- Is it worth moving budget from Google Search to ChatGPT Ads?
- What the data really says: revenue, conversions and cost per click
- How big is ChatGPT as an advertising platform compared to Google?
- What changes in practice: how do ads on ChatGPT work?
- Three real-world scenarios: what happens for different Italian companies?
- Why everyone's talking about it now: is the search market really changing?
- What tools and strategies exist to test ChatGPT Ads without wasting budget?
- What risks and limits should you keep in mind before investing?
- How to get started with ChatGPT Ads: the concrete steps
- What mistakes are most common when testing a new ad channel?
- How do I know if ChatGPT Ads is working for my business?
- So, is ChatGPT Ads worth it for your business?
- How we see it at SHM
- Frequently Asked Questions
- Data from a study of 102 sites shows that ChatGPT Ads generates about one third of the revenue per session of Google Search Ads, at a higher cost per click.
- The comparison isn't definitive: the platform is young, the data covers an early phase, and OpenAI has already launched new bidding tools (Target CPA and Target CPC).
- Google Search keeps growing (+13.4% in 2025) and remains the dominant channel, but ChatGPT has passed 1 billion weekly users: ignoring it is a strategic risk.
- The right choice isn't "one or the other": you need a controlled test with a separate budget, clear goals and comparison metrics defined before you start.
For every euro spent on ChatGPT Ads, an advertiser recovers an average of $0.52 in revenue. On Google Search, the same euro returns $3.40. Yet more than 100,000 advertisers have already tested OpenAI's platform: is it worth following them?
Is it worth moving budget from Google Search to ChatGPT Ads?
Based on the data so far, no for most companies. But a controlled test makes sense if you want to secure a spot on a fast-growing channel before costs rise.
ChatGPT Ads is OpenAI's advertising platform, launched in February 2026. It lets you show ads to users while they use ChatGPT, the world's most popular AI chatbot. The question every ads manager asks is simple: is it worth shifting part of the budget there, or is Google Search still the only sensible choice?
The first available data, collected from 102 sites tracking transactions and revenue, gives a clear but not definitive answer. ChatGPT Ads converts less and costs more per click. But the platform is young, and the numbers could change quickly.
| Your situation | Is testing ChatGPT Ads worth it? | why |
|---|---|---|
| E-commerce with mass-market products | No, for now | Low revenue per session; Google Search converts far better |
| B2B with long sales-cycle leads | Yes, with a test budget | Lower CPC, new-customer traffic (>80% new users according to OpenAI) |
| SaaS or software with a free trial | Yes, with a test budget | Criteo reports conversions 1.5x higher than other referrals |
| Professional services (legal, finance, consulting) | Yes, with a test budget | ChatGPT users look for complex answers: a favorable context |
| Retail with thin margins | No, for now | Current ROAS doesn't justify the risk on limited budgets |
What the data really says: revenue, conversions and cost per click
A study of 102 sites with real tracking shows that a ChatGPT Ads session generates $0.89 in revenue versus $2.45 for a Google Search session, with a conversion rate of 0.37% versus 1.17%.
The study analyzed by Search Engine Journal covers 60 weeks of aggregated data (through August 22, 2026) on US traffic. The numbers are descriptive, meaning they capture what happened. They don't prove that ChatGPT Ads is structurally worse for every type of business.
Of 87 advertisers who used both channels, 84 earned less revenue per session with ChatGPT Ads. That's a significant finding, but it needs context: the platform was still in its early stage during the measurement period.
| Metric | ChatGPT Ads | Google Search Ads |
|---|---|---|
| Revenue per session (US) | 0,89 $ | 2,45 $ |
| Conversion rate per session | 0,37% | 1,17% |
| Average observed CPC | $1.72 (1 account, 15 days) | $0.72 (median of 16 accounts) |
| Revenue per dollar spent (estimate) | 0,52 $ | 3,40 $ |
| Documented ROAS (agency case) | 1.49x (Opascope) | Not available in the sample |
Cost per click is the crux. OpenAI recommends a starting maximum bid of $3 to $5 per click, but Opascope, an agency that publicly documented about $60,000 in spend over 15 days, recorded an average CPC of $1.72, well below the suggested range. The maximum bid is only the ceiling you set, not the price you pay for each click.
A B2B case documented by the agency StubGroup shows a different picture: a CPC of $0.76 on ChatGPT versus $10.46 on Google Search, with a cost per lead of $92 versus $157. The conversion rate, however, was 0.8% versus 6.64%. For some B2B sectors, a higher volume of leads at a lower cost can offset the lower conversion rate.
How big is ChatGPT as an advertising platform compared to Google?
ChatGPT Ads reached $1 billion in annualized run rate in under 200 days from launch. Google Search generated $224.5 billion in revenue in 2025 alone: the difference in scale is enormous.
To understand where we stand, you need the context numbers. OpenAI has stated an annualized advertising run rate (the revenue pace projected over 12 months) above $1 billion, with more than 100,000 advertisers already running campaigns. The stated goal is $2.5 billion in advertising revenue by the end of 2026.
Meanwhile, Google isn't slowing down. According to official figures filed by Alphabet with the SEC, Google Search & other generated $224.5 billion in 2025, up 13.4% from $198.1 billion in 2024.
ChatGPT user growth is the figure that changes the outlook. OpenAI has announced it passed 1 billion weekly active users in August 2026, up from 400 million in February 2025. An audience this size can't be ignored in media planning.
What changes in practice: how do ads on ChatGPT work?
Ads appear in ChatGPT's responses as sponsored content that's contextual to the conversation. The bidding system is based on max CPC or CPM, with new Target CPA and Target CPC tools launched recently.
Unlike Google Search, where the ad appears next to search results, on ChatGPT the ad slots into the flow of a conversation. The user is asking something, and the system shows an ad relevant to the context. This changes the moment of interception: there's no explicit query (typed keyword), but a conversational intent.
OpenAI has recently introduced cost-target bidding, an automated bidding system that optimizes cost per acquisition (CPA) or cost per click (CPC) toward a target set by the advertiser. Before, there was only manual bidding with a cap. This brings the platform closer to the logic already familiar from Google Ads.
| Feature | ChatGPT Ads | Google Search Ads |
|---|---|---|
| Ad format | Contextual sponsored placement in the chat | Text above/below search results |
| Ad trigger | Conversational intent | Typed keyword |
| Bidding system | Max CPC, max CPM, Target CPA/CPC | Manual CPC, Target CPA/ROAS, Smart Bidding |
| Default maximum CPM | $60 per 1,000 impressions | Varies by auction |
| Platform maturity | Launched Feb 2026, evolving rapidly | 20+ years of optimization |
| Available optimization data | Limited (a few months) | Deep history for every account |
Three real-world scenarios: what happens for different Italian companies?
The outcome depends heavily on the industry, the sales cycle and customer value. Here are three illustrative scenarios to understand when the test makes sense.
The numbers that follow are illustrative examples: they show the logic of the calculation and aren't real client data. CPC and conversion values are based on the public benchmarks available.
| Scenario | Test budget | Estimated CPC | Expected leads/orders | Estimated cost per lead | Assessment |
|---|---|---|---|---|---|
| Fashion e-commerce (average order value €80) | €500/month | 1,72 $ (~1,60 €) | ~312 clicks, ~1.2 orders (0.37% conv.) | ~€417 per order | Not sustainable with standard margins |
| B2B consulting firm (average contract €5,000) | €500/month | 0,76 $ (~0,70 €) | ~714 clicks, ~5.7 leads (0.8% conv.) | ~€88 per lead | Worthwhile if the close rate is >2% |
| SaaS with free trial (average LTV €1,200) | €500/month | 1,72 $ (~1,60 €) | ~312 clicks, ~1.2 trials (0.37% conv.) | ~€417 per trial | Sustainable if the trial-to-paid rate exceeds 35% |
The pattern is clear: ChatGPT Ads works best where customer value is high and the sales cycle is long. For e-commerce with tight margins, the current numbers can't compete with Google Search. This is consistent with what Inc. and Criteo reported: the platform isn't right for every brand.
Why everyone's talking about it now: is the search market really changing?
Google Search is still growing at double digits, but for the first time its Chief Business Officer cited lower monetization of AI Mode as a drag on growth. The market is splitting in two.
The European search advertising market reached €56.2 billion in 2025, up 8.8% from 2024, according to the IAB Europe AdEx Benchmark 2025 report. Search remains the second-largest format in European digital advertising. This isn't a market in crisis. It's a market that's expanding as new players emerge.
The most interesting signal comes from Alphabet itself: on the Q3 2025 earnings call, the Chief Business Officer cited "the lower monetization we were seeing in AI Mode" as one of the factors slowing Search growth. He gave no figures, but it's the first time Google has publicly acknowledged internal pressure from AI.
If you handle ADV budget allocation across dominant platforms, you know that relying on a single channel is always a risk. ChatGPT Ads isn't a threat to Google yet, but it's already a real alternative for certain segments. The right time to test a channel is when it's still cheap to do so, not when everyone is already there.
What tools and strategies exist to test ChatGPT Ads without wasting budget?
The key is to isolate the test with a dedicated budget, measurable goals and a direct comparison with your active Google campaigns on the same products or services.
Before allocating budget to ChatGPT Ads, it's worth looking at what's happening on the Google side too. The recent changes to priority and budget allocation by channel in Google Ads are already changing how funds are distributed across campaigns. Adding a new channel without optimizing your main one risks amplifying inefficiencies.
| Approach | When to use it | Benefit | Limit |
|---|---|---|---|
| Isolated test (separate budget) | Always, as a first step | Clean data, no interference | Requires a dedicated additional budget |
| A/B across channels (same product) | After the initial test | Direct, comparable results | Needs enough volume for statistical significance |
| Complementary campaigns (ChatGPT awareness + Google conversion) | For brands with a long sales cycle | Leverages the strengths of both | Complex attribution to manage |
| ChatGPT Ads only | No, for now | — | Too risky without a benchmark for comparison |
What risks and limits should you keep in mind before investing?
The current data has important methodological limits: it covers an early phase of the platform, uses CPC prices from small samples and doesn't distinguish between different industries.
OpenAI itself responded to the study, stating that it "captures an early period of ChatGPT Ads development and extrapolates the platform's performance from a small sample." It's a fair criticism: 28 weeks of data on a fast-evolving platform aren't enough for definitive conclusions.
When it comes to measurement, the problem is structural. As explained in a dedicated article, advertisers testing ChatGPT Ads still struggle to understand what the metrics they see in the dashboard actually mean. Without a clear measurement framework, campaign data risks being misleading.
Another concrete risk is invalid traffic. Anyone running Google campaigns already knows how important it is to monitor anomalies: Google's new invalid traffic dashboard shows how much budget can be lost without proper controls. On ChatGPT Ads, equivalent tools don't exist yet.
| Risk | Impact | How to mitigate |
|---|---|---|
| Scarce optimization data | High: the algorithm learns slowly | Small test budget, simple campaigns at the start |
| Unpredictable actual CPC | Medium: varies widely by industry | Set a conservative cap ($3-5 as suggested by OpenAI) |
| Unclear attribution | High: hard to see the real contribution | Use UTM parameters and server-side tracking |
| No direct comparison with Google | Medium: different usage contexts | Define separate KPIs for the two channels |
| Rapidly evolving platform | Low in the short term, high in the long term | Review campaigns every 30 days |
How to get started with ChatGPT Ads: the concrete steps
You need five sequential steps: define the goal, isolate the budget, set up tracking, launch a simple campaign and compare the results with Google after 30 days.
- 1 · Marketing managerDefine the goal and the main KPIChoose a single measurable goal: lead, trial, order. Set the maximum acceptable cost for that result before you start.
- 2 · Ads managerSet aside a separate test budgetUse additional budget; don't take funds from your active Google campaigns. A test of €500-1,000 per month for 30 days is enough to get indicative data.
- 3 · Ads manager / DevSet up tracking before you go liveConfigure UTM parameters on all ad URLs. Make sure Google Analytics or your analytics system receives sessions correctly before you spend your first euro.
- 4 · Ads managerLaunch a simple campaign with a conservative bidStart with a maximum CPC of $3-4, as suggested by OpenAI. Pick the product or service with the highest customer value. Avoid overly fragmented campaigns at the start.
- 5 · Marketing managerCompare results with Google after 30 daysCalculate the cost per lead or per order on both channels. Decide whether to continue, optimize or stop the test based on your business's real numbers.
What mistakes are most common when testing a new ad channel?
The most common mistake is comparing different channels using the same KPIs without considering the different context of use. The second is cutting the test too early, before you have meaningful data.
If you've been running Google campaigns for years, you tend to bring the same expectations to a new platform. But ChatGPT Ads reaches users at a different moment than a Google search: there's no explicit query, there's a conversation. The conversion rate will almost always be lower. The question is whether the final cost per acquisition is still acceptable.
Another frequent mistake is overlooking the value of new customers. OpenAI reports that over 80% of the traffic generated by ads on ChatGPT comes from new customers. If your business has a high customer lifetime value (LTV), acquiring new users at a slightly higher cost can still make sense.
On the Google side, it's worth remembering that there are dynamics to keep an eye on there too: increasing budget on non-brand B2B keywords can push CPCs up in a non-linear way. Channel diversification isn't just about opportunity. It's also risk management.
| Error | Consequence | Solution |
|---|---|---|
| Using the same KPIs as Google Search | Wrong conclusions about the channel | Define ChatGPT Ads-specific KPIs |
| Cutting the test after 7-10 days | Insignificant data | Minimum 30 days with sufficient volume |
| Not tracking sessions with UTMs | Impossible to attribute results | Set up tracking before launch |
| Taking budget from Google to fund the test | Existing campaigns get worse | Separate, additional test budget |
| Ignoring the value of new customers | Underestimating the channel | Calculate LTV, not just the first conversion |
How do I know if ChatGPT Ads is working for my business?
The three key indicators are real cost per acquisition, the share of new customers and a direct comparison with the cost per acquisition of your active Google campaigns on the same products.
| Indicator | How to measure it | Alert threshold |
|---|---|---|
| Cost per acquisition (CPA) | Total spend ÷ conversions in the period | Higher than Google CPA on the same product |
| % of new customers out of total | ‘New user’ segment in analytics | Below 60%: the channel isn't bringing in a new audience |
| Revenue per session | Attributed revenue ÷ sessions from ChatGPT Ads | Below $0.89: below the benchmark study's average |
| ROAS (return on ad spend) | Attributed revenue ÷ ad spend | Below 1.5x: the channel doesn't pay for itself |
| Conversion rate | Conversions ÷ clicks | Below 0.37%: below the benchmark study's average |
So, is ChatGPT Ads worth it for your business?
Worth it if you run a B2B, SaaS or professional services business with high customer value and a long sales cycle. Cost per lead can be competitive, the audience is new and qualified, and the time to get in, before CPCs go up, is now.
Somewhere in between if you run an e-commerce store with medium-to-high value products or a brand that wants to reach a new audience. Test with a separate budget, measure your real CPA and compare it with Google after 30 days. Don't move existing budget: add to it.
Not yet if you run an e-commerce store with tight margins, mass-market products or Google campaigns that aren't optimized yet. The current data doesn't support shifting budget. Wait for the platform to mature and for industry benchmarks to become more solid.
How we see it at SHM
For us, ChatGPT Ads is exactly the kind of innovation that should be analyzed before it's chased. The data tells us the platform isn't competitive for everyone yet, but it also tells us it has 1 billion weekly users and growth that no other channel has shown in such a short time. Ignoring it would be a strategic mistake; throwing budget at it without a measurement framework would be an operational one.
We believe innovation should be turned into lasting value, not chased as a fad. That's why, when we work on the strategic integration of AI into marketing processes for our clients, we always start from the real data of their business, not from average industry benchmarks.
With some clients, we're already structuring ChatGPT Ads tests with isolated budgets and server-side tracking, running alongside existing Google campaigns without touching them. The goal isn't to replace one channel with another: it's to understand where every euro works best. If you're evaluating how to optimize your Google Ads strategy or how to approach new paid channels, we can help you build a test framework that produces data you can base decisions on.
You can explore all our work in the Marketing and ADV performance area or contact us directly to talk through your specific case.
Frequently Asked Questions
It depends on the industry and the customer value. For SMEs with high-value products or services (consulting, software, training, professional services), a test with €500-1,000 per month can deliver useful data. For those selling low-margin products, the current data doesn't support the investment.
No. The experts unanimously recommend a separate, additional budget for the test, without touching your active Google campaigns. Google Search still delivers the highest ROAS among the paid channels available.
A minimum of 500 euros a month for at least 30 days is the starting point for indicative data. With less budget or less time, the data isn't enough to make reliable decisions.
You need to set up UTM parameters on all your ad URLs and check that your analytics system (Google Analytics or equivalent) is receiving the sessions correctly. The native ChatGPT Ads dashboard is still limited, so external tracking is essential.
The available data suggests B2B is one of the most promising use cases. One documented case shows a cost per lead of $92 on ChatGPT versus $157 on Google Search, with much lower CPCs. The conversion rate is lower, but the cost per lead can be competitive.
The current study covers the platform's first few months. OpenAI has already launched new tools (Target CPA, Target CPC) that weren't available during the period analyzed. More complete benchmarks are expected by the first half of 2027, when the platform will have at least 12 months of mature data.
The real competitive advantage isn't being first on ChatGPT Ads. It's being first to have real data from your own business to base your decisions on.
Related articles
Discover more articles exploring similar topics, selected to offer you a more complete and stimulating perspective. Each piece of content is carefully chosen to enrich your experience.