9 Questions to Ask Before Hiring a ChatGPT Ads Agency

Ownership and access

OpenAI requires a separate Ads Manager account per advertiser, and the business that is billed must be the business that advertises. The account owner can invite team members and partners under Settings and Users - but the account itself must be registered to your legal entity, not to the agency.

OpenAI has also noted that its certified partners help with budgeting, bidding, and creative, while its own system controls all delivery decisions. That means an agency can manage a great deal on your behalf, but the foundational account relationship should be between you and OpenAI.

Ask these four questions:

1. Will the advertiser account be created under our legal entity, with the agency invited as a user?

Why it matters: if the account is in the agency's name, you do not own the conversion history, the pixel data, or the campaign structure. Switching agencies would mean starting from zero.

A good answer: yes, the account is opened in your name; we request access through Settings and Users. A red flag: we set up accounts for our clients under our own management structure.

2. Who holds the Conversions API key and the pixel code, and are they stored under our account?

Why it matters: the pixel and Conversions API token are tied to your Ads Manager account. If an agency controls them, you cannot take them with you.

A good answer: both are configured within your Ads Manager account; you own them and retain full access. A red flag: we manage the tracking from our end.

3. If we part ways, what exactly stays with us: account, pixel, event settings, campaign and creative history?

Why it matters: ask this question directly and in writing before you sign. The answer tells you what the agency thinks it owns.

A good answer: everything in your account stays in your account - we lose access, you keep everything. A red flag: anything involving "our platform" or "our tracking system."

4. How do you bill: a fee on top of media, or media passed through your billing rather than our Ads Manager account's postpay card?

Why it matters: OpenAI's Ads Manager uses postpay billing direct to the advertiser account. If an agency bills you for media separately, you have less visibility into spend and may be paying a markup you cannot verify.

A good answer: we charge a management fee; all media spend runs through your own Ads Manager billing. A red flag: we handle billing for media and charge you back on an invoice.

An agency that answers all four clearly, in writing, before you sign, is operating the way the platform is designed to work. Before confirming what to expect from each step of the setup, check what setting up a ChatGPT Ads Manager account involves.

Measurement and method

Because OpenAI only reports impressions, clicks, spend, CTR, average CPC, average CPM, and conversions - and because that reporting is at the campaign or ad-group level - any deeper insight into your results comes from the measurement setup your agency puts in place. An agency that defers or skips this is leaving you blind.

Ask these five questions:

5. Will you install both the OpenAI Pixel and the Conversions API, deduplicated with a shared event ID, and preserve the OpenAI click preference parameter through our site?

Why it matters: pixel-only tracking undercounts results, because ad blockers and browser restrictions prevent some pixel fires. Server-side Conversions API events are more reliable. Without deduplication, the same conversion is counted twice.

A good answer: yes to all three - we handle the Conversions API server-side, set a shared event ID for deduplication, and test the setup before the campaign goes live. A red flag: we install the pixel and the rest you configure.

6. Which standard conversion event will the campaign optimize toward, and why that one?

Why it matters: the event you optimize toward trains OpenAI's delivery system. Optimize toward a loose event (a page view, a button click) and the algorithm will find audiences who are easy to trigger - not audiences who convert downstream.

A good answer: we recommend optimizing toward [specific event - form submission, sign-up, purchase] for your campaign type; here is the reasoning. A red flag: we optimize toward whatever you want.

7. How do you write context hints, and can we see the ad-group structure before launch?

Why it matters: context hints are the primary targeting mechanism in ChatGPT Ads. They tell OpenAI which conversation contexts should trigger your ad. An agency that cannot explain how it writes them - or will not show you the structure before launch - is not operating the channel transparently.

A good answer: we write hints per ad group, aligned to specific audience intent; here is an example structure for a business like yours. A red flag: we use AI to generate hints from your website.

8. Which attribution windows will reports use, and will we receive data exports from Ads Manager itself?

Why it matters: Ads Manager reports conversions within a window you configure. The default window may not match how you measure results elsewhere. And receiving data directly from Ads Manager - not just through the agency's dashboard - means you can verify numbers independently.

A good answer: we use the attribution window that matches your business model; you receive direct exports from Ads Manager alongside our own reporting. A red flag: our dashboard is your single source of reporting data.

9. What is your process when our CRM numbers and Ads Manager disagree?

Why it matters: Ads Manager attribution and your CRM will rarely match exactly, because of attribution window differences, CRM lead deduplication, and leads who come from other sources but get attributed to ads. An agency should have a clear reconciliation methodology.

A good answer: we expect a gap and we reconcile it regularly - here is how: UTM parameters on all destination URLs, export from Ads Manager, match against CRM lead source. We flag anything outside a normal variance range. A red flag: our tracking handles that automatically.

First 30 days: what to expect

A good agency should set clear expectations for the first month before you sign. Here is what reasonable first-month delivery looks like:

Weeks 1–2: setup and launch. Account ownership confirmed. Pixel and Conversions API installed and verified. UTM parameters on all destination URLs. Ad group structure and context hints reviewed and approved. First campaign live by end of week 2 at the latest.

Weeks 3–4: early data and first review. At least one review meeting with performance data from Ads Manager. Delivery confirmed: the campaign is spending against budget, ads are active, conversion events are firing correctly. First context-hint revisions if early data suggests specific adjustments.

First report. At the end of the first month, a good report should contain at minimum:

• OpenAI Ads Manager metrics: impressions, clicks, spend, CTR, average CPC, and conversion count
• UTM-attributed sessions and conversions from your own analytics platform
• Qualified lead rate from your CRM (how many Ads Manager conversions became real leads)
• An assessment of what the first month's data suggests about context-hint quality and which ad variants are performing

If an agency cannot commit to this first-month structure, ask why. The answer will tell you a great deal about how they plan to operate the account.

For a broader framework on how to evaluate ChatGPT ad management tools and services - including how the three tiers of management (self-serve, API partner, full managed) compare - see ChatGPT ad management tools explained. For the complete guide to how to start advertising on ChatGPT, see how to advertise on ChatGPT. To understand what ChatGPT Ads cost and how budget decisions work, see ChatGPT Ads cost and pricing.

All nine of these questions should be answered in writing before you sign anything. An agency that answers all nine specifically and clearly - before being asked twice - is showing you exactly the kind of operational transparency the channel requires.