Opens in a new tab

All Stories

How to Choose a ChatGPT Ads Agency

Quick Answer

Choose a ChatGPT ads agency that is honest about how new the channel is and sets up OpenAI’s pixel or Conversions API before you spend. It should size your test budget from what a customer costs you today. A good one then judges results on blended revenue and new customers instead of the platform’s own ROAS. If an agency guarantees results on a channel this new, walk away.

If you run a US business, someone has probably pitched you ChatGPT ads by now. OpenAI only started testing ads in the US on February 9, 2026, so every agency is new to this channel (OpenAI, Axios). That makes it hard to judge whose pitch you should believe.

OpenAI says ChatGPT ads passed a $1 billion annualized revenue run rate in less than 200 days. The company also counts tens of thousands of advertisers across more than 40 countries (OpenAI, August 2026). That growth is why so many agencies, including plenty of paid search agencies, have added “ChatGPT ads” to their service lists almost overnight.

Testing a new channel early can pay off, because you get in before your competitors do. The same newness cuts the other way, and without the right setup the channel can turn into a money pit. Which way it goes depends mostly on your tracking and your budget, and on picking an agency that will be straight with you about both.

What ChatGPT ads actually are in 2026

ChatGPT ads are sponsored placements that appear below ChatGPT’s answers, clearly labeled and kept separate from the answer itself (OpenAI, OpenAI Help Center). They show only to users on the Free and Go plans, while Plus, Pro, Business, Enterprise, and Education accounts stay ad-free (OpenAI, OpenAI Europe update).

According to OpenAI’s Help Center, each ad carries your business name and logo, a headline, a short description, an image, and a landing page. ChatGPT picks which ad to show by reading the topic and intent of the conversation along with your ad copy and landing page. You can also add “context hints” at the ad group level to guide it.

The early test sold mainly on impressions (CPM), according to Search Engine Land. In May 2026, the same outlet reported a move to cost-per-click (CPC) bidding and a self-serve Ads Manager beta in the US, which opened the channel to smaller advertisers. Today you can buy on views, clicks, or conversions in Ads Manager. OpenAI suggests starting CPC campaigns with a maximum bid of $3 to $5 per click (OpenAI Help Center).

As of September 2026, OpenAI’s own announcements list these features for advertisers:

That list is shorter than what you get in Google Ads or Meta, and in practice the targeting options are still limited. In my experience tracking is the easier part, because the setups we have completed so far were relatively easy to get running.

OpenAI also reports more than 50 technology and measurement partners (OpenAI, August 2026). Search Engine Land reported partnerships with the agency groups WPP and Publicis Groupe and with ad tech firms Criteo and Adobe (Search Engine Land). That means many agencies can buy ChatGPT ads through tools they already use.

ChatGPT ads versus getting cited in AI answers

Buyers often mix up paid ChatGPT ads, which are placements you buy in Ads Manager, with getting your brand cited inside ChatGPT’s actual answers. The second is a separate discipline, usually called generative engine optimization (GEO).

OpenAI states that “ads do not influence the answers ChatGPT gives you,” so no agency can promise that ad spend will make ChatGPT recommend you by name (OpenAI).

When someone pitches “AI advertising,” ask which of the two they mean. A generative engine optimization agency works on your content and reputation so AI tools describe you well. A ChatGPT ads agency runs paid campaigns, so its pitch should center on budgets and conversion tracking. Both can be useful, as long as you know which one you are paying for.

Is advertising on ChatGPT worth it yet?

For most businesses, a small, well-tracked test is worth running now, while a large budget commitment should wait until you have your own results. The audience is already large, with OpenAI reporting more than 1 billion weekly active users (OpenAI, August 2026). According to Adweek, OpenAI chief marketing officer Colin Fleming says 20% of them show commercial intent.

Adweek also quoted eMarketer analyst Nate Elliott, who said OpenAI can reach parity with Google and Meta on ad tech partnerships “if they want,” but that “it’ll take time.” The ad economics are still unproven because the platform is so new. Its targeting features are limited, and it has not had time to refine how it reaches the customers who are worth paying for.

My position is that it is always worth testing, as long as you follow a clear logic and have correct tracking set up. Tracking is what teaches the system what success looks like and lets you improve results over time. Without it, you are paying for clicks you cannot learn from.

ChatGPT ads are likely a good fit for your business if:

  • You can track leads or sales on your website today
  • You can fund the test budget described below for at least four weeks
  • People research your product or service before they buy

Holding off on a test makes more sense if any of these apply:

  • You have no working conversion tracking yet
  • Your budget is far below what a single customer costs you
  • You sell in a sensitive category such as health or politics, where OpenAI does not show ads (OpenAI)

What a good ChatGPT ads agency looks like

No agency has years of results on this channel, so you are judging how an agency thinks and works. I would look for the signs below, starting with the most important.

Hands-on proof, with honesty about how little of it anyone has

A credible agency can walk you through campaigns it has built in Ads Manager, including the pixel it installed and the conversion data coming back from it. It will also tell you plainly that its experience is measured in months. It is so early that this is anyone’s game, and the promises some agencies make are where you need to be wary.

Optimistic caution about what the channel can do today

ChatGPT ads are a new channel that needs a lot of testing and development. If the person you are talking to is not approaching it with optimistic caution, I would take that as a sign they do not understand the channel yet. They are probably running a cookie-cutter setup to see what sticks. In my experience, that optimistic caution is one of the most reliable signs of a real specialist.

Tracking set up before any money is spent

A good agency will insist on the OpenAI Pixel, the Conversions API, or both before launch. The pixel records standard events such as lead_created, checkout_started, and order_created on your site, and the Conversions API sends events from your server (OpenAI Developers). OpenAI’s conversion-optimized campaigns require one of the two, plus a clear target event (OpenAI Help Center).

Clear pricing that separates fees from ad spend

A transparent agency will set up the Ads Manager account in your company’s name, with ad spend billed to your own payment method. Its fee should appear as a separate line in the contract, so you can see exactly what goes to OpenAI and what goes to the agency.

Paid search and Shopping skills that carry over

At 94n we treat ChatGPT ads as an intent channel. It behaves closer to Google Shopping than to Meta prospecting, because people are asking questions while they research. That means a strong paid search marketing agency often has the right instincts already, especially around product feeds and conversion tracking. Ask how their Google Ads or Shopping experience shapes their ChatGPT plan.

An understanding of GEO

Good agencies also check how your brand shows up in ChatGPT’s answers, even if they do not sell GEO. What the answer says about you affects whether people click your ad below it.

What ChatGPT ads cost, and how to size a test budget

How to size a ChatGPT ads test budget for lead generation and ecommerce, and how long to run the test

You pay for two things: media spend, which goes to OpenAI, and the agency’s fee for its time. Media spend is billed on impressions or valid clicks, depending on the campaign type you pick. The same billing applies to conversion-optimized campaigns, which still charge per impression or click (OpenAI Help Center, OpenAI conversion campaigns guide). For the agency fee, ask whether it is a flat monthly retainer or a percentage of spend, and whether a setup fee is charged on top.

None of the OpenAI help pages I reviewed list a minimum spend. For conversion campaigns, OpenAI’s guidance is to choose settings that give the campaign “enough time and budget to gather meaningful performance data” (OpenAI Help Center). That leaves the actual number up to you, so I use the two budget rules below to set it.

Lead generation budgets

If you generate leads, B2B or B2C, set your daily budget close to what you currently pay for one converted customer. The system needs enough traffic and engagement data to find the people you want. If a customer normally costs you $1,000 and your daily budget is $10, you are very unlikely to see success.

This is true of every ad channel, and it comes down to volume. You want to reach about 30 conversions in 30 days so the system can learn and find more people like them.

Ecommerce budgets

For online stores the same logic applies, so set your daily budget to at least one to two times your target cost per order. If a first order costs $40 to acquire and your budget is $10 a day, you are asking the system to find one buyer every four days. That is too little signal for the system to learn from.

Aim for 30 to 50 purchases in the first month, and work backward from your target cost per purchase to get your minimum monthly spend. For most of our clients, that lands between $2,000 and $3,000 a month at the low end.

How long to run the test

Give the test four to six weeks, because ecommerce has a purchase lag that lead forms do not. You need the full window to see whether week-one clicks turn into week-three orders.

How to tell if ChatGPT ads are really working

Four ways to check whether ChatGPT ads are working: blended MER, new customer share, GA4 referral conversion rate and post-purchase surveys

Judging performance is harder than launching, because the platform reports its own ROAS (return on ad spend). As with every new channel, that number tends to overstate reality. Even OpenAI’s own launch materials lead with platform-reported wins, such as an ecommerce advertiser seeing “a 3x return on ad spend” over 28 days (OpenAI, August 2026). Treat numbers like that as a starting point for your own checks.

Then check these four things in your own data, which the platform cannot fake:

  • Did your blended MER (total revenue divided by total marketing spend) hold steady or improve while the campaign ran?
  • Did your share of new customers rise?
  • Does ChatGPT referral traffic in Google Analytics 4 convert at or above your other paid traffic? You can find it by filtering on session source in GA4 (Google Analytics Help).
  • Did customers start naming ChatGPT in your post-purchase survey?

If the platform says it worked and none of those moved, it was taking credit for demand you already had. If they moved in the right direction, raise the budget and run the same checks again at the higher spend level.

Red flags when an agency pitches ChatGPT ads

Walk away, or at least push hard, if you hear any of these in a pitch:

  • Claims of “years of experience” with ChatGPT ads. The US test only began in February 2026, so nobody has years (OpenAI).
  • A guaranteed ROAS or cost per lead. Nobody knows yet how this channel will perform for your business.
  • An “OpenAI certified” badge for advertising. OpenAI’s Partner Network, launched in June 2026, offers tiers and specializations in areas such as Codex, cybersecurity, and agents (OpenAI Partner Network). It does not describe any ads certification, so ask exactly what a badge covers.
  • Reporting that stops at platform ROAS. If they cannot explain how they will check blended revenue and new customers, they cannot tell you whether the ads worked.
  • A budget far below your cost per customer. A $10-a-day test for a $1,000 customer sets you up to fail.

Brand safety and disclosure rules

OpenAI labels every ad as sponsored and keeps ads away from sensitive topics like health, mental health, and politics. It also does not show ads to users who say, or are predicted to be, under 18. Advertisers get aggregate performance data only and cannot see users’ chats or personal details (OpenAI).

Your ads still have to follow US advertising law, and the FTC’s guide for small businesses says ads must be truthful and not deceptive. Advertisers also need evidence to back up their claims before the ads run. The FTC adds that ad agencies can share legal responsibility for deceptive ads they help create.

The FTC’s native advertising guidance says an ad should be “readily identifiable” as an ad. OpenAI handles the sponsored label, but your agency should still check that your landing page and claims hold up.

Questions to ask any ChatGPT ads agency

Take these questions into every sales call, and put the most weight on the first one:

  1. How will you prove it worked, beyond the platform’s own ROAS? Listen for blended MER, new-customer share, GA4 referral conversion rates, and post-purchase surveys.
  2. Can you show me a campaign you have run in ChatGPT Ads Manager? A live account tells you more than a slide deck.
  3. How will you set up the OpenAI Pixel or Conversions API, and which events will we track?
  4. How did you arrive at the budget you are proposing? The answer should start from your cost per customer or cost per order.
  5. How long will the test run, and what result would make you recommend stopping?
  6. What do you charge, and how is that separate from ad spend?
  7. Are you selling paid ChatGPT ads, GEO, or both?
  8. What from your paid search or Shopping work applies here, and what does not?
  9. Who owns the Ads Manager account and the pixel? The answer should be you.

Frequently asked questions

Can I run ChatGPT ads without an agency?

Yes, OpenAI’s self-serve Ads Manager lets you create an account and build campaigns yourself (OpenAI, August 2026, OpenAI Ads). An agency earns its fee by getting tracking right and by measuring results honestly. If you already run Google Ads well in-house, you could test it yourself first.

How much should I budget for a ChatGPT ads test?

Start from what a customer costs you today: for lead generation, set a daily budget near that cost so you can reach about 30 conversions in 30 days. For ecommerce, spend one to two times your target cost per order each day. For most of our ecommerce clients, that means at least $2,000 to $3,000 a month for four to six weeks.

Do ChatGPT ads change what ChatGPT recommends?

No, according to OpenAI, which says ads do not influence the answers ChatGPT gives and are labeled and shown separately from the answer (OpenAI). Getting cited inside answers is a separate discipline called GEO.

Is there an official OpenAI-certified ChatGPT ads agency?

Not that OpenAI has documented, since its Partner Network, launched in June 2026, lists specializations in areas like Codex, cybersecurity, and agents, but no advertising certification (OpenAI Partner Network). If an agency says it is certified, ask which program the badge comes from and what it covers.

Hire now, pilot, or wait?

A careful pilot with the right agency is a sensible bet if you can track conversions. You also need to be able to fund a test sized to your cost per customer for four to six weeks. If tracking is missing or the budget math does not work, fix those first. Either way, use the questions above to find an agency that can show you its work and explain how it will measure yours.

Tracking is the part to get right before anything else. If you want help setting up conversion tracking or a second opinion on a proposal in front of you, the team at 94n is happy to talk it through.

Alex Tkacovs

Connect On LinkedIn
Alex Tkacovs is the CEO and co-founder of 94n, a paid media and SEO agency that helps online brands grow revenue through transparent, measurable campaigns. He works alongside a team of Google Ads, Meta, and tracking specialists serving DTC, eCommerce, and lead-generation brands across fashion, beauty, tech, and home goods. His focus is on accurate measurement and real business outcomes over vanity metrics. Accounts 94n takes over see an average lift of more than 30% in key metrics, and the agency keeps over 95% of its clients.

Related Stories