As of 5 November 2026, OpenAI lists South Africa as available for self-service ChatGPT Ads. The legal entity being advertised and billed must be based in an available country. Access does not predict results: check the current beta terms, choose one offer, and measure qualified leads or sales before increasing spend.

Updated 5 November 2026: confirmed South Africa’s current self-service availability and refreshed OpenAI’s beta, bidding, reporting and product-feed details.

ChatGPT Ads are in beta. As of 5 November 2026, OpenAI lists South Africa as available for self-service Ads Manager. The legal entity being advertised and billed must be based in a listed country. Availability confirms that a business can sign up; it does not predict whether ads will be profitable. Check OpenAI’s current country list before setting a budget.

For a South African service business, ecommerce store or professional firm, start by choosing one offer and tracking the cost of qualified enquiries or sales. Keep the test small enough to learn from without relying on clicks alone.

What to check before spending

Before you put budget into ChatGPT Ads, check the foundation.

1. Can the landing page convert?

The page should have a clear offer, proof, pricing context, a visible call to action and a working form or booking path. Test it on mobile before launch.

2. Is the ad promise aligned with the page?

If the ad speaks to a specific problem, the page must continue that same thought. Do not send decision-stage traffic to a generic homepage.

3. Is tracking clean?

Use UTM parameters, confirm analytics receives the traffic, and test the conversion event. If you use server-side tracking, make sure the event names match the outcomes you actually care about.

4. Are status, billing and schedule controls ready?

Before publishing, confirm the campaign is in the intended status, the billing entity and payment method are correct, the budget is set, and the dates match your plan. OpenAI’s Quickstart says start and end controls may not appear before publishing a first campaign; future campaigns can be edited before they go live. Do not publish a first campaign only to reveal those controls: an approved campaign may begin delivering and accruing spend. If the first-campaign dates cannot be set yet, verify the current status and budget and follow the current Ads Manager guidance before continuing.

5. Can you absorb the spend and learning period?

Set the campaign budget before launch and account for OpenAI’s seven-day daily-budget pacing. If the maximum possible weekly spend would strain cash flow, lower the budget, set a lower campaign-total budget or wait until the offer is ready. Confirm the billing currency, payment terms and any applicable taxes in Ads Manager; South African availability does not establish those account-specific details.

6. Are you comparing against business outcomes?

Do not compare ChatGPT Ads to Google Ads only by CPC. Compare cost per qualified enquiry, cost per customer, lead quality, sales cycle and actual revenue.

7. Are you ready for limited query visibility?

You may not get the exact conversation that triggered the ad. Build your test around outcomes and landing-page performance, not a search-terms workflow.

8. Is the website ready for AI-era discovery?

Paid ads are only one part of this shift. Your organic website content still needs to be crawlable, useful and easy to cite. The Google AI Mode and ChatGPT Search checklist covers that wider visibility work.

What has actually changed

On 5 May 2026, OpenAI announced partner access and a beta self-service Ads Manager. In August, OpenAI reported that self-service access had expanded to additional regions; its current country list now includes South Africa. The Ads Manager Beta overview describes campaign setup and reporting, while warning that the beta continues to change.

OpenAI says ads do not influence ChatGPT’s answers and advertisers do not receive users’ conversations. Ads may use the current conversation for relevance; when ad personalization is enabled, selected signals from a user’s broader ChatGPT experience may also be used. These signals are not shared with advertisers. Read OpenAI’s advertiser guide for the current details.

Where ads appear

OpenAI says ads can appear below ChatGPT responses and is testing a multi-image visual format during image generation. The standard ad unit can include an advertiser name, logo, headline, description, landing page and image. Formats and placements may change during the beta; see OpenAI’s ad format details.

These are paid placements, not recommendations or endorsements by ChatGPT.

A conversation can show what a person is considering, but it does not make a click profitable by itself. The offer, landing page, tracking and follow-up still affect whether a click becomes a qualified lead or sale.

Do not treat ChatGPT Ads like Google Ads

Google Ads and Microsoft Ads train advertisers to think in keywords, search terms, match types and query reports. ChatGPT Ads are different.

OpenAI says ad selection considers signals such as conversation context, context hints, landing page, ad title and ad copy. Context hints are advertiser-provided signals about relevant topics; they are not exact-match keywords and do not guarantee delivery in specific conversations.

That means a business should not simply export a Google Ads keyword list and expect the same control.

Build ad groups around decision moments:

  • Someone comparing local providers.
  • Someone choosing between service packages.
  • Someone looking for a product with constraints.
  • Someone asking what to check before hiring a supplier.
  • Someone trying to solve a business problem and considering paid help.

Use a keyword-led search campaign when keyword control and query reports matter most. Consider ChatGPT Ads when you can test a context-based placement and judge results through conversion tracking rather than query reports. If the landing page or conversion tracking is not ready, prepare those before spending on either channel.

CPC makes testing easier, but not risk-free

OpenAI currently lists impression-based and valid-click buying options. For cost-per-click (CPC) campaigns, OpenAI recommends starting with a maximum bid of $3 to $5 USD per click. The maximum bid is the most you are willing to bid for a valid, billable click under OpenAI’s rules; it is not a guaranteed charge per click or a forecast of average cost. OpenAI notes that ad-review crawler visits to pages with tracking parameters are not billable clicks; its guide does not describe every click-validity rule. Check the current pricing and bid guidance before launch.

A billable click can still fail to become a lead or sale. Set the test budget from your business economics, not from the bid recommendation alone.

A hypothetical example, not an observed campaign result:

  • Your average CPC is $5.
  • You spend $250 for 50 clicks.
  • Two people enquire.
  • One becomes a paying customer.

That would be $125 per enquiry and $250 per customer before delivery costs. Whether it works depends on your margin, sales process and lead quality. A high-value service can support different economics from a low-margin product. The featured image’s R4.25 display is also illustrative; it is not OpenAI’s recommended bid, an exchange conversion or observed campaign data.

Check budget pacing before setting a limit. OpenAI says a daily budget is an average over seven days, not a same-day cap. Spend on an individual day can reach twice the daily amount, but cannot exceed seven times that amount over the applicable seven-day period. A campaign-total budget caps campaign spend. The payment threshold is different: it triggers a charge to the saved payment method as accrued costs reach that amount; it is not an account-level spend cap and may be reached more than once. See OpenAI’s billing guidance and choose a campaign budget your cash flow can absorb.

Click price alone does not show whether the channel works. Compare spend with qualified enquiries, sales and margin.

Pausing may not stop delivery immediately

OpenAI says ads may continue delivering for up to 24 hours after a campaign is paused, and spend reports may change after delivery stops. A pause does not automatically refund spend already delivered. Check the current billing guidance before relying on a pause as an immediate spending cutoff.

Measurement must be ready before the first campaign

OpenAI’s Help Center says Ads Manager Beta reports impressions, clicks, spend, click-through rate (CTR: clicks divided by impressions), average CPC, average cost per thousand impressions (CPM) and conversions when conversion measurement is set up. Impressions, clicks and CTR usually refresh within about 15 minutes; spend can take 7 to 8 hours to appear. A temporary zero-spend figure does not mean no charges have accrued. See OpenAI’s billing and reporting guidance.

OpenAI also supports static UTM parameters (tags added to a landing-page URL to identify campaign traffic in analytics). Use them to identify ChatGPT Ads visits in your existing analytics.

OpenAI’s Conversions API sends server-side events (conversion records sent directly from your website server) and complements the browser Measurement Pixel (code that records eligible conversion events in a visitor’s browser). A small business can start with UTM tracking and a tested conversion event; add server-side tracking if its setup and volume justify the added work.

The launch checklist above covers the page, tracking, offer, budget and campaign controls to confirm before publishing. Without analytics and a tested conversion event, you may see clicks and spend but have limited evidence for attributing enquiries or sales.

What advertisers can see

Advertisers do not receive individual ChatGPT conversations. OpenAI reports aggregate campaign results, such as views and clicks. Ad selection may use the current conversation and, if personalization is enabled, selected signals from a user’s broader ChatGPT experience. Those signals remain within ChatGPT; advertisers do not see the user’s chats, history, memories, name, email or precise location. OpenAI’s ads FAQ explains the current privacy controls and data boundaries.

You may know that a campaign drove leads. You should not expect a report of the exact conversation that preceded an ad.

That changes the optimisation process. Instead of mining search terms every day, you will need to judge:

  • Which ad groups produce qualified enquiries.
  • Which landing pages convert.
  • Which offers attract the wrong type of lead.
  • Which locations or product categories deserve more budget.
  • Whether ChatGPT Ads improves total business results or merely shifts clicks from other channels.

Product feeds matter for ecommerce

For ecommerce businesses, product feeds can connect a catalog to product ads. They require a separate data-feed setup and technical work.

OpenAI’s product-feed documentation requires an ad account with feed access, an API key, a catalog in its product schema (the required fields and format for product data), publicly accessible product pages and images, and an SFTP (Secure File Transfer Protocol) upload. Product-feed campaigns cannot currently be created through Ads Manager’s CSV bulk upload. Confirm feed access and technical support before budgeting for this route.

A feed carries product titles, descriptions, prices, availability, images and destination URLs. OpenAI can use eligible catalog items in a product-ad template, subject to campaign, product and serving requirements.

Store campaigns can use product questions with specific constraints, for example:

  • “What running shoes suit wide feet?”
  • “Which laptop is good for design work under this budget?”
  • “What gift should I buy for a client?”
  • “Which skincare product suits this concern?”

If your catalog is incomplete, outdated or incorrectly formatted, products may be missing or ineligible. Feed quality can affect which products are available to the campaign.

For service businesses, the equivalent is not a feed. It is clear service pages, pricing context, proof, FAQs and useful comparison content. A vague landing page gives people less information to decide whether the service fits.

Start with a small decision-led test

Keep the first test narrow.

Do not launch every service, every location and every audience at once. Choose one offer where you already know the economics.

For example:

  • A website redesign audit.
  • A fixed-scope starter website.
  • A local SEO review.
  • A business email setup service.
  • A specific ecommerce product category.

Then build the campaign around one decision moment and one landing page. Use the opening checklist to confirm that the offer, page, tracking and follow-up are ready before you spend.

The article on fixing conversion problems before paying for more ads applies directly here. Paid traffic cannot fix a landing page that does not explain the offer or provide a working next step.

What beta status means for your budget

OpenAI describes Ads Manager as a beta that continues to change. Delivery, inventory, formats, bidding and measurement may evolve. A feature listed in documentation does not establish that it is enabled for every account or that it will produce profitable results.

OpenAI’s August platform update includes platform-wide and partner-reported results, but those figures do not predict performance for a South African small business. Treat your own measured campaign results as the decision evidence. Compare qualified enquiries or sales, not clicks alone, and record the offer, audience, landing page and dates so you know what the test actually measured.

Conclusion

ChatGPT Ads may suit a business whose offer, landing page, billing and measurement are ready. A controlled test can show whether the channel fits its sales economics; it cannot guarantee profitable results.