All answers

The Prevailing
Answers for
Business model:
Business model, pick up to two

Should we test ads inside AI assistants as they launch?

The PrevailingUpdated Nov 9, 2026

Our answer

Test ChatGPT ads only with spare research money once your current ads are profitable. Size the test for about 30 new-customer orders and judge it on your sales against a holdout. Other assistants' ads come through your existing campaigns.

Confidence
Low
Strongest proof
1 controlled test
7 official rules, 3 datasets, 4 practitioners, 10 opinions or claims
Sources
19 sources
25 statements

New here? Take the 1-minute tour

Why this is the answer

3 reasons and our judgment

Google ads converted better than ChatGPT ads in all four published same-advertiser cases (PPC ONE, a Czech paid search agency). A working paper under review found a paid ChatGPT visit earned $0.89 in US traffic against $2.45 for Google search (Kaiser and Schulze).

You'll have to read results in your own records. OpenAI has no benchmarks six months in, clicks cost from $3 to $22 and there's no auction or competitor reporting (Search Engine Journal).

OpenAI's minimum ChatGPT ad budget buys too few orders to judge the test. StoreHero, a profit reporting software maker, uses 30 new-customer orders as the minimum read. At OpenAI's $25 daily minimum, a month buys about 6.

Our judgment

  • Buy ChatGPT as its own channel and leave the other assistants to existing campaigns. ppc.io says ChatGPT is the only big chatbot selling ads inside its answers. Google says existing campaigns enter AI Overviews and AI Mode automatically, and Microsoft and Amazon sell their assistant placements, in Microsoft Copilot and Amazon's shopping assistant, through their own ad platforms.
  • Spend only money you can treat as research, once your current ads are profitable. The published results are mixed, the same-advertiser comparisons and the working paper lean toward Google, and OpenAI has no benchmarks.
  • Size the test for about 30 new-customer orders over about 30 days, and wait if that sum doesn't fit a spare budget. This follows StoreHero's minimum read and is in line with the first-test rules in Budget for a first small-channel test and How long a small-channel test should run.
  • Judge the test on new customers and margin in your own sales records against held-out regions. Platform conversion counts have run both above and below real sales. WorkMagic reports one brand's geo holdout found 2.3 times the incremental orders click attribution showed, and Ads Manager has no auction or competitor reporting.
  • Waiting costs little if your current ads aren't yet profitable, as StoreHero says. The case for getting in early comes from firms that sell ChatGPT ads or ad data, and MarTech reports that growth to about 60,000 advertisers may mean many cheap experiments. A local business with a small budget should wait, as ppc.io advises, even though OpenAI's targeting can reach a region or metro market.

Our judgment, built from ppc.io, Google, Microsoft, Amazon, StoreHero, PPC ONE, Search Engine Journal, Opascope, Similarweb, Common Thread Collective, MarTech, OpenAI, Kaiser and Schulze (PPC Land) and WorkMagic: our reading of these sources, not a finding from any one of them.

What the sources say

25 statements from 19 sources
  • GoogleUndatedSupports

    Google says existing Search, Shopping and Performance Max campaigns can serve in AI Overviews and AI Mode automatically, with no way to target or opt out of them separately, and no extra charge.

    “you can't target ads exclusively to AI Overviews or AI Mode, or opt out of showing ads in them separately from Search.”

    Read it on support.google.com

  • OpenAIUndatedSupports

    OpenAI says ChatGPT ads can be bought by impressions, clicks or conversions; conversion bidding is in open beta, charges per valid click, and needs pixel or Conversions API tracking.

    “Treat the CPA bid as an optimization input, not as a conversion charge.”

    Read it on developers.openai.com

  • OpenAIUndatedCaveat

    OpenAI's ad reports count conversions from clicks and views within a default 30-day click and 1-day view window. Advertisers can pick 7 or 14 days, or drop views, which changes CPA and ROAS.

    “Goal conversions from clicks plus views within the selected attribution windows.”

    Read it on developers.openai.com

  • OpenAIUndatedCaveat

    OpenAI's ads target by geography, platform and custom audiences; inclusion audiences need at least 25,000 matched users, product-feed campaigns target by country only, and context hints are not exact-match keywords.

    “Inclusion and bid multipliers require at least 25,000 matched users; ready exclusion audiences have no minimum.”

    Read it on developers.openai.com

  • OpenAIUndatedSupports

    OpenAI's ads docs say location targeting lets you choose countries, regions or markets where ads deliver. A campaign with no location targeting can target all available locations.

    “Use location targeting to choose the countries, regions, or markets where your ads can deliver.”

    Read it on developers.openai.com

  • PPC LandNov 2, 2026Supports

    A working paper relayed by PPC Land found a paid ChatGPT visit earned $0.89 in US revenue against $2.45 for a paid Google search visit, and ads coincided with a 6 to 8% drop in organic ChatGPT referrals.

    “generated $0.89 in revenue, against $2.45”

    Read it on ppc.land

  • WorkMagicNov 2, 2026Caveat

    WorkMagic, which sells incrementality measurement, says a Dose geo holdout found 2.3x more incremental DTC orders than click attribution reported, 67% from new customers, plus a 34.5% halo on Amazon.

    “2.3x more incremental DTC orders than click attribution reported”

    Read it on workmagic.io

  • Search Engine RoundtableNov 2, 2026Caveat

    OpenAI's weekly advertiser email, per Search Engine Roundtable, says ChatGPT ads added 26 countries, a wider standard ad unit, multi-image ads in open beta and an international Shopify integration. No performance data.

    “Multi-image ads are now available in open beta to all advertisers”

    Read it on seroundtable.com

  • PPC LandOct 30, 2026Caveat

    PPC Land reports ChatGPT ads run in 89 countries for 100,000+ advertisers on Free and Go plans, but early delivery was thin: one advertiser spent $2,500 of a $250,000 commitment in four weeks.

    “Less than 10% of the eligible base was part of the pilot”

    Read it on ppc.land

  • PPC LandOct 29, 2026Caveat

    PPC Land reports StackAdapt lets advertisers who buy ChatGPT ads through it retarget people who saw them across its CTV and display inventory. No markets, prices or match rates are given.

    “advertisers buying ChatGPT ads through its platform can retarget people who merely viewed those ads across its connected TV and display inventory.”

    Read it on ppc.land

  • MarTechOct 21, 2026Caveat

    MarTech reports, citing New Street Research, that ChatGPT advertisers rose from about 20,000 in late August to roughly 60,000 in early October 2026, with clicks about $5 against $30 to $50 on Google.

    “Sixty thousand advertisers can mean strong repeat demand, or it can mean 60,000 companies running inexpensive experiments.”

    Read it on martech.org

  • ppc.ioOct 17, 2026Supports

    ppc.io says ChatGPT is the only big AI chatbot selling ads, and one agency's test got about 0.6% CTR versus about 2% on search. It advises a fixed 30-day test against a region with no ChatGPT ads.

    “The good numbers all come from OpenAI or its measurement partners. The bad one comes from an agency that ran its own test.”

    Read it on ppc.io

  • StoreHeroOct 15, 2026Supports

    StoreHero, which sells profit software, says a ChatGPT ads test needs about 30 new-customer orders, roughly $3,000 over 30 days, so the $25 daily minimum gives only a look. Its rules are its own, not OpenAI's.

    “The $25 daily minimum only buys $750 a month, which at $3 a click and 2.5% is about 6 orders: a look, not a read.”

    Read it on storehero.ai

  • Operators (podcast)Oct 8, 2026Supports

    Jessica Schinazi says her team is currently testing ads on ChatGPT to learn how the platform works for customer acquisition.

    “we're also actually doing ads now on Chai GPT, for example. So we're just testing those things.”

    Read it on podcasters.spotify.com

  • Barn2Oct 6, 2026Supports

    Barn2, a small software seller, spent $477 on ChatGPT ads over six weeks: 370 clicks at a 1.93% click rate and zero sales, where its organic ChatGPT conversion rate pointed to about 5.

    Read it on barn2.com

  • PPC ONESep 29, 2026Supports

    PPC ONE, a search agency, reviewed 49 published ChatGPT ad cases: half of 24 campaigns had CTR under 1.25%, and in all four same-advertiser comparisons ChatGPT converted worse than Google Ads.

    “I wouldn’t move budget away from search that already works.”

    Read it on ppcone.cz

  • Search Engine JournalSep 8, 2026Supports

    Search Engine Journal says six months in, ChatGPT ad CPCs range from $3 to $22 and OpenAI has no benchmarks or competitive reports, so set your own bar and measure outside Ads Manager.

    “OpenAI says it does not yet have performance benchmarks across advertisers, industries or campaign types.”

    Read it on searchenginejournal.com

  • MicrosoftAug 27, 2026Caveat

    Microsoft claims ads in Copilot drive 101% higher engagement than traditional Search ads, and Copilot paths saw 53% more purchases within 30 minutes. It gives no method beyond its own data.

    “paths with Copilot saw a 53% increase in purchases within 30 minutes of that interaction.”

    Read it on about.ads.microsoft.com

  • SE RankingAug 10, 2026Caveat

    SE Ranking found ChatGPT ads on about 26% of 50,006 US commercial prompts. About 14% of ads were off-topic, and the advertiser was cited in the answer in only 3.63% of placements.

    “In 96.37% of placements, the advertiser is not cited.”

    Read it on seranking.com

  • Common Thread Collective2026-08Disagrees

    Common Thread Collective, which sells ChatGPT ad buying, says ChatGPT ads do not replace Google or Meta, and suggests purchase as the conversion event and $500 to $1,000 a month to start. No data.

    “ChatGPT Ads is not a replacement for Google or Meta.”

    Read it on commonthreadco.com

  • ShopifyJun 17, 2026Supports

    Shopify's Shop Campaigns, which target customers at a set price across Shop and outside channels, added ChatGPT, Pinterest and open web ads in June 2026. Merchants pay only when a customer converts.

    “You only pay when the customer converts.”

    Read it on shopify.com

  • OpascopeJun 15, 2026Disagrees

    Opascope, which sells ChatGPT ad management, says its scaled client campaigns spent about $60K for about $89K revenue, a 1.49x ROAS at $1.72 a click, with choppy days and undercounted platform conversions.

    “stand up first-party tracking before you read the dashboard as truth”

    Read it on opascope.com

  • SimilarwebJun 15, 2026Disagrees

    Similarweb, which sells ad intelligence, counts active ChatGPT advertisers rising from 2,585 to 3,770 in the week to June 8, 2026, with the top five taking nearly 17% of impressions.

    “Eight of the top 10 advertisers are B2B SaaS or productivity brands.”

    Read it on similarweb.com

  • AmazonJun 11, 2026Supports

    Amazon says existing sponsored ads campaigns automatically serve in Alexa for Shopping conversations and can be optimized as AI-powered prompts, with no extra setup.

    “Existing sponsored ads campaigns are automatically eligible to be optimized as a prompt.”

    Read it on advertising.amazon.com

  • AmazonJun 11, 2026Caveat

    Amazon says its Rufus assistant had over 300 million users in 2025, and that adding prompts to a Sponsored Brands ad lifts conversions 6%. It gives no method for these figures.

    “Nearly 20% of shoppers who interact with a prompt continue the conversation about that brand, and adding prompts to a Sponsored Brands ad drives a 6% increase in conversions.”

    Read it on advertising.amazon.com

In the full answer

For readers who sign in
  • What to do this week3 steps
  • The numbers behind it3 figures from cited sources
  • Where each source standsA chart placing each source on the question
  • The strongest disagreementThe best case against the answer, from a cited source
  • How sure we areWhat its confidence rests on, and what would change the answer

Read next

More in Other emerging channels