What We Learned From Running ChatGPT Ads in Malaysia for a Week

Since ChatGPT ads came to Malaysia, the first thing people ask us is, “Does it actually work?” Fair question. So before telling anyone to spend on it, we ran ads on our own account and tracked every number in OpenAI's Ads Manager and Google Analytics. Here's what we saw, what surprised us and what we'd do differently.

Written by

Cleverus Team

Published on

BlogChatGPT

Since ChatGPT ads came to Malaysia, the first thing people ask us is, “Does it actually work?” Fair question. So before telling anyone to spend on it, we ran ads on our own account and tracked every number in OpenAI's Ads Manager and Google Analytics. Here's what we saw, what surprised us and what we'd do differently.

New to this? Start with how ChatGPT Ads work in Malaysia first.

Do ChatGPT Ads Actually Work?

Yes, people see them and people click. Our ads were shown about 6,300 times and got 153 clicks, a CTR of around 2.4%. We paid an average of RM1.45 per click, which works out to about RM39 for every 1,000 views.

People also noticed. A contact forwarded us a WhatsApp message showing we are running ads on ChatGPT, with a screenshot of our ad sitting inside a ChatGPT answer. That's a small thing, but it tells you the ad is visible enough that people talk about it.

Two screenshots side by side: a labelled ad shown inside a ChatGPT answer, and a forwarded WhatsApp message reading you guys running ads on chatgpt?
Our Cleverus ad inside ChatGPT, and a WhatsApp message asking if we are running ads on ChatGPT

We ran two ads on different topics, and one clearly beat the other. The GEO ad had a 2.6% click rate and the SEO ad 2.0%, and the GEO ad brought in far more of the conversions Ads Manager counted. So test at least two before you decide what works for you.

The early days looked expensive, though. Like most paid platforms, ChatGPT ads seem to have a learning phase, where the system works out who to show your ad to. On day one we paid RM2.63 per click. By day five it was down to RM1.17, and it was still drifting lower. Google's own help page says its bidding systems can take a few conversion cycles to calibrate, and our numbers look similar. We can't say exactly how long ChatGPT's learning phase lasts, but plan for a week or more before judging.

Line chart of average cost per click in ringgit, falling steadily from 29 September to 3 October
Ads Manager chart showing cost per click falling from RM2.63 on day one to RM1.17 on day five

In our view, this is where many advertisers lose their nerve. They see an expensive first day, switch it off and decide ChatGPT ads don't work. Decide before you launch how long you'll wait. To be fair, this is one account over a short run, and we can't yet say how many clicks turned into real customers.

Who Sees Your ChatGPT Ad and What Do They Do Next?

About 70% of our ad views happened inside the ChatGPT iOS and Android apps. Around 25% were on desktop web and only about 4% on mobile web. We're a B2B business, where you'd expect more people on a work laptop, and the app still won by a wide margin. If that's true for us, retail and B2C businesses may well see an even heavier app share, since their customers browse on their phones all day. We haven't tested that ourselves, so treat it as a likely guess.

Line chart of ad impressions by platform from 29 September to 4 October, with the iOS app and Android app lines highest, then desktop web, then mobile web
Ads Manager chart of impressions by platform, with iOS and Android apps well ahead of desktop and mobile web

What happened after the tap surprised us. Google Analytics recorded 129 sessions from the ad, and those visitors stayed about 11 seconds on average. The 50 sessions that came from ChatGPT mentioning us in an answer stayed about 53 seconds. Same app, very different mood. In our view, an ad catches someone mid-conversation, and they're not ready to read a long page yet.

Google Analytics traffic acquisition report: chatgpt paid with 129 sessions and 11 seconds engagement, chatgpt.com ai-assistant with 50 sessions and 53 seconds
Google Analytics table comparing visitors who clicked our ChatGPT ad with visitors from ChatGPT mentions

So the page does as much work as the ad. Open your landing page on your own phone right now. Does it load fast? Can someone see the price or the WhatsApp button within three seconds? Here's how that plays out by business type:

  • Café or restaurant: show the menu and location first, not your story.
  • Tuition centre: lead with the fee and a free trial button.
  • Property agent: put the project name, price range and WhatsApp button on the first screen.
  • Online shop: send people to the product, not the homepage.

Why Doesn't My ChatGPT Ads Conversion Number Match Google Analytics?

This was the biggest lesson of the week. Ads Manager showed 56 conversions. Google Analytics showed 27 conversion events from the same ad traffic, and only 15 visitors tapped our WhatsApp button. One person asked for a free audit and nobody filled in the contact form. If we'd trusted the dashboard alone, we'd have thought the ads were doing far better than they were.

We're still tracing exactly why. Part of it is plain counting. A conversion is only whatever you told the system to count, and if you pick something easy, like a tap on a WhatsApp button, it will happily count a lot of them. A tap isn't a lead, because tracking can't tell you whether anyone actually sent a message or got a reply. Here's what we'd tell any business before the first ringgit goes in:

  • Pick one action that really is a lead, such as a booking, a purchase or a form you can match to a real customer. Treat a WhatsApp tap as a signal only.
  • Do the action yourself and check it shows up once, and only once.
  • Every week, compare the ad platform, Google Analytics and your own records, like your WhatsApp inbox or sales sheet.
  • If the numbers don't match, trust your own records first and fix the tracking before spending more.

Until the numbers match, don't raise the budget. Our first campaign guide walks through the tracking setup step by step.

Four numbered cards: pick one real lead, test it yourself, match it every week and fix before you scale
Four checks to run before you trust your ChatGPT Ads conversion numbers

Should You Start ChatGPT Ads Now or Wait?

It depends on what you sell. This is our view after one test, not a rule.

Business typeOur viewWhy
Property, education, business servicesTest small nowPeople ask detailed questions in chat before choosing, so an ad can meet them early.
Online shopsTest small nowIt's worth a test if product pages load fast on a phone and purchases are tracked properly.
Cafés, salons and other local businessesPrepare firstGet your mobile page and WhatsApp replies ready, so the first clicks aren't wasted.
Finance, health and legalCheck the policy firstOpenAI's Ad Policies say finance and health services ads are generally prohibited outside the US, and legal services ads are currently prohibited.

Whatever industry you are in, set a test budget you can afford to lose and judge it after a full week or more, not a day.

If you're not sure and need help, feel free to contact us at Cleverus to schedule a consultation.

References

Business Today Malaysia. (2026). ChatGPT Rolls Out Advertising Services In Malaysia. Available at: https://www.businesstoday.com.my/2026/09/24/chatgpt-rolls-out-advertising-services-in-malaysia/.

Google Ads Help. (2026). Duration of the learning period for campaigns and what affects it. Available at: https://support.google.com/google-ads/answer/13020501.

OpenAI. (2026). Ad Policies. Available at: https://openai.com/policies/ad-policies/.

OpenAI Help Center. (2026). Ads in ChatGPT. Available at: https://help.openai.com/en/articles/20001047-ads-in-chatgpt.

FAQs on ChatGPT Ads in Malaysia

  • How much do ChatGPT ads cost in Malaysia?

    You pay for views or clicks, billed in ringgit, and you set your own budget and bid. In our test the average was RM1.45 per click, or about RM39 per 1,000 views, with day one at RM2.63 per click before it settled. Your cost will depend on your topic and how much competition there is, so treat our figures as a reference point, not a promise.

  • Can I pay to make ChatGPT recommend my business?

    No. Ads are labelled, and OpenAI says ads do not influence ChatGPT's answers. A paid ad is separate from the answer itself. Getting named in the answer is GEO work, which is a separate job from buying ads.

  • Are ChatGPT ads better than Google Ads?

    It's too early to say, and we'd treat them as different jobs. Google Ads catches people searching to buy right now. A ChatGPT ad meets people in the middle of a conversation. We'd test ChatGPT ads next to Google, not instead of it.