Skip to content
TAGBRUT

How is AI changing Google Ads account management?

Guide6 min readUpdated October 28, 2026

In short

In Google Ads, AI now sets the bid on every search, part of the targeting, the ad copy and sometimes the landing page. The expert's job does not disappear, it moves: sending the AI an accurate signal (reliable conversions and their value), setting the guardrails (exclusions, brand, URLs), and judging results on margin rather than on the reported figure. Alongside it, AI agents prepare the recurring work; the expert decides.

In this guide

What Google already automates

FeatureWhat it decidesWhat it needs
Smart BiddingThe bid on every search, based on signals about the person and the contextReliable conversions, and their value to bid on revenue
Performance MaxWhere to serve across Google's inventory (Search, YouTube, Display, Discover, Maps) and which asset combinations to showConversion goals, creative assets, audience signals
AI Max for SearchQueries served beyond your keywords, adapted headlines and descriptions, the landing pageA conversion-based bid strategy, a budget that does not hold the campaign back
Generated assetsText and assets created from your site and your adsUp-to-date landing pages: the advertiser remains responsible for what is generated

A few dated reference points. AI Max is not a new campaign type but a layer added to Search campaigns, and Google turns it on by default for new ones. According to Google, advertisers who turn it on typically see 14% more conversions or conversion value at a similar CPA or ROAS (Google internal data, 2025, non-retail advertisers). From June 2026, "Maximize conversions with a target CPA" and "Maximize conversion value with a target ROAS" are simply called "Target CPA" and "Target ROAS".

What stays with the expert

  1. Signal quality. Smart Bidding optimizes toward the conversions you send it. A sale counted twice, a lead never qualified or a broken consent setup, and the AI learns fast, but in the wrong direction. Checking measurement comes before any setting.
  2. Value. An AI maximizing conversions treats a small quote like a large contract. Giving each conversion a value, or importing sales from your CRM, teaches it what pays.
  3. Guardrails. Brand exclusions, negative keywords, URL exclusions, review of generated assets: the AI explores, the expert sets the boundaries.
  4. Reading what is actually served. The search terms report now shows the "AI Max" match type and the source of each match. Reading it every week shows where the budget goes once keywords no longer control it.
  5. Testing before rolling out. An AI feature is judged against a baseline: a control campaign, a comparable period, the same margin metric. Without a comparison, you credit the AI with what seasonality did.
  6. Allocating on margin. A rising ROAS can hide brand sales you already had, or low-margin products. Deciding where the next euro goes remains a business decision.

Common mistakes

  • Turning on AI Max or Performance Max on broken measurement. AI amplifies the error: more volume, on a wrong signal.
  • Leaving your brand in Performance Max. Part of the reported ROAS then comes from searches for your own name that you would have won anyway.
  • Capping the budget. According to Google, AI Max is not effective when the campaign is limited by budget.
  • Judging too early, or without a control. The first weeks are a learning period; a conclusion with no comparison period is worthless.
  • No longer reading search terms. Fewer keywords to manage does not mean less control to exercise.

AI agents, outside Google

Google's AI optimizes inside Google Ads, toward Google Ads goals. AI agents do something else: they prepare the work your team redoes every week, from account data read through the official Google Ads API.

  • What they do well: sort search terms and propose negatives with the cost avoided, spot spend that does not convert, date a drop in conversions, prepare a monthly report where every figure links to its source.
  • What they should not do: write to the account without approval. At TAGBRUT, our agents write nothing without your go-ahead, every recommendation comes with its evidence, and every action is logged.
  • What they assume: accurate measurement. An agent optimizing on wrong conversions speeds up the mistake; we check measurement before connecting the first one.

What we found at two clients

A company buying luxury goods from private sellers, in France and Italy, needed to reach owners ready to sell, while sellers and buyers type almost the same words. Keywords and negatives isolated the intent to sell, with campaigns by category and by area, and AI Max was turned on to show up on new queries. From January to May, year over year: +101% conversions and cost per conversion down 26%, with investment up 49%.

Read the case study

An online business filings platform, about twenty Search campaigns, counted leads without knowing their value. Each filing type was assigned a value to bid on revenue, then AI Max was tested against standard Search and kept where it wins. In Q2 2026, revenue generated by Google Ads grew 149% year over year, with 46% more budget.

Read the case study

Frequently asked questions

Will Google's AI replace a Google Ads expert?

No, it moves the expert's work. Google's AI sets bids, part of the targeting and the copy, but it optimizes toward the conversions it is sent and within the limits it is given. Checking measurement, assigning value to conversions, setting exclusions, reading the queries served and allocating on margin remain human decisions.

Should you turn on AI Max for your Search campaigns?

Not blindly. AI Max requires a conversion-based bid strategy and a budget that does not hold the campaign back; it adds nothing if conversions are wrong. The right approach: check measurement, turn AI Max on for part of the account, compare it with standard Search over a defined period and metric, then keep it where it wins.

What should you monitor in a Performance Max campaign?

The share of brand in the results, the landing pages picked by final URL expansion, the generated assets, the split between profitable and unprofitable products, and whether reported ROAS matches actual revenue. Google reminds advertisers that they remain responsible for landing page content and all dynamically generated assets.

Can AI agents manage a Google Ads account?

They can prepare much of the recurring work: search term triage, conversion checks, wasted spend detection, reporting. They should not apply changes without human approval. At TAGBRUT, agents read the account through the official Google Ads API, deliver work ready for review, and write nothing without your approval.

Does AI work on a small budget?

It needs data. With few conversions a month, Smart Bidding learns slowly and expansion features spend before they find what works. On a small account, first make measurement reliable, focus budget on the searches that convert, and open up AI features only once the signal is strong enough.

When to call in an expert

  • You are considering AI Max or Performance Max and want to know whether your measurement can support it.
  • Reported ROAS keeps rising, but revenue does not follow.
  • You no longer know which queries trigger your ads.
  • Your team spends its weeks extracting and checking instead of deciding.

The diagnosis is free: the five most costly findings, with a euro figure, and what we would do first, in writing within 2 business days of read access to your account.

For a first check in two minutes, test your site with the tracking self-check.

Request my free audit

Sources (7)
Get my free audit
Your session, live
    Consent: Pending See details