Google Ads AI does not invent your business objective. You choose the campaign goal, conversion actions and, when appropriate, a target CPA or ROAS. Google’s systems then use those instructions together with campaign settings, account information, creative assets, optional feeds, audience signals and live auction context to predict which opportunities are most likely to produce the selected outcome.
The exact data depends on the campaign type and features you enable. Performance Max and AI Max for Search use overlapping information, but they automate different parts of campaign creation and delivery. Google publishes examples of signals and inputs, not a complete inventory of every model feature, so the documented list should be treated as representative rather than exhaustive.
As an Amazon Associate I earn from qualifying purchases.
The data categories Google Ads AI uses
Google Ads Help describes several classes of information that can be supplied to, or learned by, its advertising systems. Some are explicit settings you enter; others are contextual observations used at auction time.
| Data category | Examples | What the system uses it for | Where it applies |
|---|---|---|---|
| Goals and measurement | Campaign objective, conversion actions, conversion values, optional target CPA or ROAS | Defines the outcome to optimize and the value assigned to outcomes | Performance Max and Smart Bidding; comparable conversion settings in other campaigns |
| Business and account information | Business name, website URL, linked YouTube channel, Google Business Profile, app or Merchant Center account | Business understanding, recommendations, prefilled content and, where enabled, campaign inputs | Setup and account-linked features; not every linked property is necessarily used in every model |
| Creative and page content | Ad text, images, videos, landing-page copy, final URLs and asset-group material | Combines supplied assets, creates variations and selects relevant text or pages | Performance Max and AI Max for Search features |
| Feeds and structured product data | Optional data feeds, such as Merchant Center information | Provides additional product, service or business details for eligible campaigns | Feature- and campaign-dependent, especially Performance Max |
| Audience guidance | Customer, visitor, app-user, video-viewer and Firebase lists; custom segments based on interests, keywords, URLs or apps | Helps discover likely customers without necessarily restricting delivery to those lists | Performance Max audience signals and related automation |
| Auction and performance context | Query, device, location, time, language, browser, operating system, audience membership, creative, search partner and app ratings or reviews | Estimates conversion probability or value and adjusts bids for each opportunity | Smart Bidding and other automated bidding systems |
Google’s documentation also says signals can interact. A bid may reflect the combination of a query, device, location and prior conversion patterns rather than one isolated demographic or keyword.
#1 Best Overall
You define the goal; AI estimates how to reach it
Campaign objectives and conversion actions
When setting up a campaign, you select the business outcome that matters, such as sales, leads or another available objective. You then identify the conversion actions that represent success. For Performance Max, Google says advertisers can define conversion actions and optionally add a target CPA or target ROAS. These settings tell the system what to optimize; they do not come from an independent decision by the AI about what your company should value. See Google’s Performance Max setup guide.
Conversion tracking quality therefore matters as much as the creative. If a form submission is counted as a conversion, the bidding system learns toward more form submissions. If purchases carry reliable values, it can distinguish a high-value sale from a low-value one. Incorrect, duplicated or incomplete actions teach the model the wrong outcome even when the ads look polished.
How Smart Bidding turns that definition into bids
Smart Bidding uses conversion data to estimate the likelihood and, where values are available, the expected value of a conversion. Google describes query-level learning across ad groups and campaigns. When an account uses cross-account conversion tracking, manager-account query-level data may also contribute. The system is still predicting performance against your selected measurement definition; it is not selecting a new business objective.
A target CPA or ROAS is a control, not a guarantee. The model may bid more aggressively when the predicted probability or value supports the target and less when the context looks weaker. Actual results can vary with budget, competition, conversion delays, tracking changes and inventory.
Business and account information can supply context
Google’s getting-started material says setup may request a business name and website URL, with potentially linked information such as a Google Business Profile. It describes these fields as optional and says they can help Google learn about the business, make recommendations and prefill content. The same guide lists optional links to a YouTube channel, Google Business Profile, mobile app or Merchant Center account.
Optional does not mean irrelevant when you choose to provide the information, but it also does not mean every connected account is always read as a model feature for every campaign. Treat linked properties as available sources for eligible setup and asset workflows, not as a promise that every item will affect delivery.
Assets, landing pages and feeds power creative automation
Performance Max asset inputs
Performance Max can use supplied text, images and videos together with campaign and account settings, goals, audience signals and optional data feeds. Google says its generative tools can combine creative assets and generate images and videos. The generated material is therefore grounded in the information and assets you provide, subject to the campaign’s available formats and controls. The Performance Max overview explains this asset-based approach.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Good source material gives the system more accurate claims, brand language and product details to work with. It does not remove the need to inspect generated combinations for factual errors, unsuitable promises, duplicate messages or an image that represents the offer incorrectly.
AI Max for Search text customization
AI Max for Search is a Search campaign feature set rather than a separate cross-channel campaign type. Its text customization can use existing ad text, landing-page copy and assets with generative AI to produce ad wording tailored to an individual search. Google documents this behavior in its AI Max for Search documentation.
When Final URL expansion is enabled, the system can select a relevant page from your domain based on the query and ad-group theme. This makes the accuracy, accessibility and topical structure of your site part of the potential input. URL, brand and asset controls remain available so you can limit where automation can draw from.
Audience signals guide discovery, but they are not a hard fence
What you can provide
For Performance Max, audience signals can include customer lists, website-visitor lists, app-user lists, video-viewer lists and Firebase lists. You can also create custom segments from interests, keywords, URLs or apps. These signals give Google a starting point for finding people who resemble the audiences associated with your goal.
Google explicitly cautions that an audience signal is guidance rather than a strict targeting boundary. An ad may be shown beyond the supplied signal when the system predicts that doing so will help achieve the campaign’s selected goal. If you require a hard audience restriction, use a campaign or targeting control designed for that purpose instead of assuming a signal will act as a fence. The setup details are in Google’s audience signal instructions and the Performance Max FAQ.
Operational limits that affect audience signals
Google’s current instructions include requirements and timing estimates that can change as the product evolves. They state that a new signal integration can take up to two weeks, while a new or updated list may need 24–72 hours for its audience-size figure to populate or update. The instructions list more than 1,000 active and eligible users and recommend refreshing lists every 540 days.
| Operational point | Published guidance | Practical implication |
|---|---|---|
| Where signals are attached | Typically at the asset-group level | Check each relevant asset group rather than assuming one account-wide signal applies everywhere |
| Model integration | Up to two weeks for new signal integration | Do not judge a newly added signal immediately |
| List-size update | 24–72 hours for new or updated lists | Allow processing time before troubleshooting an apparently empty size estimate |
| Minimum list guidance | More than 1,000 active and eligible users | Small or inactive lists may not be usable in the same way |
| Refresh recommendation | Every 540 days | Keep membership current so an old list does not become stale |
Because these are changeable platform requirements, verify the live Google Ads Help page and your account’s current interface before relying on a number for a time-sensitive launch.
Auction-time signals shape each bid
Google’s automated-bidding documentation lists the actual search query, device, location, local time and day, list-based audience information, ad creative, interface language, browser, operating system, search network partner and app ratings or reviews among contextual inputs. The system combines these with learned conversion performance to estimate the value of an individual auction.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
- Query: The precise wording can reveal intent that a broad keyword label does not.
- Device and operating system: Performance may differ between phones, tablets and computers or between platforms.
- Location and time: Local availability, opening hours and regional demand can change the probability of conversion.
- Language, browser and network: The interface and delivery environment add context around the same search.
- Audience and creative: List membership and the particular ad variation shown can alter predicted response.
- App ratings or reviews: For eligible app-related inventory, these provide additional context.
Google does not publish a complete inventory of every internal model input. The list above should not be expanded into claims about undisclosed personal attributes or data sources.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Performance Max and AI Max for Search are different systems
Both use automation, but they do not have the same scope or controls.
| Axis | Performance Max | AI Max for Search |
|---|---|---|
| Campaign scope | One goal-based campaign can access Google inventory across channels | Feature set applied to a Search campaign |
| Advertiser inputs highlighted by Google | Goals, creative assets, audience signals, optional data feeds and account settings | Existing ads, landing-page copy, assets, URLs and search-term matching controls |
| Creative automation | Combines supplied assets; generative tools can create images and videos | Text customization creates search-tailored ad copy from existing material |
| Reach guidance | Audience signals and optional search themes guide discovery | Search-term matching uses broad-match, asset-based and landing-page-based technology |
| Controls and reporting | Asset-group material, campaign settings and reporting | Text customization, Final URL expansion, URL controls, brand controls and asset/search-term reporting |
Do not assume that a control or input documented for one feature automatically applies to the other. Choose the documentation for the campaign and automation setting you are actually using.
How to give the system useful, controllable inputs
- Define the outcome first. Select the campaign objective and identify the conversion actions that genuinely represent success.
- Audit conversion tracking. Remove duplicate actions, confirm values where appropriate and check that delayed conversions are being recorded consistently.
- Set a target only when it reflects your economics. Add a target CPA or ROAS when you have enough reliable data to make the target meaningful; avoid treating it as a guaranteed result.
- Prepare source assets. Supply accurate text, images, videos and landing pages. Make claims, prices, availability and legal disclosures consistent across the material.
- Add optional business and feed information deliberately. Link eligible properties or feeds when they contain current information that should be available to the campaign.
- Use audience signals as starting evidence. Add relevant first-party lists or custom segments, but plan for delivery beyond them.
- Review automation and reporting. Inspect generated assets, search-term or asset reporting, URL choices and campaign settings. Pause or replace material that is inaccurate or unsuitable.
What Google Ads AI does not establish from the published documentation
Google’s help pages do not disclose a complete model specification, a fixed list of every data field or a guarantee that a particular linked account will influence every decision. They also do not support the claim that AI independently chooses a company’s objective. The defensible description is narrower: advertisers provide goals and inputs; Google’s systems use documented creative, audience, account and contextual signals to predict and optimize toward those settings.
Free tools Windows power users keep installed
One-click scans. No signup required.
Feature names, defaults, eligibility, list requirements and generated-asset controls can change. Before publishing a campaign or relying on a documented limit, compare your account’s current controls with the latest Performance Max setup guide, Performance Max overview, AI Max documentation and automated-bidding documentation.
The practical answer
Google Ads AI uses your definition of success, the material and settings you supply, optional audience and feed information, and a changing set of auction and learned-performance signals. It generates or selects assets from eligible advertiser content and predicts which impressions are most likely to produce the conversion outcome you specified. The more accurate your goals, tracking, source content and landing pages are, the more meaningful those automated decisions become; audience signals guide discovery, but they do not guarantee that delivery stays inside the lists you provide.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
