Skip to main content

    AI advertising: what the machine decides and what you still decide

    Short answer

    AI advertising is paid advertising in which the ad platforms' own machine learning does work people used to do by hand: bidding in every auction, finding likely buyers and assembling ads from your headlines, images and videos. In Google Ads that means Smart Bidding, Performance Max and AI Max; on Meta it means Advantage+. A person still owns the budget, the offer, the message and what counts as a conversion.

    By Jakub Cambor, founder of AI for Marketing, updated

    What is AI advertising?

    AI advertising is paid advertising in which the platform's machine learning makes the decisions that happen too fast or too often for a person to make well: the bid in each auction, who sees the ad, and which mix of headline, image and video each person is shown. The advertiser supplies the goal, the budget, the creative and the conversion data the system learns from.

    In Google Ads the oldest part of this is Smart Bidding, which Google describes as bid strategies that use its AI to aim for conversions or conversion value in every auction, weighing signals such as device, location, time of day, the actual search and whether the person is on a remarketing list. Performance Max goes further: one goal-based campaign that reaches YouTube, Display, Search, Discover, Gmail and Maps, built from the text, images, videos and audience signals you give it. AI Max for Search campaigns brings the same approach to search ads, matching searches beyond your keyword list, rewriting headlines and descriptions from your ads and landing pages, and sending each searcher to the page it judges most relevant.[1][2][3]

    On Meta the equivalent is Advantage+. Meta says Advantage+ audience lets its AI find the people to show your ads to, learning from past conversions, pixel data and how people responded to earlier ads, and it will show ads beyond the audience you suggest when it expects a better result. Advantage+ sales campaigns hand the audience, the placements and the budget to the same system at once.[4][5]

    Which decisions the platforms' AI makes, and which stay with you
    DecisionGoogle AdsMetaWho should own it
    The bid in each auctionSmart Bidding: target CPA, target ROAS, maximise conversions or conversion valueSet by the delivery system inside each campaignThe platform, inside a target you set
    Who sees the adAudience signals in Performance Max, broad match and AI Max search matchingAdvantage+ audience, which can reach past your suggestionsThe platform, steered by your data and exclusions
    Which creative runsPerformance Max combines your assets; AI Max rewrites textCombinations of the images, videos and copy you uploadYou write and approve what goes in; the platform mixes it
    Where the budget goesSpread across channels by Performance MaxSpread across ad sets by Advantage+ budgetYou set the total and the limits
    The offer and the messageNot the platform's jobNot the platform's jobYou
    What counts as a conversionThe conversion goals you chooseThe events your pixel and Conversions API sendYou, and it matters more than any other setting

    Do AI ads perform better?

    AI ads often do better on cost per result when the platform has clean conversion data and enough volume to learn from, but almost all the published evidence comes from the platforms themselves, so the answer that holds for your business is a controlled test in your own account.

    Here is what the platforms claim. Google says advertisers that switch on AI Max in Search campaigns typically see 14% more conversions or conversion value at a similar cost per acquisition or return on ad spend, from its internal data for 2025 and for non-retail advertisers. Meta says advertisers using Advantage+ sales campaigns saw a 9% improvement in cost per conversion on average, and that Advantage+ audience could lower cost per result by 7.2% for sales campaigns and by 9.7% for traffic, engagement and leads campaigns.[3][4][5]

    What Google and Meta say their own AI features add
    • Google AI Max: more conversions or value14%
    • Meta Advantage+ sales: better cost per conversion9%
    • Meta Advantage+ audience, leads: lower cost per result9.7%
    • Meta Advantage+ audience, sales: lower cost per result7.2%

    Each figure is the platform's own measurement of its own product, as published on its help pages in September 2026. None is independent, and none is a forecast for your account.

    Sources: Google Ads Help, About AI Max for Search campaigns; Meta Business Help Centre, About Advantage+ audience; Meta Business Help Centre, About Advantage+ sales campaigns

    The same pages say when those gains do not arrive. Google warns that AI Max will not be effective if a campaign is limited by budget. Meta says an ad set usually leaves its learning phase after about 50 results in the week after its last significant edit, and that until then results are less stable and cost per result is usually higher. A small account that changes its campaigns every few days never gives the system enough to learn from.[6][7]

    How to test an AI feature without betting the account
    1. 1

      Fix the conversion first

      Make sure the goal the AI chases is a real sale or qualified lead, checked against your store or CRM.

    2. 2

      Change one thing

      Test AI Max, Advantage+ audience or a new bid strategy on its own, never all three at once.

    3. 3

      Split the traffic

      Run it as a Google Ads experiment or a Meta A/B test. Google recommends half the budget each side.

    4. 4

      Wait for enough data

      Let the test run past the learning phase before reading it, and do not edit it midway.

    5. 5

      Judge on your numbers

      Compare cost per sale or qualified lead in your own systems, then keep the winner.

    Sources: Google Ads Help, About custom experiments; Meta Business Help Centre, About the learning phase

    At ToastPal, an AI speech-writing product, Jakub worked this way for 12 months as a freelance consultant, before he founded the company. AI wrote ad variants in volume for weekly tests across the US, UK, Australia and Canada, while every budget, bid strategy and market decision stayed with him. What that year produced in tracked revenue is shown below, with its context.[9]

    $528K

    Revenue tracked from Google Ads

    $528K over 12 months (October 2023 to October 2024), with the highest search impression share among tracked competitors (20.51%, October 2024).

    ToastPal, Google Ads across the US, UK, Australia and Canada. Delivered by Jakub as a freelance consultant before founding AfM.

    Read the case study

    Who can run AI-assisted ad campaigns for my business?

    Anyone who runs ads today is already using the platforms' AI, so the real choice is who will set it up properly and hold it to account: a marketer on your own team who knows the platforms, a paid media freelancer or agency, or a managed service that runs the ads together with the landing pages and follow-up they feed.

    Far more people search for AI ads and AI advertising in the US than in the UK, about five times as many each month.

    What people search for each month
    SearchUKUS
    ai ads1,3006,600
    ai advertising1,3006,600
    All of these2,60013,200

    Monthly Google searches for each phrase, in each country.

    Source: Google Ads monthly search volume, via DataForSEO (September 2026)

    Whoever you choose, these are the signs they will use the AI well rather than simply switch it on:

    • They start with tracking and can tell you exactly which event each campaign is optimising for, and how it matches your revenue or qualified leads.
    • They can explain in plain words which features they use, such as Performance Max, AI Max or Advantage+, which they leave off, and why.
    • They test new AI features against your current setup instead of switching everything over on the same day.
    • The ad accounts, pixels and analytics stay in your name, and they work in them with access you granted.
    • They report cost per sale or per qualified lead from your own systems, next to the platform's figures.

    Two of Jakub's freelance engagements, both before he founded the company, show the difference judgement makes on top of the platform's AI. For a global technology platform, cost per lead fell 97%, from $270 to $8.86 over about 11 months, mostly by moving the budget to Meta and testing creative written with AI help in German and English. For Mumshandmade, a handmade knitwear brand selling in more than 10 countries, Meta reported a peak weekly return on ad spend of 4.72x in week 12, and Advantage+ shopping campaigns were tested against standard prospecting in each market along the way.[11][12]

    AI for Marketing runs paid ads as part of one managed service. Jakub sets the budget, the offer and the conversion goals, the platforms' AI and our own system handle the volume, and you approve every new ad before it runs. More on specific platforms is on choosing a PPC agency and choosing a Meta ads agency.

    Proof from our own work

    • 97%

      Lower cost per lead

      97%, from $270 to $8.86 per lead over about 11 months (Q4 2024 to Q3 2025), with $110K+ of ad spend managed and 1,700+ leads.

      Global technology platform (anonymised), paid media on LinkedIn then Meta across Germany, Austria and Switzerland. Delivered by Jakub as a freelance consultant before founding AfM.

      Read the case study
    • 4.72x

      Peak weekly Meta return on ad spend

      4.72x (week 12, platform-reported), during a four-month engagement from September 2024 to January 2025.

      Mumshandmade, ecommerce, Meta ads in 10+ countries. Delivered by Jakub as a freelance consultant before founding AfM.

      Read the case study

    What we do

    We run your marketing for you and make it pay for itself.

    • First comes a free written breakdown: we look at your site and any numbers you choose to share, with no logins, and show you where the money leaks and the first thing we would fix.
    • Then we fix and run the connected commercial journey: the work that earns attention, the pages that turn it into a lead or sale, and the follow-up that creates the next valuable conversion.
    • Everything is actively managed and improved, with a weekly report showing what changed and what it made, and you approve everything before it goes out.
    • Rolling monthly. No lock-in, cancel any time.

    What it is not

    • Not a single-channel agency: we do whatever makes you money next, not ads forever or SEO forever.
    • Not a junior team on a six-month lock-in: you work with Jakub directly, rolling monthly.
    • Not AI left to run on its own: every piece is reviewed by a person and approved by you.

    We also build custom automations, such as reporting digests, lead enrichment and routing, and support triage in tools like n8n, Make and Zapier. They are an add-on to the managed service, not a separate offer.

    More questions

    Is Performance Max the same thing as AI advertising?

    It is one kind of it. Performance Max is Google's most automated campaign type, reaching every Google channel from one campaign, but Smart Bidding in an ordinary Search campaign, AI Max and Meta's Advantage+ are all AI advertising too. Most accounts use a mix, with some campaigns kept tightly controlled so you can see what is working.

    Can AI write my ads?

    It can draft them, and the platforms now do some of it themselves: Google says AI Max text customisation writes headlines and descriptions from your existing ads, landing page copy and assets. Treat everything it writes as a draft. Check every claim, keep the offer and the tone yours, and approve each ad before it goes live, because the platform will not know what your business can promise.[6]

    Does AI advertising need a big budget?

    It needs enough conversions to learn from more than it needs a big budget. Meta, for example, says an ad set usually leaves its learning phase after about 50 results in a week. If your budget cannot buy that many, optimise for an earlier step that happens more often, or keep fewer campaigns so each one gets enough data.[7]

    Every answer, by topic

    Sources

    1. 1.Google Ads Help, About Smart Biddingsupport.google.com
    2. 2.Google Ads Help, About Performance Max campaignssupport.google.com
    3. 3.Google Ads Help, About AI Max for Search campaignssupport.google.com
    4. 4.Meta Business Help Centre, About Advantage+ audiencefacebook.com
    5. 5.Meta Business Help Centre, About Advantage+ sales campaignsfacebook.com
    6. 6.Google Ads Help, How AI Max for Search campaigns workssupport.google.com
    7. 7.Meta Business Help Centre, About the learning phasefacebook.com
    8. 8.Google Ads Help, About custom experimentssupport.google.com
    9. 9.Revenue tracked from Google Ads: AI for Marketing case study
    10. 10.Google Ads monthly search volume, via DataForSEO (September 2026)dataforseo.com
    11. 11.Lower cost per lead: AI for Marketing case study
    12. 12.Peak weekly Meta return on ad spend: AI for Marketing case study