ChatGPT Ads Case Studies: What the Published Results Show for Arab Advertisers

Reading time 15 min
Summarize with AI
Oct. 28, 2026

ChatGPT ads can work, but the evidence so far is thin and mostly sold. OpenAI says its ad business passed a $1 billion annualized run rate in under 200 days. Independent tests tell a harder story: in all four head-to-head comparisons one roundup could document, Google Ads converted better. None of the published cases comes from an Arab market.

Key takeaways

OpenAI’s own figures (a 3x return in 28 days, 80%+ new customers) come from single, unnamed advertisers and are not audited.
Panel data puts typical click-through rates between about 0.5% and 1.9%, against about 6% on Google search.
The clearest wins are B2B signups and lead forms with a short path from click to action. The clearest losses are demo requests and software purchases.
Pricing fell from a $60 CPM with a $200,000 minimum in February 2026 to self-serve CPC bidding with no minimum by May.
Advertisers in Saudi Arabia, the UAE and Egypt can buy ads today, but nobody has published Arabic-language results.

$1B

Annualized ad revenue run rate, in under 200 days

OpenAI, 31 August 2026

0.5% to 1.9%

Typical ChatGPT ads click-through, against about 6% on Google search

Panel data, 2026

Demo requests from 363 clicks in one enterprise test

Nakora, 2026

What does OpenAI itself report about ChatGPT ads results?

OpenAI reports scale, not benchmarks. On 31 August 2026 it said ChatGPT ads had reached a $1 billion annualized revenue run rate less than 200 days after launch, with tens of thousands of advertisers. It has published two performance examples. One ecommerce advertiser saw a 3x return on ad spend over 28 days, and a technology partner reported that more than 80% of ad-driven traffic came from new customers.

A run rate is current monthly revenue multiplied by twelve, so it measures how much advertisers are spending, not whether that spending pays back. The only named result came on 5 October 2026, when DoubleVerify reported that WeightWatchers’ attributed cost per acquisition ran 15.3% below an “established benchmark”, without publishing spend, dates or the benchmark itself. OpenAI’s help centre still says the platform has no performance benchmarks.

What do independent ChatGPT ads case studies show?

Independent cases split almost evenly between modest wins and clean failures. We reviewed more than a dozen published tests and roundups in October 2026. Almost every positive number comes from an agency or tool that sells ChatGPT ads, and almost none discloses a full method.

Independent cases split almost evenly between modest wins and clean failures.

Case and date	What it reports	Caveat
Opascope, one advertiser account, 2026	About $60,000 over 15 days, 1.49x return, $1.72 per click, 2.35% conversion rate	Daily return swung between 0.2x and 2.9x.
InterTeam Marketing, three B2B SaaS clients, mid-2026	150+ qualified leads at a cost per lead about 60% below Google	Agency that sells the service. No spend disclosed.
Nakora, five US and UK tests	Cost per signup 18.1% below Google ($80 vs $98 on $3,600 over 28 days); one enterprise test stopped after 0 demos from 363 clicks	Nakora sells ChatGPT ads and ran every test. No dates.
ScaleWise VA, 10 Shopify stores, 28 March 2026	4.8x average return on $178,400 over 90 days, range 1.9x to 8.4x	Agency that sells the service. No method.
MalachiSoft, Texas wellness practice, July to August 2026	22 leads on $1,139: $52 blended, $114 per ready-to-book lead	12 of the 22 were ebook downloads. No revenue reported.
GravityKit, WordPress software, about three weeks	1.7% click-through rate, about EUR 2.22 per click on roughly EUR 475	Zero product views, checkouts or purchases. Stopped early.
Workshop Digital, B2B, 2026	739 clicks on $2,319 at a 1.1% click-through rate, against 5.2% on Google search	0 qualified leads, while Google produced 2.
Adthena, enterprise brands, 2026	0.91% click-through rate against a 6.4% Google benchmark for the same sector	One advertiser spent 3% of a $250,000 budget in several weeks.

Which results held up, and which did not?

The tests that worked had a short path from click to action: a free signup, a lead form, or a product the user could buy on the next page. Nakora’s two SaaS tests cut cost per signup by 18.1% against Google Ads and 51.6% against Meta. InterTeam reports qualified B2B leads at well under half its Google cost.

The tests that failed asked a chat user to do something heavy. Nakora’s enterprise test bought 363 clicks at about $14.80 each and got zero demo requests, with engaged sessions averaging 7 seconds against 54 on its benchmark channel. GravityKit’s 1.7% click-through rate looked healthy and produced no purchases at all.

A cheap click from ChatGPT is only worth something if the next page finishes the job the answer started.

How much do ChatGPT ads cost in 2026?

Prices have fallen fast as OpenAI moved from a managed pilot to self-serve buying. The US pilot launched on 9 February 2026 at a $60 CPM with a $200,000 minimum commitment. CPMs dropped as low as $25 by mid-April, and on 5 May the self-serve Ads Manager opened with cost-per-click bidding and no minimum spend.

Date	What changed
9 February 2026	US pilot for Free and Go users, $60 CPM, $200,000 minimum
Mid-April 2026	CPMs reported as low as $25
5 May 2026	Self-serve Ads Manager, CPC bidding, no minimum
6 June 2026	UK launch, with CPMs reported around $30
17 September 2026	Saudi Arabia and the UAE open; SAR billing at a 50 riyal daily minimum
21 October 2026	Availability list grows to 89 countries

On clicks, the PPC ONE roundup lists US campaigns from $0.76 to $14.80 per click, with half below $4.41. Nakora paid about $3.70 per click in the UK against $9.40 in the US in the same 18-day test. Ads Manager offers CPM, CPC and conversion-optimised bidding, and smaller advertisers usually start on CPC.

A cheap click from ChatGPT is only worth something if the next page finishes the job the answer started.

What click-through rate should you expect from ChatGPT ads?

Plan for a click-through rate between 0.5% and 2%. Similarweb reported ads in 26% of responses to Free and Go users with a 0.50% click-through rate as of 17 August 2026. Its market panel, published by Graphite on 29 September, puts the all-time US rate at 0.73% and Brazil, the highest of nine markets, at 1.91%. No Arab market is in that data.

Market	All-time CTR to 16 September 2026
Brazil	1.91%
Mexico	1.55%
Japan	1.53%
South Korea	1.21%
Canada	0.75%
United Kingdom	0.74%
United States	0.73%
Australia	0.67%
New Zealand	0.60%

The four markets above 1% are all non-English. Whether Arabic prompts follow that pattern is untested, and we would not budget on it. Creative matters at the margin: Workshop Digital saw about 1.2% with specific context hints against 0.9% with broad conversational ones. Relevance is not guaranteed either, because SE Ranking found about 14.35% of ads had no topical link to the prompt across 50,006 commercial prompts.

Is paid ChatGPT traffic worth more than organic answers?

On the evidence so far, organic visibility inside the answer is worth more than the ad next to it. In a dataset of US home-services contractors from June and July 2026, leads from paid ChatGPT ads averaged about EUR 1,087 per job. Leads that found the business through ChatGPT’s organic answers averaged about EUR 3,008.

The ad and the answer are also separate. SE Ranking found that the advertiser was cited in ChatGPT’s organic answer in only 3.63% of ad placements. Paying for the slot below the answer does not put you inside it. For a Gulf brand, that means a paid test and organic AI visibility are two different projects, and the second one compounds.

What did the UK trial teach advertisers?

Mostly how the platform works, not how it performs. A UK agency in the trial recommended starting at a GBP 45 CPM, with a GBP 15 daily minimum and a GBP 500 one-week test, and reported no CTR, CPA or ROAS. Ads reach only Free and Go users, with no audience targeting or retargeting yet.

Two setup lessons carry straight into Arabic campaigns. Ad copy is truncated early, so the offer has to sit in the first words, and the image renders small enough that it works as a brand cue rather than a message. A global network agency writing in Campaign from its own trial seat made a similar point: do not run the channel like paid search.

Why can’t these numbers be compared?

Each study measures a different outcome on a different attribution model. ScaleWise reports return on ad spend, MalachiSoft reports cost per lead, and the Similarweb panel divides clicks by ad-bearing exchanges. Ads Manager also credits view-through conversions inside a one-day window when no click claims them, so its totals can run ahead of what your CRM sees.

The gap runs in both directions. In the PPC ONE roundup, one US advertiser saw 36 conversions in Ads Manager against 2 real inquiries, and a Brazilian agency saw a dashboard showing zero conversions while three sales existed. Agency executives quoted by MediaPost also saw OpenAI’s click counts diverge from Google Analytics sessions. Your CRM has to be the source of truth.

How do you judge a ChatGPT ads case study?

Ask five questions before you trust any published result. Most case studies in this article fail at least two of them, which is why we label each one with its caveat rather than averaging them into a benchmark.

Who published it? An agency selling ChatGPT ads has a reason to publish wins and skip losses.
Is the spend and period disclosed? A ratio without a budget cannot be checked.
What counts as a conversion? A newsletter signup and a paid order are not the same result.
Was there a comparison channel in the same period? Without one, seasonality looks like performance.
Which market and language? Every result here is English-language and mostly American.

We applied five similar checks to 44 published ChatGPT ads cases in October 2026: disclosed spend, disclosed dates, a business outcome, a result checked outside Ads Manager and a comparison channel. Only 2 passed all five, and both showed ChatGPT ads losing to the advertiser’s own baseline. Of 11 cases where the ads beat a comparison channel, 9 came from a seller of ChatGPT ads or were selected by OpenAI. The full scorecard is in our evidence audit of 44 ChatGPT ads case studies.

What does this mean for advertisers in Saudi Arabia, the UAE and Egypt?

Thirteen Arab markets can now buy ChatGPT ads, but nobody has published how those ads perform. As of 27 October 2026, OpenAI’s availability page lists 89 countries, including Saudi Arabia, the UAE, Egypt, Qatar, Kuwait, Bahrain, Oman, Jordan, Iraq, Lebanon, Morocco, Algeria and Tunisia. Self-service access depends on where the advertising and billing legal entity is based.

The audience is there. In September 2026, ChatGPT sent 75% of the AI chatbot referrals StatCounter tracked in Saudi Arabia, 80.99% in the UAE and 72.16% in Egypt. Keep that in proportion, though: in the same Saudi data, Google search sent 99.61% of combined Google and ChatGPT referrals, and ChatGPT sent 0.39%. ChatGPT dominates chat, not discovery.

Two local constraints change the plan. First, ads for health, financial and legal services outside the US were prohibited in most cases when we checked the policy on 19 October 2026. That rules out most clinics, labs, lenders, brokers and law firms in the region for now. Second, new self-serve accounts start limited to their home country until identity checks clear, so a Cairo company cannot target Riyadh on day one. Our record of which Arab markets opened, and when, is in ChatGPT Ads in the Arab World: 13 Countries Now Open, Including Saudi Arabia and the UAE.

How should an Arab advertiser run a first test?

Run it as an experiment with a pass mark, not as a channel launch. At the PPC ONE median of $4.41 a click and a 0.9% conversion rate, the top of its head-to-head range, 30 conversions need about 3,300 clicks, or roughly $14,700 in the US. The SAR minimum of 50 riyals a day buys you delivery data, not a verdict.

A first test in the Gulf or Egypt should cover five points:

Eligibility. Confirm your billing entity’s country is on the list and your category is not restricted.
Measurement. Install the pixel or Conversions API and reconcile every conversion against orders or booked calls in your CRM each week.
Language. Run Arabic and English ad groups separately, each sending traffic to a landing page in the same language.
Comparison. Keep a search campaign live on the same offer and the same dates, so you have a baseline.
Kill rule. Write the stop condition before launch, for example zero qualified leads after a set click count, or engaged time far below your search traffic.

My read on this

The number I watch is the one nobody headlines: time on site after the click. Nakora’s failed test showed 7-second sessions before it showed zero demos, and that is the earliest warning an Arab advertiser can get without waiting for conversions. Click-through rate tells you the ad was noticed. Engaged time tells you the click came from someone with a real need.

I would also resist the agency case studies with 3x to 8x returns until someone publishes the method behind them. The honest position for an advertiser in Riyadh, Dubai or Cairo in late 2026 is that the channel is open, the audience is real, and the economics are unproven in Arabic. That is a reason to test carefully, and to work on being the answer, not only the ad below it.

Questions advertisers ask about these results

The agency showed me a 4.8x return. Why shouldn’t I plan around it?

Because it is an average from 10 unnamed Shopify stores, reported by the agency that sells the service, across a range from 1.9x to 8.4x with no disclosed method. One store at 1.9x and another at 8.4x are not the same business case, and neither of them sold in Arabic.

Our dashboard shows conversions but sales did not move. What is happening?

This is an attribution gap, and the published cases show it. Ads Manager counts click and one-day view-through conversions on its own rules, and your analytics counts on different ones. Reconcile against orders or booked appointments in your CRM before you judge the campaign.

Is a small budget in Egyptian pounds or riyals enough to learn anything?

It is enough to check delivery and click quality, not conversion economics. MalachiSoft learned its lead cost on $1,139 over six weeks, while a ScaleWise client that started at $400 a month had to restart at a higher budget. Size the test to the number of conversions you need, not to a round figure.

Can my clinic in Riyadh or my law firm in Cairo advertise?

Probably not yet. On the policy page we checked on 19 October 2026, ads for health, financial and legal services outside the US were prohibited in most cases, with manual review for some categories. Check the current wording before you plan around it, because OpenAI is rolling these categories out gradually.

Should we wait until there is Arabic data?

Only if you cannot measure. A business that can track a conversion in its own CRM and fund a test of a few thousand clicks will produce Arabic data before anyone publishes it. A business that cannot should fix measurement first, because no channel can be judged without it.

The better question

The useful question is not whether ChatGPT ads work, because the published cases show they sometimes do and often do not. It is whether your offer can produce a measurable action inside a small, time-boxed test. If you can name that action, count it in your own system and keep a search baseline running beside it, a test is a reasonable experiment. If you cannot, start with measurement and with how ChatGPT already describes your brand in organic answers.

Plan my ChatGPT ads test or Show me how ChatGPT answers about my brand