HTTP 200 · buying infrastructure
Page title: “New ways to buy ChatGPT ads | OpenAI”. The response contained CPC, conversion and Ads Manager terms.
A rollout page is not a placement observation.
A restrained readiness sample for an AU-first ChatGPT Ads rollout and placement ledger: preserve official source status, account eligibility, actual observation evidence and unknowns before selling an agency brief.
Disclosure: the six placement rows are fictional/composite preflight records and every test is marked
not_run. This page reports zero observed ads and zero “no ad observed” tests. It is not an advertiser case study, campaign report or claim that any account can currently buy or receive ads.1. What the public source check established
At 20:33 AEST on 23 July 2026, a no-login script fetched four official pages. HTTP and term presence help locate evidence; they do not settle country/account eligibility or prove delivery.
HTTP 200 · measurement surface
Page title: “Ads | OpenAI Developers”. The response contained CPC and conversion terms.
HTTP 200 · country status source
The response contained Australia, New Zealand, Canada and “coming soon”. A test row must still capture the current country-specific label.
HTTP 200 · rollout context
The response contained Australia, New Zealand, Canada, Free and Go. Mention is not proof that a particular account or session is eligible.
Method: public HTTP GET, browser-like user agent, no login, ad account, prompt execution or personal conversation data. Download the timestamped machine-readable snapshot.
2. The fixture keeps “not tested” visible
Six AU preflight rows cover Free/Go and three low-regulation intent groups. Nothing has been promoted to an observation.
| Row | Country | Plan | Language | Intent | Account eligibility | Placement result | Observed evidence |
|---|---|---|---|---|---|---|---|
| FX-AU-01 | Australia | Free | en-AU | DTC apparel | Needs verification | Not run | Empty by design |
| FX-AU-02 | Australia | Free | en-AU | Home goods | Needs verification | Not run | Empty by design |
| FX-AU-03 | Australia | Free | en-AU | Productivity SaaS | Needs verification | Not run | Empty by design |
| FX-AU-04 | Australia | Go | en-AU | DTC apparel | Needs verification | Not run | Empty by design |
| FX-AU-05 | Australia | Go | en-AU | Home goods | Needs verification | Not run | Empty by design |
| FX-AU-06 | Australia | Go | en-AU | Productivity SaaS | Needs verification | Not run | Empty by design |
3. The founder-readable five-bullet review
1
Split rollout, advertiser eligibility and user-side delivery
A country on an official page does not establish that one advertiser account can buy or one eligible user saw an ad. Store these as separate dated fields.
2
Use three observation states, never one boolean
Keep not_run, no_ad_observed and ad_observed distinct. A dated no-ad outcome is not proof that inventory does not exist.
3
Require a proof packet before publishing a card
Country, plan, language, intent group, eligibility check, observed time, evidence reference, advertiser domain, format, disclosure and landing page belong together.
4
Keep the smoke test manual, adult and bounded
Start with one verified eligible Australia account, three low-regulation verticals, ten commercial-intent prompts each and two dates. Do not use an account farm, automated scraping or personal-conversation capture.
5
Sell evidence before media buying or software
Test a fixed-scope country × vertical brief with unknowns and source dates. Do not promise ROAS, placement, complete monitoring or API coverage.
4. The bounded paid scope
A$149–299 Public-source Ad Rollout Ledger Readiness Review
One public sample ledger/brief, returned in 48–72 hours with a source/status schema, observation-proof rules, unknown-field policy, one buyer-readable placement-card template, pricing/CTA edits and an approval-gated agency distribution queue.
One public sample ledger/brief, returned in 48–72 hours with a source/status schema, observation-proof rules, unknown-field policy, one buyer-readable placement-card template, pricing/CTA edits and an approval-gated agency distribution queue.
Promote gate
At least ten real evidence-backed placements replicated across two dates, plus two agency requests for the full brief or one paid brief.
Kill gate
Eligibility cannot be stably verified, observations are too sparse, collection requires automated multi-account scraping, or buyers only want free tutorials.
Boundary: public or explicitly supplied redacted evidence only. No ad-account login, campaign setup, media spend, prompt automation, personal-conversation storage, policy/legal advice, ROI/ROAS guarantee, placement guarantee or complete-market-coverage claim.