AI in marketing: how machine learning changes the industry

February 10, 2023

Artificial intelligence has made its mark on marketing. Companies use it to understand customer behavior, predict what customers are likely to want and show them relevant advertising. AI can examine millions of data points, from buying habits to on-site behavior, faster and at lower cost than a team of analysts, and help marketers choose the right channel and the right moment to reach each person.

Where machine learning earns its keep

• Predictive segmentation and lead scoring. Models group customers by likely behavior rather than demographics alone and rank leads by their chance of converting, so sales teams call the right people first.
• Recommendation engines. Product and content recommendations, familiar from Amazon and Netflix, are now within reach of mid-sized shops and apps through e-commerce platforms and cloud services.
• AI-driven ad campaigns. Google's Performance Max runs a single campaign across all Google Ads inventory, including Search, YouTube, Display, Gmail and Maps, while Meta's Advantage+ tools use AI to optimize targeting, bidding, creative and placements across Facebook, Instagram and Meta's other apps. The advertiser sets goals, budgets and creative; the algorithm decides where the ads appear and to whom.
• Chatbots and assistants. AI assistants answer product questions, suggest suitable products and qualify leads around the clock.
• Generative content with human review. Language and image models draft ad variants, emails and product descriptions in minutes. They work best as a first draft that an editor checks; see five ways content creators can use generative AI.

What has changed since 2023

Generative AI has become an everyday tool, and search has changed with it. In May 2024 Google began rolling out AI Overviews, AI-generated summaries at the top of results, to all users in the US. A Pew Research Center analysis of March 2025 browsing data found that users clicked a traditional search result on 8% of visits to pages with an AI summary, against 15% without one. Original data and real expertise, which a summary cannot replace, are now worth more than generic informational content.

Data and compliance

Under the GDPR, processing personal data needs a lawful basis such as consent, which makes first-party data collected with clear consent through accounts, purchases, newsletters and apps the most valuable input. Third-party cookies have not disappeared on schedule: Safari has blocked them by default since 2020, but Google kept its user-choice approach in Chrome and in October 2025 retired most of its Privacy Sandbox technologies, including Topics and Protected Audience.

The EU AI Act's transparency rules in Article 50 apply from August 2, 2026. People must be told when they are talking to an AI system such as a chatbot, unless that is obvious; deepfake images, audio and video must be disclosed as AI-generated; and providers of generative AI tools must mark their output in a machine-readable way, with a deadline of December 2, 2026 for systems already on the market.

Risks to manage

• Hallucinated claims. AI can state prices or policies that are wrong, and the company remains responsible. In 2024 a Canadian tribunal ordered Air Canada to compensate a customer after its chatbot gave wrong information about bereavement fares.
• Brand voice. Unedited generated copy tends to sound generic. Style guides and human editors keep it consistent.
• Bias. Models trained on historical data can exclude or stereotype groups of customers, a legal risk as well as a reputational one in areas such as credit, housing and employment.
• Attribution. Automated campaigns report results by their own logic. Holdout tests and independent analytics show whether AI is creating demand or taking credit for sales that would have happened anyway.

How to start

Start with clean, consented customer data in one place, usually the CRM. Pick one measurable problem, such as lead scoring or churn, and test the AI features already built into your ad platforms, CRM and email tools before building anything. Custom models pay off when your data is a competitive asset that off-the-shelf tools cannot use, or when you need control over how predictions are made and where data is processed. Most of the engineering effort goes into integrating them with the CRM and ad accounts. Our guide to getting the most out of AI in business covers the wider picture.