For agencies

ChatGPT Ads for agencies: build a repeatable client launch process

Build a repeatable agency workflow for ChatGPT Ads: client access, campaign briefs, account separation, creative approval, launch checks and useful reporting.

Editorial illustration: three colleagues review materials around a meeting table in a small studio.

Adding ChatGPT Ads to an agency offering starts with a bounded client test. Agree on the offer, the budget, who approves the work and the evidence you need before recommending another month.

The main operational challenge is repeatability. One specialist can remember a client’s service area and prohibited claims. A team managing ten accounts needs those decisions written down and visible before the next person edits an ad.

Here is a practical workflow for that first client, with a clear place for AthillyAds in an agency setup.

1. Establish account ownership before campaign work

Have the client confirm their advertiser account and invite the appropriate agency users. For account setup through ChatGPT, OpenAI’s current guidance specifically excludes agencies creating accounts on a client’s behalf. Business verification and permission to serve ads remain OpenAI processes.

Record the account owner, billing contact, agency lead and approver in the client brief. Check the intended country, currency and time zone before the client confirms account setup; fixing a wrong assumption later can disrupt both reporting and budgeting.

AthillyAds access and Ads Manager access are separate. A ChatGPT workspace role also does not grant advertising permissions. OpenAI documents the distinction in Ads Manager identity and access.

2. Separate client accounts and team access deliberately

AthillyAds organizes work into an organization containing ad accounts. Use a separate ad account for each client’s brand information, campaigns, assets and reporting. Give the account a recognizable client name before onboarding it.

There is an important planning detail: organization members can access all ad accounts within that AthillyAds organization. Admin and User roles distinguish team administration; they are not a per-client permission system.

That works for an internal delivery team with a shared client roster. It also means you should review the access model before inviting clients, freelancers or a specialist who should only see one customer. Keep the approval process in the collaboration arrangement you have agreed with the client.

3. Turn the website into a verified client brief

AthillyAds reads the pages you select, asks about gaps and produces an editable Brand book. Treat that as the first reviewable deliverable.

A hypothetical heating installer might advertise “heat pumps throughout the region,” while the sales team only accepts installations within a defined service area. The site may also mention emergency repairs even though that service has been discontinued. Correct those points before generating campaigns.

Ask the client to verify five things:

  • The offer being tested and its destination page.
  • The customer situation the offer solves.
  • Geography, delivery conditions and exclusions.
  • Claims the business can substantiate.
  • The next step after a click: enquiry, booking or purchase.

This review creates a reusable brief for subsequent campaigns. It also makes disagreement concrete: “we do not offer weekend installation” is easier to act on than “the ads don’t sound like us.”

4. Agree on the test before expanding the plan

Choose one commercial question. For example: can a specific installation service attract qualified enquiries at a cost the client can support?

Define what counts as qualified. A form completion outside the service area may be a conversion in an analytics tool but an unusable lead for the client. Agree on who checks enquiries and how the result returns to the agency.

Then set the spending limit, review date and reasons to stop earlier, such as a broken booking flow or the business becoming unable to fulfil the offer. Use the ChatGPT Ads cost guide to structure the arithmetic. Avoid promising a market-wide acquisition cost before you have client-specific evidence.

5. Review campaigns in a predictable order

AthillyAds generates a campaign plan with ad groups, context hints and ad cards. Review the offer and audience situation first, then the hints, copy, image and destination as one connected message.

For a heating business, “replacing an old boiler” and “comparing heat-pump running costs” can justify different creative angles. The examples are planning hypotheses. They do not establish that one phrase will produce cheaper leads.

Use the campaign builder to inspect and edit the plan, then run pre-flight checks. Share the specific version the client needs to approve, including the campaign settings. An approved headline alone is not approval of a budget or a broader geography.

6. Choose a launch and editing workflow

The full AthillyAds app supports an Ads Manager workbook for creating campaigns and CSV update files for objects already linked to platform IDs. It also supports a direct connection through an eligible account’s own advertising API key. Select the route your client can operate and maintain.

Assign one person to coordinate changes. If a client edits in Ads Manager while the agency edits in AthillyAds, review the current state before publishing the next change. A saved export should not become an unattended source of truth.

For manual bulk changes, OpenAI recommends starting from its current export-for-edit workflow and preserving the relevant IDs. Editing campaigns in Ads Manager.

7. Report a decision, with supporting evidence

A useful client review answers four questions: what ran, what it spent, what business outcomes were observed and what you propose next. Distinguish delivery data from website analytics and from sales-team qualification. A decision-focused scorecard brings those answers together with the evidence limits and the person responsible for the next action.

AthillyAds provides account performance history and breakdowns. Pair those with consistent UTM tracking and the client’s outcome records. Do not describe an individual context hint as a winner when the evidence is only available for the group.

Track agency fees, AthillyAds licenses and media spend separately. The paid license model is based on ad accounts; consult current pricing when scoping a multi-client package. That makes the economics reviewable before the roster grows and gives each client a clear choice about continuing the test. Where a retainer covers several assignments, allocate agency fees to the specific test on a documented basis so its total cost remains reviewable.

Your next chapter

Put the ideas to work.

Give the free generator one of your pages and explore targeting ideas and ad drafts. No account needed.

No credit card needed · Ad spend is separate