The AI-era Google Ads vocabulary, explained in terms that inform the business decision. Google Ads today predicts intent and bids at the impression level — it is no longer a platform for buying keywords. Each term has a short technical definition and, where it matters, a note on what it means in practice.
The umbrella campaign that runs one AI logic across all Google channels at once: Search, Display, YouTube, Discover, Gmail, Maps. You supply assets, audience signals and conversion goals; the algorithm handles bidding, placement and creative assembly. Replaced Smart Shopping in 2022.
Maximum reach with minimal granular control — the value is in the quality of the signals and assets you feed it.
The campaign type that replaced Discovery, optimized for visual discovery and top-of-funnel demand (YouTube, Shorts, Discover, Gmail, Display). Integrates the product feed and omni-channel (online + offline) bidding.
A layer of AI features you switch on over classic Search campaigns (not a new campaign type). It extends matching beyond your set keywords, dynamically generates headlines and descriptions, and picks the optimal landing page. Out of beta in 2026; campaigns with DSA, campaign-level broad match and automatically created assets auto-upgrade to AI Max.
It is the mechanism that redefines keywords. For many accounts the transition is no longer optional — the question is who controls it.
The AI Max feature that extends ad matching toward new, high-performing queries you did not set explicitly. A broad-match + keywordless technology.
The feature where AI automatically picks the most relevant landing page for each search, instead of sending all traffic to a fixed URL.
Dynamic generation of headlines and descriptions from your page content (formerly “automatically created assets”).
Targeting based on intent inferred from context and signals, not on matching a keyword. It underpins eligibility in conversational surfaces (AI Overviews, AI Mode).
The widest match type, which in the AI era no longer means “vague match” but machine-learning interpretation of intent. Google positions it as a broad theme, not an exact word.
The organizing unit in PMax: a set of headlines, descriptions, images and video that AI combines dynamically for each placement.
Hints you give the algorithm about what a good customer looks like (e.g. buyer lists). They are not a targeting restriction but a suggestion — the system looks for similar users, it is not limited to your list.
The classic granular structure with a single keyword per ad group. In the AI era it has generally become counterproductive: the algorithm needs consolidated structures and data volume, not fragmentation.
The informal name for the AI Max + Performance Max + Demand Gen combination, conceived as a single system covering the whole funnel.
The set of automated strategies that use Google's AI to optimize for conversions or conversion value.
Adjusting the bid at the level of each impression, in milliseconds, based on hundreds of signals (context, device, time, history, intent). No human can replicate this computation at that speed and volume.
A strategy that maximizes the number of conversions within budget. It optimizes for volume, regardless of each conversion's value.
Target cost-per-action. A variant of maximize conversions with a target cost per conversion. Still volume-driven, but with an efficiency constraint.
A strategy that maximizes total value generated within budget, rather than the number of conversions.
Target return on ad spend. You tell the system how much value you want back per unit spent, and the AI bids aggressively on searches with high predicted value and pulls back on weak ones. Mechanically: max bid = predicted conversion value ÷ ROAS target.
The family of strategies (Maximize Conversion Value with or without Target ROAS) that optimize for value brought to the business, not volume. Requires at least two different reported conversion values and a minimum of data.
It is the key strategic shift of the AI era — from “how many conversions” to “how much profit”. A 20%-margin sale and a 60%-margin sale are no longer treated the same.
A strategy applied across several campaigns that lets budget flow dynamically toward the best-returning opportunities, instead of locking it in per-campaign silos.
The conversion value the AI estimates for a given user/search before bidding. The basis of the bid calculation.
The window (usually 1-2 weeks) after setting or majorly changing a strategy, during which the system collects data and performance is unstable. Too-frequent changes reset learning.
Automatic production and assembly of text, images and video directly in the platform, based on Google's models.
The creative hub in Google Ads for generating, bulk-editing and testing assets. Integrates Google's generative models.
The multimodal Gemini model integrated in Asset Studio (2026), enabling video generation, a natural-language interface and a “brief-to-asset” flow with conversational refinement.
Google's image-generation model, used in Asset Studio to create and extend visual assets.
Google's video-generation model, used to create clips from images and product feeds.
The feature that replaces old “text guidelines” with “messaging guidelines”: you give the system a message-and-tone brief, not rigid text rules, and the AI generates within that frame.
Automatic generation of headlines and descriptions from your site. In AI Max these evolved into “text customization”.
The feature where AI generates variations of your headlines, descriptions and images and tests them alongside the originals. AI content is flagged in reporting and can be turned off — you keep approval control.
One-click A/B testing of assets, extended to asset-group level in PMax.
The search ad that dynamically combines multiple headlines and descriptions you supply into the most relevant variant for each search.
Real-time assembly of the best-fitting combination of headlines, descriptions and images, according to the user's context.
Data collected directly from your customers, with consent (email, phone, CRM records). The most durable marketing asset: unlike cookies, it belongs to you and survives any platform identifier.
Every AI system — bidding, targeting, creative — is only as good as the signals it receives. Without quality first-party data, the AI is blind.
Third-party cookies, which underpinned digital advertising for two decades, are blocked or drastically limited (Safari, Firefox) and on their way out. They triggered the shift to first-party data architectures.
Sends hashed first-party data (email, phone, address) alongside conversion events. Google matches it to signed-in users and recovers conversions lost to cookies, cross-device behavior or delays. Typical reported lift: ~5-17% more conversions. It supplements existing tags, it does not replace them.
Technology that adjusts data collection based on user consent. If the user declines cookies, Google reports modeled conversions instead of losing the data entirely. Legally required for EU/EEA and UK traffic; recommended everywhere for data completeness.
AI estimation of conversions that cannot be observed directly (non-consented users, blocked cookies), using probabilistic models that recover much of the otherwise-lost journeys.
Importing real sales data from the CRM back into Google Ads (e.g. which leads actually became customers). Lets the AI optimize for real profit, not just generated leads.
The difference between “1,000 leads” and “the 80 that brought money”. OCT closes the loop between the ad and profit.
Uploading your (hashed) customer lists into Google to target or exclude audiences and train targeting with first-party data.
Moving data collection from the browser (fragile, client-side) to a server you control. Increases accuracy and future-proofs measurement against browser restrictions.
The irreversible encryption algorithm applied to first-party data before it is sent to Google, so sensitive information cannot be reconstructed.
The event-based analytics platform, a source of audiences and conversion signals for Google Ads.
Sending conversions server-to-server to the platform, more robust than the browser pixel. A natural pair with server-side tagging.
The AI model that distributes credit for a conversion across touchpoints based on real contribution, not fixed rules (e.g. last-click).
Google's Marketing Mix Modeling (MMM) solution, back as measurement infrastructure in the signal-loss era, with GA 360 integration.
The generative summary that answers a question directly on the Google results page, synthesizing sources from the web. Increasingly it also shows ads (e.g. Shopping ads on a “how do I clean the oven” search).
The conversational search experience that anticipates intent and follow-up questions. It has passed one billion monthly users. Ads appear triggered by intent, not strictly by keywords.
The original (older) name of Google's generative search experience, later evolved into AI Overviews and AI Mode.
The mechanism by which AI Mode generates extra answers to follow-up questions it anticipates, beyond the initial query.
The shift from short keywords (“running shoes”) to long, assistant-style phrases (“best running shoes for sensitive knees under 500 lei”). It fundamentally changes which ads become eligible.
A format where Gemini selects your most relevant products for a search and writes, on the spot, a personalized explainer for why your product fits that user.
A brand agent built with Gemini, placed directly in the ad. Instead of a static form, the user taps “Chat” and gets instant answers based on your site — turning the interaction into a lead.
Promotions that appear naturally in the AI Mode answer as the user explores options, with Gemini building a deal per query.
Google's infrastructure for agentic shopping: a cross-retailer cart that works in Search, Gemini and other surfaces, with native checkout.
Optimizing presence for generative engines and AI assistants (ChatGPT, Perplexity, Claude, AI Overviews). The complement to classic SEO in the era of conversational search.
The terms are not islands. They work as a system: first-party data feeds value-based bidding, generative creative provides the raw material that AI campaign types combine, and all of it becomes eligible in conversational surfaces. Weakness in one pillar breaks performance in the others. That is what we do at WebSEM: we don't switch features on in isolation; we build the system in which Google's AI produces business results — measured on profit, not vanity metrics.
Definitions reflect the state of the Google Ads ecosystem in mid-2026, including the Google Marketing Live 2026 announcements. Terms evolve fast; we update periodically.
Want these features run as one system, measured on profit?