Thailand has access. Your target country is a separate check.
OpenAI announced the Thailand rollout on 23 September 2026, alongside Indonesia, Malaysia, the Philippines, Singapore, Vietnam and Taiwan. Its current self-service table lists Thailand as available. Creating a login still doesn't establish account approval or permission to advertise everywhere.
This matters for a Thai hotel selling to overseas guests. Some new accounts initially have home-country targeting only; international eligibility requires identity approval and actual home-country ad spend, with no public spending threshold. Check the countries your account permits before planning an overseas campaign. Don't assume Bangkok-level targeting or Thai-language performance from the country launch.
A ChatGPT ad reaches a different buying moment.
Google Search ads start with what someone searches for. Facebook and Instagram ad selection also draws on predicted interests and behaviour. ChatGPT ads can match the context of a conversation, which creates an opportunity when a buyer is working through constraints rather than typing a finished product search.
Suppose someone is comparing weekend courses in Bangkok and needs English instruction. A training business can describe that actual offer and send the visitor to a page covering dates, language and price. This is an illustrative scenario, not a DLVX campaign. Context hints describe useful customer situations; they aren't exact-match keywords, exclusions or guaranteed placements.
Before asking for an enquiry, the example course page should answer the questions that could disqualify a booking: total price, teaching language, available dates, required experience and location. Keep those facts consistent with the ad. If someone discovers the wrong language only after opening LINE, the business has paid to start a conversation that should never have reached staff.
Reach has limits. Free and Go users may see ads; Plus, Pro, Business, Enterprise and Edu are ad-free. Temporary Chats and accounts identified as under 18 don't receive them either. For a B2B seller, that is a reason to investigate reachable buyers, not proof that every business buyer is absent.
THB 205 is a budget setting, not your cost per customer.
For THB billing, OpenAI publishes a THB 205 minimum daily budget. Daily spending can reach twice the selected amount, while the applicable seven-day budget limit and proration still apply. Choose a campaign-total cap if the experiment has a firm overall loss limit; don't treat the daily number as a hard daily ceiling.
Actual click prices come through an auction. A Conversions objective also doesn't mean you pay only when somebody buys: billing still uses clicks or impressions. We found no defensible Thailand-specific CPC or customer-acquisition benchmark in the evidence reviewed. Calculate your own affordable price before treating someone else's cheap click as a reason to spend.
Work backward from a paid customer: an example in baht.
Imagine a Thai training business keeps THB 3,600 per paid booking after variable delivery costs, before advertising and fixed overhead. It chooses to allocate at most THB 1,800 to acquiring that customer. Every number here is a teaching assumption, not a DLVX price or a ChatGPT benchmark.
Assume 4% of ad clicks become qualified enquiries and 25% of those enquiries pay. Multiplying those rates gives a 1% click-to-customer rate. The affordable media cost per click is therefore THB 1,800 × 1% = THB 18. A qualified enquiry must meet the business's actual booking criteria; tapping a LINE button doesn't qualify by itself.
At THB 10 per click, that assumed funnel gives a THB 1,000 media acquisition cost per customer. At THB 25 it becomes THB 2,500, already above the chosen allowance. Neither price is a prediction. If the real click-to-customer rate halves, the affordable CPC also halves to THB 9.
The remaining contribution pays overhead and supports profit; it isn't net profit. Add setup, creative and staff costs when judging all-in acquisition cost. Use actual completed-customer economics where you have them, and avoid assumed lifetime value that the business hasn't demonstrated.
Public cases show why a cheaper lead can disappoint.
StubGroup's case, updated 13 October 2026, compared one B2B compliance-services business over the same 30-day period. It reported ChatGPT clicks at USD 0.76 versus Google's USD 10.46, and leads at USD 92.21 versus USD 157.57. Google leads became customers more often, producing materially lower customer-acquisition cost.
That is useful evidence against judging by cheap leads alone. It isn't a controlled verdict on either platform: StubGroup sells advertising services, didn't disclose the underlying customer counts or numerical CAC, and compared a new channel with Google campaigns that had years of optimisation. The sales cycle was still incomplete.
PromptMarketing reported a more promising ecommerce signal: EUR 1,003.21 spent and EUR 3,006.03 in platform-attributed order value across selected European webshops during 1 to 25 September. Its roughly 3x ratio included 30-day click and one-day view attribution. Verified orders, returns and margins weren't connected, so the authors didn't claim verified profit. These selected account reports give us questions to ask, not Thai performance forecasts.
Follow the LINE handoff all the way to payment.
For the hypothetical training business, the useful journey is a matching mobile landing page, a genuine enquiry, qualification and a paid booking. LINE can sit between those steps. An outbound button click records a handoff; it doesn't prove a conversation happened, that the person qualified or that payment arrived. Record campaign source on enquiries where possible and reconcile paid customers with your booking records.
Use distinct paid-campaign tags so ChatGPT ad visitors don't merge with organic ChatGPT referrals. OpenAI supports Pixel and Conversions API measurement; if you send the same event through both, use matching event IDs to avoid duplicates. Review the information you collect and the applicable disclosure and consent requirements before enabling tracking.
Keep your attribution settings beside the results. Current reporting offers 7-, 14- or 30-day click windows and optional one-day view-through attribution; enabling the latter includes eligible view-through conversions in the total. Conversion reports can take 24 to 48 hours to update. Reconcile spend, qualified enquiries and paid customers after the normal sales cycle, rather than declaring success on launch day. An attributed sale still doesn't prove the ad caused an additional sale.
Paying for an ad does not buy an organic ChatGPT answer.
OpenAI explicitly separates ads from its answers. Advertisers cannot pay to change answer ranking or obtain users' private conversations and memories. Organic mentions and citations require their own discovery work. Keep paid campaign outcomes separate from organic AI visibility when someone reports results.
Our view is that conversational advertising makes a precise offer and an honest landing page more valuable. A business that explains who it fits, its constraints and its price gives the buyer something useful to evaluate. That doesn't establish that Google will disappear, or justify moving a working search budget wholesale.
Run a bounded test when these answers are ready.
We would prepare the following before committing budget. These are DLVX's decision rules, not OpenAI performance guarantees.
- Choose one offer and verify account approval, permitted target countries and the buyers who can actually see ads.
- Set a maximum customer-acquisition cost and an overall experiment cap that the business can afford to lose.
- Test a real enquiry through the mobile page and handoff to the booking or sales record; label test events so they cannot become campaign results.
- Agree what counts as qualified, reconcile paid outcomes and allow the actual sales cycle to finish before deciding whether to expand.
Sources and downloads.
Sources are linked directly. Market prices and engine answers can change, so use the checked date when citing this page.
- OpenAI: current Ads Manager country eligibility
- OpenAI: campaign budgets, objectives and targeting
- OpenAI: who sees ads and how ads relate to answers
- StubGroup: same-business ChatGPT versus Google case, updated 13 October 2026
- PromptMarketing: September 2026 ecommerce test and attribution limits
- OpenAI: conversion windows and reporting delays