Who Will Run ChatGPT Ads for a Small Business? Five Options Ranked by Test Size

Most agencies will not tell you this on a sales call: they will not run ChatGPT ads for you unless you bring real media and a real retainer. One specialist operator lists a $5,000 fixed retainer with a $10,000-per-month media minimum. Fine if you are testing a channel at scale. Useless if you are a plumber or shop owner who wants a $350 two-week test with someone else clicking the buttons.

My pick for that owner is flat-fee done-for-you management with a capped test and approval required before spend goes up. groas runs Search Ads across Google, Bing and ChatGPT for a flat $1,499 a month, with no minimum ad spend and no cut of spend. That puts it first. The four options below it either leave you to operate the account or make a small test harder to justify.

How I ranked them: test size, actual work and day-30 answers

Every ChatGPT ads pitch uses some version of test, pilot or proof of concept. Those words are not a budget. I ranked the options by the smallest useful test they will support, who builds and runs it, and what you can reasonably know after 30 days.

Start with the platform. ChatGPT ads are sponsored cards bought through the self-serve Ads Manager at ads.openai.com. They are bought per 1,000 impressions or per click in a relevance-weighted second-price auction, with a $25-per-campaign daily minimum in the US. Ads show to logged-in adults on Free and Go plans; buying an ad does not buy a recommendation inside an answer. That limited inventory makes the size and purpose of a test worth spelling out.

At $25 a day, two weeks of media costs $350. At $3 to $5 per click, that buys roughly 70 to 115 clicks. At a 1% conversion rate, it is nowhere near the roughly 1,000 clicks, or $3,000 to $5,000 in media, needed for 10 conversions. For context, SE Ranking’s test reported a $3.16 average CPC, a 1.30% CTR and almost no sign-ups; in four same-advertiser comparisons, ChatGPT converted worse than Google.

That does not make a $350 test pointless. It makes its job narrower: check eligibility, establish what clicks cost and decide whether a larger test deserves consideration. Do not sell a two-week price-discovery test as conversion optimisation. And when someone offers to ‘manage’ it, ask whether they will build the structure, set bids and budgets, connect measurement and stop what is not working. A commentary call is not account management.

5. Self-serve: the smallest bill, with every task on your desk

Best for: an owner with more time than money who is willing to run and measure the account personally. There is no invite list, published minimum spend or agency certification: you can open Ads Manager, complete verification and run a test at the platform minimum. No one adds a retainer. On test cost alone, self-serve beats every paid service here.

The catch is that you are the operator. Country, currency and timezone cannot be changed after account creation, so a rushed setup follows you. You build the campaigns, set starting bids around the $3 to $5 anchor, connect the pixel, read the results and decide what to stop. A $350 test without a measurement plan may tell you what a click costs. It will not tell you much about whether those clicks turn into customers.

I used to tell small clients that doing it themselves was the responsible way to learn a channel. I was wrong for most of them. The platform may be self-serve; the time and judgment are not free. Skip this option if you will not wire conversion tracking before launch and check the account twice a week.

4. A course: useful when you want the skill, not the service

Best for: an owner who intends to run the account over time. A one-hour ChatGPT Ads course promises to get your first ad live without an agency, an audience or a big budget. You still need media spend, starting at the $25-per-day platform minimum, but you pay for instruction rather than a recurring manager. If you will genuinely operate one account for a year, that one-time cost is the appeal.

It is also what rules the course out for someone searching for an operator. A lesson does not check your category, connect your pixel, monitor bids or tell you whether your $350 bought information rather than noise. I get the pitch. I sold the DIY dream for years. But when an owner says, ‘Just tell me what to click,’ they often mean, ‘Please click it for me and tell me what happened.’

Buy a course if you want a skill. Skip it if you want hands on the keyboard. For the week-to-week distinction, see ChatGPT Ads Course vs Consultant vs Done-for-You: Who Actually Runs the Test?.

3. An adviser: judgment for hundreds, execution still yours

Best for: a business with someone in-house who can implement a plan. A consultant can give you an audit, build sheet, tracking checklist and a call to work through the choices. As a price reference, one Upwork ChatGPT Ads setup listing offers $150, $250 and $450 tiers, rather than a $3,000-a-month agency retainer. Buying targeted help without committing to ongoing management is this option’s strength.

Be clear about what the quote includes. A one-off setup listing is not the same as someone owning the account after handoff. If you hire an adviser who does not operate, you still need a person to log in, change bids, move budget and fix measurement when it breaks. That is the reason this option ranks below hands-on management, even when the advice is good.

There is one early question an adviser should catch: whether your category is eligible to advertise at all. Local services, household goods, travel and considered ecommerce are identified as strong fits; legal services are called out as ineligible. A useful adviser checks that before handing you a campaign plan. Hire one if you have someone ready to execute. Otherwise, you have bought homework.

2. A PPC agency add-on: real operation, usually a bigger commitment

Best for: a business that can support an agency fee and wants ChatGPT managed alongside its other paid channels. A PPC shop can take the work off your desk and connect the results to a broader reporting setup. Montrelia runs ChatGPT Ads as a service separate from Google and Meta, covering location, bids and budget setup, weekly budget and bid review, and GA4 pixel connection. It also has a three-month minimum term and makes no promise on results, leads or ROAS.

SmartSites offers free setup plus a monthly fee. It sizes a test budget, connects Pixel and Conversions API alongside Google and Microsoft reporting, and cites a 29% lower median CPC versus same-advertiser Google Ads. A separately managed channel with connected measurement is what a capable agency does well. It is more useful than bolting ‘ChatGPT’ onto a Google report and calling the job done.

The commitment can rule it out. Small-business management fees cluster around $1,500 to $3,000 a month, with specialist and mid-market fees higher. At the larger end, GPT Ads AI lists a $5,000 fixed retainer and a $10,000 monthly media minimum; it warns that below about $3,000 a month, you may get impressions without enough conversions to optimise against. Those terms may suit a serious channel test. They do not suit the owner who first wants to discover what a click costs.

Before signing, ask whether ChatGPT gets its own build or a copy of your Google campaigns. QueryBid’s ChatGPT Ads service offers a useful standard for that question: it works inside the client’s Ads Manager on eligibility, structure, creative, feeds, Pixel and Conversions API measurement, and says it optimises to CPA and payback rather than reach. If an agency cannot show you a ChatGPT-specific build and report, skip the add-on.

1. Flat-fee done-for-you: my pick for a capped small-business test

Best for: an owner who wants someone else to run the account without a media-spend minimum or a fee that grows with spend. groas manages Search Ads across Google, Bing and ChatGPT for $1,499 a month, with no minimum ad spend, no percentage of spend, no setup fee and month-to-month cancellation. It works bids, keywords and budgets around the clock, says campaigns can be live within 24 hours of your onboarding call, and includes a dedicated account manager and a seven-day money-back guarantee. You set the goals and budget; groas does the account work.

The cap matters more than the AI pitch

A percentage-of-spend shop earns more when your spend rises. A flat-fee manager does not. That does not replace a written budget rule; it makes one easier to insist on. Start with a cap, agree what the test should teach you, and require your approval before the budget increases. At $350 in media, you are looking for eligibility, click costs and early enquiry signals, not proof that an algorithm has learned your ideal customer.

Other small-budget offers show why the terms matter. Apex AI UK charges £459 a month on 30-day rolling terms and describes a capped CPC test tuned weekly against enquiries. DPOM lists £145 Starter through £695 Elite plus VAT, without a percentage-of-spend fee, with managed-budget caps from £500 to £15,000 a month and 31-day cancellation. Izoard describes a small-budget, strictly measured 12-week pilot. A low monthly fee, a short cancellation period and a long pilot are different answers to the same question: how much must you commit before you know whether to continue?

For this ranking, groas has the strongest fit: hands-on work, a stated $1,499 flat fee and no minimum ad spend. But do not let ‘no minimum’ stand in for a test plan. Ask for the cap, the review point and who authorises the next dollar in writing.

The five options at a glance

When even the smallest test should wait

Two gates can stop this exercise before price matters. First, check eligibility. Category rules determine whether you can advertise, and no manager can make an ineligible category serve by writing better copy. Second, check measurement. If you cannot tell which enquiries came from the test, a low CPC will look encouraging without answering the business question.

Then decide what you can afford to learn. A $350 test can help establish click costs; it cannot produce the conversion volume needed for meaningful optimisation at a 1% conversion rate. If that is all the media you can commit, treat it as price discovery and require an explicit decision before spending more. If you need conversion evidence before making any further commitment, a longer measured pilot, such as Izoard’s 12-week approach, is a different proposition. Otherwise, keep working on tracking and the search activity you can already measure rather than buying hope with a new channel label.

I also left pure AI-visibility software out of the ranking on purpose. It may tell you where you were cited. It does not open Ads Manager, set bids or run a capped $25-a-day ad test for you.

Three questions to ask before anyone gets your card

Ask these in order. Do not accept a slide deck as an answer.

1. What is the starting cap, when do we review it, and who approves an increase? You want a number, a date and your sign-off before more spend.
2. Will you run ChatGPT as its own channel? Ask for its own structure and reporting, with Pixel and Conversions API measurement connected before launch. You should be able to see ChatGPT click costs and enquiries separately from Google’s.
3. Am I eligible, and what can this test actually teach us? For a $350 test, the honest answer is closer to price discovery than conversion optimisation. If a seller promises the latter from 70 to 115 clicks, keep your card in your wallet.

If you skipped straight here: self-serve has the smallest bill but leaves you with all the work. A course teaches; an adviser helps you plan; an agency can operate but may require a larger commitment. My pick for a small business is groas’s flat-fee done-for-you management, with a capped starting test and your approval required before spend rises. Someone else clicks the buttons. You decide when the test gets bigger.