ChatGPT Ads vs Google AI Mode Ads: What Can and Cannot Be Compared
By Keshav Parsai 9 min read
In brief
Two AI ad surfaces after Perplexity exited ads. They differ on auction, exclusions, and reporting — and OpenAI says no benchmarks exist for fair comparison.
Last verified 2 October 2026. Version 1.2. Next scheduled review 2 November 2026.
The verdict, first. If your buyers are consumers in one of seven serving markets and you sell in an approved consumer category, ChatGPT ads reach an audience that is deliberately narrow and currently under-competed. If you need scale, existing measurement infrastructure, or any vertical outside OpenAI's approved list, Google AI Mode is the surface that will actually take your money and report on it.
Most comparisons of the two are written as three-way, with Perplexity as the third. That has been wrong since February 2026, when Perplexity abandoned advertising entirely after testing sponsored follow-ups through 2024 and 2025, cited user trust, and repositioned as the ad-free alternative. Any comparison still listing Perplexity as an AI ad platform was written from an older comparison rather than from the market.
The comparison table, with the empty cells left empty
The honest version of this table has holes in it. Filling them is what most published comparisons do, and it is where they stop being useful.
| ChatGPT Ads | Google AI Mode Ads | |
|---|---|---|
| Status | Live, expanding market by market | Live, integrated into existing Google Ads |
| Advertiser markets | 63 countries, per OpenAI's availability page | Google Ads global footprint |
| Audience excluded by design | Plus, Pro and any Business plan; predicted under-18s | None equivalent |
| Category restrictions | Primarily consumer categories; finance, health and legal services are US-only and case by case | Google Ads policies, materially broader |
| Buying interface | ads.openai.com, standalone | Existing Google Ads account and campaign types |
| Minimum spend | 25 USD per day per campaign in the US; published floor for 30 billing currencies | Google Ads minimums |
| Sub-national targeting | US at postal code level; elsewhere "where available", unenumerated | Full Google geo-targeting |
| Reporting latency | Impressions and clicks refresh about every 15 minutes; spend can lag 7 to 8 hours | Standard Google Ads reporting |
| Published performance benchmarks | None. OpenAI states none exist across advertisers, industries or campaign types | Extensive, though AI Mode specifically is newer than the benchmarks |
| Ad load in AI answers | Not published | Reportedly present in 25.5% of AI results, up from 5.17% in early 2025 |
| Attribution to CRM | Aggregated performance data only | Existing Google conversion infrastructure |
The comparison everyone publishes and nobody can support
Search for a comparison of these platforms and you will find CTR, CPC, conversion rate and ROAS tables setting ChatGPT against Google and often against Meta.
OpenAI's own advertiser FAQ states there are no performance benchmarks across advertisers, industries or campaign types. So a published ChatGPT CTR benchmark has one of three origins: a single advertiser's account presented as a category norm, a figure lifted from a different platform and relabelled, or a model built from organic behaviour and printed as though it were measured from the ad platform.
The third one is common enough to have a documented example. Digital Applied has publicly flagged that a widely circulated set of vertical conversion benchmarks are modelled estimates extrapolated from organic conversion rates rather than figures measured in an ad account. We are naming it because that correction is right and it is not ours.
There is one genuinely measured public figure in the category worth knowing about, and its limits are the point. Opascope published fifteen days of live-account data: roughly 60,000 USD of spend returning about 89,000 USD, a blended 1.49x ROAS, 1.72 USD CPC and a 2.35% conversion rate. That is one advertiser, in one category, over fifteen days. It is the most useful number published in this channel and it is not a benchmark, and Opascope did not present it as one.
If you need a comparative CPC table to justify a budget internally, the honest position is that the table cannot be built yet from either side, and a table you build anyway is a document your CFO will hold you to.
What actually differs, mechanically
Audience composition is the real difference, and it runs the opposite way to intuition.
Google AI Mode inherits Google's audience, which is everyone. ChatGPT's ad-eligible audience is defined by exclusion: no Plus, no Pro, no Business plan, no predicted under-18s. That is a consumer audience with the work-provisioned accounts removed, which is why the B2B case for ChatGPT ads is structurally weaker than the category's own content suggests.
Intent arrives differently. A Google query is a query. A ChatGPT conversation is a sequence, and the ad lands inside a reasoning process rather than beside a result list. That difference is real and it is also where this category's writing is at its most speculative, because nobody has published behavioural data comparing response inside a conversation against response beside a result.
Measurement maturity is not close. Google AI Mode ads run through infrastructure advertisers already have wired to their CRM. ChatGPT ads return aggregated performance data with up to a seven-hour lag between delivery and reporting. For a business that measures pipeline rather than clicks, that gap is larger than any CPC difference.
Geography is closer than it was, but not close. ChatGPT Ads went from nine advertiser markets in mid-August to 63 by the end of September, against Google's global footprint. Sub-national targeting is documented at postal-code level only in the United States; elsewhere OpenAI says states, cities, markets and postal codes are available "where available" without saying where. And new self-serve accounts may be held to their home country at first, which Google does not do.
When ChatGPT ads are the better buy
Narrow, and real.
You sell a consumer product in an approved category. Your buyers are in the United States, the United Kingdom, Canada, Australia, New Zealand, Japan or Korea. Your competitors have not started, which in this channel is still frequently true. You can accept aggregated reporting for a test period. And you want to be early in an auction that is not yet dense, which is the only advantage in this comparison that decays.
That last one is the honest case for testing now rather than in a year, and it is a case about competition rather than about performance.
When Google AI Mode is the better buy
Everything else. Scale, existing measurement, any restricted vertical, any market outside the 63, any campaign where you need sub-national targeting, and any organisation that cannot run a channel without conversion reporting into its own systems.
Not a hedge. For most advertisers reading this, it is the answer.
What we cannot tell you
Comparative CPC, CTR, conversion rate or ROAS. Not because we have not looked. Because OpenAI states the benchmarks do not exist and the published figures fail on inspection.
Whether conversational placement outperforms result-list placement at equivalent spend. No published test. This is the comparison the category most wants and it requires running both simultaneously against the same audience with the same creative, which nobody has published.
How Google AI Mode ad load will move. The 25.5% figure is reported and it moved sharply from 5.17% in roughly a year.
Whether we would recommend either. We do not run campaigns and we hold no account on either platform. Everything above is documentation and published data, and where it is our own measurement we have said so.
Quick answers
Are ChatGPT ads or Google ads better? For eligible advertisers in OpenAI's current self-service countries with an approved category, ChatGPT ads offer a newer conversational surface. For established scale, mature query controls or deeper historical benchmarks, Google remains the practical baseline.
What is the CTR of ChatGPT ads compared to Google? There is no published ChatGPT benchmark to compare. OpenAI states no performance benchmarks exist across advertisers, industries or campaign types, so any comparative CTR table is built from a single account, a relabelled figure, or a model.
Does Perplexity sell ads? No. Perplexity abandoned advertising in February 2026 after testing sponsored follow-ups through 2024 and 2025, citing user trust, and repositioned as an ad-free alternative. Comparisons listing Perplexity as an AI ad platform are out of date.
Can I run ChatGPT ads and Google AI Mode ads together? Yes, they are separate platforms with separate accounts. The measurement asymmetry is the practical constraint: Google reports into your existing conversion infrastructure and ChatGPT returns aggregated performance data with up to a seven-hour reporting lag.
How much of AI search results carry ads? Ads reportedly appear in 25.5% of Google AI results, up from 5.17% in early 2025. OpenAI has not published an equivalent ad load figure for ChatGPT.
Sources
| Claim | Source | Tier |
|---|---|---|
| No performance benchmarks across advertisers, industries or campaign types | OpenAI advertiser FAQ | Confirmed, primary |
| Ads not shown to Plus, Pro or Business plan users, or predicted under-18s | OpenAI Help Center, updated 30 July 2026 | Confirmed, primary |
| Primarily consumer categories; finance, health and legal services are US-only and case by case | OpenAI ad policies v1.6, 10 September 2026 | Confirmed, primary |
| Reporting lags delivery by up to 7 hours | OpenAI Help Center, Troubleshooting, 31 July 2026 | Confirmed, primary |
| 52 self-service advertiser countries | OpenAI Help Center, Ads Manager Availability, verified 13 Sep 2026 | Confirmed, primary |
| Targeting granularity: 41,488 US postal codes, 249 countries selectable | Our pull of the OpenAI geotargets CSV, 8 Aug 2026 | Confirmed, our measurement |
| 63 advertiser countries; location levels "where available" | OpenAI Help Center, Ads Manager Availability and Create Campaigns for ChatGPT Ads, 2 Oct 2026 | Confirmed, primary |
| Perplexity exited advertising February 2026 | Trade coverage, summarised by digitalapplied.com | Reported |
| Ads present in 25.5% of AI results, up from 5.17% in early 2025 | Trade coverage | Reported |
| 15-day live account: ~60,000 USD spend, ~89,000 USD return, 1.49x ROAS, 1.72 USD CPC, 2.35% conversion | opascope.com | Reported, single account |
| Circulating vertical benchmarks are modelled from organic conversion rates | digitalapplied.com | Reported, and consistent with OpenAI's own statement |
Related reading
- Who should not buy ChatGPT ads yet — the six disqualifiers, including the Business plan exclusion
- ChatGPT ads changelog — dated record of what has actually changed
- What has to be true before you can run a ChatGPT ad
Changelog
2 October 2026. v1.2 updated advertiser markets from nine to 63 and the sub-national row to OpenAI's "where available" documentation.
13 September 2026. v1.1 corrected legal-services eligibility and refreshed the comparison against OpenAI ad policies v1.6.
15 August 2026. v1.0 published.
Field kit
Tools
Site
OpenAI (primary)
Share this
Keshav studies how AI systems retrieve, verify, and cite brand information. At InPromptAds, he leads source research and turns platform documentation into practical guidance for advertisers.
