AI in marketing already creates measurable value in a few clear areas: creative variations at scale, research and synthesis, continuous campaign monitoring, early message testing and automated reports. In strategy, in content published in the brand's name and in performance forecasts, it can help with analysis and proposing alternatives, but the decision and accountability stay with people. What separates the two sides is mainly the workflow, the metric and the oversight around the technology.
In brief
- ערך מדיד כבר היום. Creative variations, research, campaign monitoring, message testing and reporting are areas where AI can save time and support decisions.
- שליטה אנושית עדיין חובה. In strategy, positioning, content before publishing and forecasts, AI can help, but the decision and accountability belong to the people who know the business.
- הפלטפורמות כבר אוטומטיות. Google's Performance Max and AI Max and Meta's Advantage+ make a large share of targeting and bidding decisions, and your role is to give them good data and assets.
- הצד השני של המטבע. Users ask ChatGPT and Gemini about product categories, so it's worth knowing whether the brand appears in the answers and how it is described in them.
Where does AI in marketing already create measurable value, and where does it still need human control?
AI in marketing creates measurable value in tasks with a lot of data, repetition and fast feedback, and needs human control in tasks that require judgment, accountability and business context.
There is a lot of talk today about artificial intelligence in marketing. To understand where it actually changes the work, it helps to sort the uses by two questions: can the AI's contribution to the result be measured, and who is responsible when it makes a mistake. The following table organizes common uses by these questions.
| Area | Where AI already creates measurable value | Where human control is still required |
|---|---|---|
| קריאייטיב | Generating variations of headlines, copy and formats, so advertising systems learn faster | Keeping the brand voice, legal approval and choosing the central concept |
| Research | Synthesizing market research, social listening, competitor mapping and summarizing customer reviews | Framing the research question and deciding which findings are relevant to the business |
| Message testing | Narrowing down creative directions before they reach the media budget | The final decision, and testing with a real audience when the stakes are high |
| Campaign monitoring | Continuous monitoring and alerts about anomalies in performance and spend pacing | Diagnosing the cause and deciding what to change |
| דיווח | Reports that update automatically with a written summary of what changed | Business interpretation and recommendations going forward |
| Strategy and positioning | Proposing alternatives and analyzing scenarios | The choice itself, which relies on knowing the business, the brand and the market |
| Content for publication | Fast first drafts | Checking facts, context and accuracy before every publication |
| Forecasts | An indication for planning and for comparing scenarios | Not treating a forecast as a commitment to a result |
In most rows the AI's contribution can be measured: preparation time, number of variations tested, response time to an anomaly. What the rows in the last column have in common is that a mistake there is costly. Models can make mistakes, invent details and miss context, so any content published in the brand's name needs to pass a human eye.
What are Google's and Meta's AI tools already doing inside campaigns?
Google and Meta already build AI into the core of their buying systems, so a large share of targeting, placement and bidding decisions is made automatically.
Performance Max is a goal-based campaign type in Google Ads that makes it possible to reach all of Google's inventory from one campaign: Search, YouTube, the Display Network, Discover, Gmail and Maps. The advertiser sets conversion goals and provides creative assets and audience signals, and Google's AI combines them and serves them across the available inventory. Smart Bidding sets the bid in every auction according to the likelihood of conversion, and generative tools can produce additional images, videos and text.
AI Max for Search campaigns is a bundle of expansions for Google Search campaigns. It includes improved search term matching, customization of headlines and descriptions, and expansion of final URLs to a page that better matches the user's intent.
At Meta, Advantage+ is an umbrella name for automating campaign components. According to Meta's developer documentation, a sales, app or leads campaign is defined as Advantage+ when the audience, campaign-level budget and placements are all set to automatic. The advertiser still controls the campaign objective, the bidding strategy, geographic location, creative and audience exclusions, such as existing customers.
For brands, this means that when the system decides, the quality of the input determines the result. Three inputs are especially important: complete and accurate conversion data, a variety of creative assets that stay on brand, and a goal that reflects profit, not just volume. Control shifts from setting targeting manually to setting goals, measurement and quality control. As of October 2026 the platforms continue to update these tools at a fast pace, so it's worth checking the official documentation before any structural change in the account.
What is an AI agent for business, and how does it help with campaign monitoring?
An AI agent for business is software that performs a defined task continuously, for example tracking data, identifying anomalies and sending an alert, without someone having to run it each time.
In campaign management, an agent's value is time. Instead of discovering in the monthly report that a campaign spent its budget too fast, or that conversions dropped because of a broken tag, the agent alerts when it happens. The team can then focus on diagnosis and decision, rather than manually scanning accounts and reports.
It's important to distinguish between an agent that alerts and an agent that acts on its own. An agent that alerts and suggests an action fits most uses. An agent that changes budgets or pauses campaigns by itself requires clear boundaries: what it is allowed to change, within what range, and who gets a report on every action. Without these boundaries, automation meant to save time can create mistakes that are hard to spot.
A related use is early message testing. Synthetic audience research is a method in which ideas and messages are tested against personas built on data, before investing in production and media. The method doesn't replace research with real people for big decisions, but it quickly narrows the number of directions that reach the next stage.
How do brands appear in ChatGPT's and Gemini's answers?
Brands appear in the answers of AI engines when there is clear, consistent and reliable information about them online that the engines can find and cite.
Users ask ChatGPT, Gemini or Perplexity about a product category and get one composed answer, instead of a list of results. GEO/AEO (Generative Engine Optimization) is the field that deals with brands' visibility in the answers of AI engines. A brand that doesn't appear in these answers, or is described inaccurately in them, doesn't always know it, so it's worth checking.
The work starts with measurement: what the engines say today about the brand and its competitors, which questions it appears in and which it doesn't, and which sources the engines draw on. From there you build content that answers these questions directly, with clear definitions, verified data and a structure that is easy to cite. In this sense, GEO/AEO and SEO complement each other rather than compete.
How do you choose AI tools for marketing, and what may you put into them?
An AI tool for marketing should be chosen by the workflow it improves and by a clear data policy, not by a list of features.
When evaluating a vendor or a tool, the question "do you use AI?" isn't enough. Ask which specific process changed and by how much, and ask to see the metric. It's just as important to check what may be entered into the tool. OpenAI, for example, states that in the business versions of ChatGPT and in the API it does not use business data to train models by default, and that the customer owns the input and output to the extent permitted by law. The retention period for input and output depends on the product, the specific service and the agreement, so it needs to be checked separately for each tool and each type of data.
The safe way to start integrating AI into marketing follows a fixed order:
- הגדירו את המשימה. Which workflow the tool is supposed to improve, and what metric will show it works.
- מפו את הנתונים. Which data will go into the tool, and especially whether it includes personal information about customers.
- בדקו את תנאי הספק. Whether the data is used for training, how long it is retained and who owns the outputs.
- הגדירו בקרה אנושית. Who reviews the output before it goes out, and in which cases publishing without approval is prohibited.
- הריצו פיילוט מוגבל. One task, a defined period and a comparison with the previous state.
- החליטו לפי המדד. If the improvement is measured, expand. If not, stop and move on to the next use.
When personal information is involved, it's worth also involving the privacy protection officer or legal counsel, especially since Amendment 13 to the Privacy Protection Law came into force in August 2025.
How we approach this at CONNECT
At CONNECT, part of the Bruckner Yaar Levi advertising group, we build our use of AI around existing workflows, not the other way around. AudienceLab tests ideas, scripts and messages against a panel of 200 Israeli synthetic personas, with in-depth interviews, a group discussion room and a Go, Revise or No-Go recommendation, before a direction reaches production and media. Our AI agents for campaign monitoring monitor the accounts and flag anomalies in performance and spend pacing, so the team can review and address them earlier.
Wherever the decision is strategic or the content goes out in the brand's name, the decision stays with people. In parallel, we track how ChatGPT and Gemini present the brand as part of our GEO/AEO work, to know what the engines say rather than just guess.
Frequently asked questions
What is AI in marketing?
AI in marketing is the use of artificial intelligence models to perform or accelerate marketing tasks: generating creative variations, researching and synthesizing information, monitoring campaigns, testing messages and reporting. Some of the uses are already built into the advertising platforms themselves, such as Performance Max and Advantage+. The real value comes when AI is integrated into a workflow with a clear metric and human oversight.
Which AI tools for marketing are worth starting with?
It's best to start with tools that improve an existing, measurable process, not with the most talked-about tool. Good starting uses are summarizing research and reviews, generating variations of ad copy, and reports that update automatically. Before entering data into a tool, check whether it is used for training, how long it is retained and who owns the outputs. Then run a short pilot and measure.
Can AI replace an advertising agency?
AI changes the nature of an advertising agency's work more than it replaces it. AI speeds up repetitive tasks such as variations, reports and monitoring, and frees time for work that requires judgment: strategy, positioning, choosing the concept and budget decisions. Models can make mistakes and invent details, so someone needs to review the outputs, take responsibility for them and connect them to a deep understanding of the business.
What is an AI agent for business?
An AI agent for business is software that performs a defined task continuously and independently, such as tracking campaign performance, identifying spend anomalies and sending alerts. In marketing, such an agent shortens the time between a problem appearing and it being addressed. When the agent is also authorized to take actions, for example changing a budget, it's important to define clear boundaries in advance and a reporting mechanism for every action.
Is it allowed to enter customer data into AI tools?
It depends on the tool, the vendor's terms and the purpose of use. Before entering personal information, check whether the vendor uses the data for training, how long it is retained and who can access it. Since Amendment 13 to the Privacy Protection Law came into force, it's right to involve legal counsel or the privacy protection officer in the decision. Where possible, it's better to work with aggregated or masked data.
ראו גם: Technology & AI in Marketing · מה זה GEO/AEO · Synthetic audience research · פרטיות ומדידה ותיקון 13 · הטכנולוגיה שלנו
Sources: Google Ads: About Performance Max campaigns · Google Ads: About AI Max for Search campaigns · Meta for Developers: Advantage+ campaign experience · OpenAI: Enterprise privacy · Privacy Protection Authority: professional guide to Amendment 13
רוצים לבדוק איפה AI יכול לחסוך זמן בתהליכי העבודה שלכם? השאירו פרטים.