Quick answer: A ChatGPT Ads audit checks eight areas: account and eligibility, campaign structure, ad groups and context hints, bids, creative, landing pages, measurement and reporting. The most common money leaks are vague keyword-style context hints, ad groups with only one ad, bids below OpenAI's $3–$5 CPC starting guidance, landing pages that block OpenAI's crawlers, and conversion campaigns without real conversion events arriving. Every check below links to OpenAI's own documentation. (OpenAI, OpenAI, OpenAI, OpenAI)
ChatGPT Ads gives you fewer levers than Google Ads. There are no keywords, no search terms report and no conversation data (Search Engine Land). That makes the levers you do have more important. Use this checklist on a new account before launch, then repeat it monthly. It's also a good list to send to whoever runs your campaigns.
Time-sensitive: OpenAI retires Max CPC and Max Bid on January 1, 2027. If your campaigns still use them, check #16 is urgent (OpenAI).
Part 1: Account and eligibility (checks 1–5)
| # | Check | Why it matters | Source |
|---|---|---|---|
| 1 | A billing profile exists and the payment method is valid | Missing or invalid billing blocks delivery | OpenAI |
| 2 | Business verification is approved, not just "in progress" | Verification can take up to 7 business days, and delivery may be restricted until it's approved | OpenAI |
| 3 | No recent Brand Name or Legal Name change is pending review | Changing Brand Name pauses serving until brand review is approved. Changing Legal Name restarts business review. | OpenAI |
| 4 | Your category is allowed in every country you target | An ad can pass review and still be ineligible in some countries | OpenAI |
| 5 | A new self-serve account isn't targeting countries outside its home country | Some new accounts can advertise only in their home country at first | OpenAI |
Part 2: Campaign structure (checks 6–11)
| # | Check | Why it matters | Source |
|---|---|---|---|
| 6 | Separate campaigns for meaningfully different objectives, regions or product categories | This is OpenAI's best practice, and it keeps reporting readable | OpenAI |
| 7 | Every campaign has the right objective (Views, Clicks or Conversions) | Objectives can't be changed after creation, so a wrong one means rebuilding | OpenAI |
| 8 | US-only campaigns explicitly include the United States | Excluding other countries alone doesn't restrict delivery to the US | OpenAI |
| 9 | Campaign dates include today, and the budget has allowance left this week | Daily budgets average over a Sunday–Saturday week. Using the weekly allowance early can stop delivery later in the week. | OpenAI, OpenAI |
| 10 | Daily budgets are set knowing that a single day can spend up to 2× | A daily budget is a delivery target, not a hard daily cap | OpenAI |
| 11 | Custom audiences used for inclusion have at least 25,000 matched users, and none are used for EEA or Swiss campaigns | Inclusion targeting needs 25,000+ matches. OpenAI advises avoiding custom audiences in the EEA and Switzerland while ads personalization is unavailable there. | OpenAI |
Part 3: Ad groups and context hints (checks 12–15)
This is the part most likely to need work, because it's the part of ChatGPT Ads that's least like Google Ads.
| # | Check | Why it matters | Source |
|---|---|---|---|
| 12 | Each ad group focuses on one product, service, theme or customer need | Separate ad groups are recommended when use cases need different messaging or landing pages | OpenAI |
| 13 | Context hints are specific, natural phrases covering what, who or when | OpenAI recommends one clear idea per hint "rather than a list of disconnected keywords" | OpenAI |
| 14 | No hint is written as a delivery instruction ("only show to…") | Hints can't enforce geography, schedules or exclusions. Use the product's targeting controls instead. | OpenAI |
| 15 | Hints reflect current pricing, features and availability | OpenAI asks advertisers to update hints when offers change | OpenAI |
Quick test for check 13: compare "running shoes" with OpenAI's example, "cushioned everyday running shoes for beginners training for their first 5K" (OpenAI). If your hints read like the first one, rewrite them. One agency did see keyword-style hints outperform full questions in a single test (Search Engine Land). Test both styles instead of assuming either one wins.
Part 4: Bids and bidding strategy (checks 16–19)
| # | Check | Why it matters | Source |
|---|---|---|---|
| 16 | Campaigns on legacy Max CPC or Max Bid have a migration plan to Target CPC, Target CPA or Intelligent bidding | Legacy controls retire January 1, 2027 | OpenAI |
| 17 | Click bids are in line with OpenAI's $3–$5 starting guidance, and low-bid warnings are reviewed | OpenAI recommends $3–$5 to start, and Ads Manager flags bids likely to limit delivery | OpenAI, OpenAI |
| 18 | Target CPC and Target CPA values come from recent actual results | A target below recent performance may restrict delivery | OpenAI |
| 19 | In conversion campaigns, the bid reflects the value of a conversion, not a click | OpenAI lists this as a delivery check | OpenAI |
Remember that Target CPC and Target CPA are average-cost goals. Individual clicks can cost more, and the average can land above target (OpenAI). Audit against actual averages, not the number in the settings. More on pricing in how much ChatGPT ads cost.
Part 5: Creative (checks 20–23)
| # | Check | Why it matters | Source |
|---|---|---|---|
| 20 | Every ad group has several ads, each taking a distinct angle | OpenAI recommends "a high volume and diverse set of ads" for coverage | OpenAI |
| 21 | Titles state the value clearly, and the copy adds new information | Benefit-focused copy is "more likely to be relevant to users" | OpenAI |
| 22 | No ads are stuck as "Not serving" or rejected without action | Hover to see the reason, then edit and resubmit or request review. Appeals are the fastest path. | OpenAI, OpenAI |
| 23 | No exaggerated claims, false endorsements or copy that imitates ChatGPT | Both are prohibited under the baseline ad standards | OpenAI Ad Policies |
Need copy inspiration? See our ChatGPT ads examples. For a quick pass on one ad and its landing page — character limits, policy language, reachability, the OpenAI Pixel and robots access — use the free ChatGPT Ads Checker. Bonus: if you sell visual products, consider testing the multi-image format (4–6 square images), which has been in open beta since October 26, 2026 (OpenAI).
Part 6: Landing pages (checks 24–26)
| # | Check | Why it matters | Source |
|---|---|---|---|
| 24 | Each ad links to the most relevant page, not the homepage by default | OpenAI recommends product, collection or content pages for a seamless path to action | OpenAI |
| 25 | Landing pages don't block OAI-AdsBot or OAI-SearchBot, through robots.txt, a WAF or bot protection | Blocked pages are a common rejection cause | OpenAI, OpenAI |
| 26 | The landing page offers exactly what the ad promises | Ads must be consistent end to end, and destinations can't introduce disallowed content | OpenAI Ad Policies |
Note for your analytics: OpenAI's ad-review crawlers may visit landing pages with your tracking parameters. Analytics tools may count those visits as paid traffic, but they aren't billable clicks (OpenAI).
Part 7: Measurement (checks 27–29)
| # | Check | Why it matters | Source |
|---|---|---|---|
| 27 | The OpenAI Pixel and/or Conversions API is installed, and real events are arriving at the linked data source | Setup alone doesn't prove usable signals are coming in. Weak signals can reduce delivery and leave budget unspent. | OpenAI |
| 28 | If you use both Pixel and Conversions API, each conversion shares the same event ID | Without a shared event ID, OpenAI can't deduplicate the same conversion | OpenAI |
| 29 | Conversion campaigns optimize toward a standard event with enough volume, and Event Quality Score warnings are resolved | Custom events aren't supported for conversion optimization, and rare events are hard to optimize | OpenAI, OpenAI |
Part 8: Reporting hygiene (check 30)
| # | Check | Why it matters | Source |
|---|---|---|---|
| 30 | Every ad URL has UTM parameters, and reports are compared with the same date range, timezone, filters and attribution windows | UTMs persist on ad clicks. Mismatched settings are the most common cause of "the numbers don't match." | OpenAI, OpenAI |
Also keep in mind that spend insights can lag 7–8 hours behind impressions and clicks. Don't make bid decisions on today's spend number (OpenAI).
ChatGPT ads not delivering? A 5-minute triage
Work through these in order. They follow OpenAI's troubleshooting guide (OpenAI):
- Is it too early? For a new campaign with no clear blocker, allow at least 24 hours before reporting zero delivery.
- Status. Are the campaign and ads active, review complete and today within the campaign dates? Delivering ads show "Serving."
- Account and billing. Is verification complete and the payment method valid? Was the brand name changed recently?
- Budget. Is there daily and weekly allowance left? Is a lifetime budget ahead of pace?
- Bids and goals. Are there low-bid warnings? Does the conversion setup match the campaign goal?
- Countries. Is a new account trying to deliver outside its home country?
If nothing turns up, contact ads-support@openai.com with your campaign name, launch date and a screenshot showing zero delivery (OpenAI).
How often should you audit?
Our suggestion: run the full checklist before launch, again after the first 2 weeks once you have data, and monthly after that. Also run it whenever OpenAI ships a major change, like the November 2026 bidding update. The platform has changed fast. Self-serve opened in May 2026 (OpenAI), and target bidding arrived in November (OpenAI).
Skip the spreadsheet: get this audit done for you, free
Blobbo automates the data-heavy parts of this checklist. Connect your Advertiser API key (Blobbo only makes read requests) and get a scored report covering account structure, context-hint quality, bids against OpenAI's guidance, wasted spend, creative gaps and budget pacing, with the exact change to make for each issue. Your key is used once for the audit and never stored.
Get your free ChatGPT Ads audit → or click through a sample report first.
Blobbo is an independent tool and is not affiliated with or endorsed by OpenAI.
FAQ
What should a ChatGPT Ads audit include?
Account eligibility, campaign structure, context hints, bids, creative, landing pages, conversion tracking and reporting setup. All eight areas are covered in OpenAI's Help Center guidance (OpenAI, OpenAI).
Why is my ChatGPT ad not showing?
Common causes are incomplete billing or verification, ads still in review or rejected, campaign dates or budget allowance, low bids, and new accounts targeting outside their home country (OpenAI, OpenAI).
Why was my ChatGPT ad rejected?
Usually because the ad copy or landing page doesn't comply with OpenAI's Ads Policies, or because the landing page blocks OpenAI user agents or IP addresses. You can edit and resubmit, or request a review (OpenAI).
How do I track conversions from ChatGPT ads?
Use the OpenAI Pixel, the Conversions API or both, with a shared event ID for deduplication. Add UTM parameters for your own analytics (OpenAI, OpenAI).
What changes on January 1, 2027?
Max CPC and Max Bid retire. Campaigns need to move to Target CPC, Target CPA or Intelligent bidding (OpenAI).
Sources
- OpenAI Help Center, Ads in ChatGPT: The Basics
- OpenAI Help Center, Creating campaigns for ChatGPT Ads
- OpenAI Help Center, Create Ad Groups for ChatGPT Ads
- OpenAI Help Center, Writing context hints for ChatGPT Ads
- OpenAI Help Center, Creating ads for ChatGPT Ads
- OpenAI Help Center, Bidding strategies and how pricing works in ChatGPT Ads
- OpenAI Help Center, Conversion Measurement
- OpenAI Help Center, Conversion-optimized Campaigns
- OpenAI Help Center, Troubleshooting Common Issues
- OpenAI Help Center, Troubleshooting Common Onboarding and Policy Issues
- OpenAI, Ad policies
- OpenAI, New ways to buy ChatGPT ads (May 5, 2026)
- Search Engine Land, How to run ChatGPT ads: A step-by-step guide from early campaigns (August 10, 2026)