the CRM-reconciliation test for double counting.
Field kit
Tools
Site
OpenAI (primary)
Ansh works across GEO strategy, B2B research, and execution. At InPromptAds, he translates new AI advertising products into clear operating advice, tests, and measurement questions for marketing teams.
