Paid traffic · A decision before a budget

Before you test ChatGPT ads,
check what happens after the click.

You are ready to plan a test when access is confirmed, one offer is clear, the customer journey works, and you can distinguish an enquiry from a customer. A new ad format is not evidence that it will convert for your business.

By Website Traffic Editorial · Published · Sources checked October 27

Blue glass speech bubble beside a paper checklist and coral magnifying glass
Original conceptual illustration—not a product screenshot or campaign result.

What changed

A new format. The same need for evidence.

On October 5, 2026, OpenAI announced a visual ad format and expanded measurement capabilities. Its initial format test was described for a limited group of advertisers in the US, around image-generation conversations on Free and Go plans. Ads are separate from generated images and do not influence ChatGPT’s answers.

That announcement does not confirm access for your business, today’s available targeting, a minimum budget, or a likely return. Check the current official announcement and your actual account before making a plan. Do not advertise a placement you cannot buy.

Paid placement is not organic AI visibility. Buying an ad is not a route to a guaranteed recommendation in an AI answer. For the unpaid search and discovery work, use the separate search visibility plan.

Six gates · all need evidence

Could you explain why this test is ready?

Check a box only when you can point to evidence. These are editorial planning gates, not platform eligibility rules. Nothing here creates a campaign, installs tracking or grants spending approval.

0 of 6 gates checked. Resolve the missing evidence before a paid test.

This worksheet stays in this tab: it does not submit or save your answers. Avoid personal or confidential information. Print or copy your brief before leaving.

Worked example · fictional, not a forecast

A studio with enquiries, but a weak next step.

Imagine a small design studio planning to promote one landing-page review. Its contact form works, but the confirmation says only “We’ll be in touch.” No one has tested the handoff to a conversation. Buying attention now would test the offer and a broken follow-up process at the same time.

The first experiment is not a campaign. The owner defines a qualified enquiry, writes a truthful confirmation with the next step, assigns the response owner and checks the journey with a labeled test. Test submissions are excluded from business results. Only then does the owner consider a separately approved paid test.

Example measurement definitions—not observed portfolio results
StageCountDo not confuse it with
Ad clickA platform-reported clickA unique person, qualified lead or sale
EnquiryA genuine request, excluding spam and QAA newsletter signup or automated test
Qualified enquiryA request matching written service-fit criteriaEvery completed form
Conversation bookedA confirmed booking associated with the requestA booking-link click
Paid workA payment actually received and reconciledAn accepted proposal, all-site revenue or profit

If a hypothetical test spent $120 and produced three verified qualified enquiries, the observed cost per qualified enquiry would be $40. That says nothing by itself about profit. If none paid, do not report revenue. If attribution is uncertain, record “unknown” instead of assigning every sale to the campaign.

Your smallest evidence-producing test

Write the decision before the campaign.

A useful brief tells you what to do if the idea fails—not just what you hope happens. Complete this with your own evidence. There are no default budget or conversion promises.

When the answer is “not yet”

Fix the closest problem first.

  1. No confirmed access? Keep this as a preparation brief. Do not build a forecast around an unavailable channel.
  2. No clear offer? Clarify the audience and next step before testing more traffic.
  3. Enquiries already arrive but stall? Check the confirmation, reply and booking handoff before increasing spend.
  4. Tracking or consent is unresolved? Pause the paid experiment until the appropriate review is complete.
No universal winning cost or conversion rate. A small sample can be noisy, platform attribution is not causal proof, and other channels may contribute. Use the AI Growth Decision Lab to organize assumptions; do not treat a scenario as a result.

Keep planning for free

Turn a test into a useful month.

The existing Traffic Growth Pack includes a 30-day plan and marketing templates. Review the signup details on the pack page.

Get the free growth pack →

If the customer journey needs help

Scope one conversion fix.

Already getting traffic or enquiries? Waymaker’s Growth Fix focuses on one agreed journey problem. Review the current scope, price and scheduling terms before deciding; this guide does not reserve availability.

Review the Growth Fix →

Website Traffic and Waymaker Agency are related businesses. Buying a service is optional and does not provide ad-platform access.

Sources & limits

OpenAI: New ChatGPT ads format and measurement capabilities, October 5, 2026; checked October 27, 2026. This supplies the product-announcement facts, not evidence of demand or results for this site.

The checklist, definitions, example and worksheet are original editorial planning resources. No live campaign was run for this guide. Access, features and policies can change. Confirm current account details and obtain the necessary business, privacy and spending approvals before launching.