When we first wrote this post in 2019, artificial intelligence was an increasingly fashionable marketing slogan, with great implications for the cost-effective scaling of marketing automation systems. It was obvious that AI would play a big role in marketing. The only question was how big.
Since then generative AI has arrived. OpenAI launched ChatGPT in November 2022, and in McKinsey’s State of AI survey published in March 2025, 71% of respondents said their organizations regularly use generative AI in at least one business function, with IT and marketing and sales the functions where they use it most often. Technology still develops faster than business can follow, but AI applications are now a mainstay of marketing teams around the world, often inside the tools those teams already pay for.
What types of AI are used in marketing
Four kinds of AI do most of the work, often inside the same tool:
- Predictive machine learning: models that score leads, forecast demand, predict churn and pick the next best offer for each customer.
- Natural language processing: software that reads reviews, social posts, search queries and support chats to find intent and sentiment.
- Generative AI: language models that draft copy, and image and video models that produce visuals, from ad variants to product photos.
- Agentic AI: systems that carry out multi-step tasks, such as building a campaign, launching it and adjusting it, within limits a person sets.
Which tools are best changes every few months, but the ones most teams use fall into a few groups: general assistants such as ChatGPT, Claude and Gemini for drafting and analysis; creative tools such as Adobe Firefly and Canva; the AI built into ad platforms, such as Google’s Performance Max and Meta’s Advantage+; and the AI inside CRM and email suites such as HubSpot and Salesforce. How can artificial intelligence help in practice?
Generating valuable content
Given a data set, AI content writing programs could already produce fairly human-like articles in 2019. Automated Insights’ Wordsmith, the best known tool of that era, produced over 1.5 billion pieces of content in 2016, mostly short texts built from data and templates, such as company earnings stories for the Associated Press.
Large language models changed both the scale and the kind of content. They draft blog posts, ads, product descriptions, social posts and video scripts in seconds and in any language, and image models produce visuals to match. The weak spots are just as clear: generic text that sounds like everyone else’s, invented facts and a brand voice that drifts. Google’s position, published on 8 February 2023, is that it rewards high-quality content however it is produced, but that using automation, including AI, mainly to manipulate rankings breaks its spam policies. In practice the strongest teams use AI for research, drafts and variants, and keep people for the expertise, the facts and the final edit.
Reaching audiences and dividing buyers into segments
Big data about the broad demographic population matters less and less to marketers. What matters is your own data: if you cannot organize and segment it and build trend models on it, your ROI will start to fall as the market grows. AI can predict which segments of recipients are more likely to convert, show which customers are most loyal and which are about to leave, and find new prospects who look like your best customers.
Because these models learn from first-party data, collected with consent on your own site, app and stores, the quality of that data now decides the quality of the targeting.
Buying and optimizing ads
Advertising is where AI took over the most decisions. In November 2021 Google opened Performance Max to all advertisers: one campaign that buys ads across YouTube, Display, Search, Discover, Gmail and Maps, with the system choosing placements, bids and combinations of creative. Meta’s Advantage+ campaigns work in a similar way. The marketer’s job moved from setting bids to feeding the system good inputs: clear conversion goals, accurate conversion data, strong creative and sensible budgets. When those inputs are wrong, the AI optimizes efficiently toward the wrong thing.
Guiding a customer through the sales funnel
Marketers no longer have to lead every potential client through the entire customer journey by hand. If you have accurate data, your artificial intelligence may actually be better than your marketing specialists at telling which leads are ready to buy and which ones will convert only with a discount.
You know the small chat window that sometimes appears on a shop page, asking if you want to talk to a representative or have any questions? Have you ever wondered how it answers so quickly? Often the one talking to you is an AI chatbot. Many e-commerce websites rely on customer support with artificial intelligence. Thanks to machine learning and language models, websites can handle a huge volume of customer service requests at a pace no single representative could match. AI gives websites a 24-hour service, which online shops can hardly do without today.
How AI changes marketer-customer communication
Artificial intelligence changes almost every part of marketing, from customer service and social media to the reach of email and the online experience, through website optimization and personalized product recommendations. AI also changes the way marketers work, giving them access to new insights and scale, so they can make better decisions and work more efficiently. Artificial intelligence is at the forefront of creating engaging experiences online, and bringing it into your marketing processes has become a condition for staying competitive.
Search, AI Overviews and AI answers
The biggest change since 2019 is in how people find things. Google brought AI Overviews, AI-written summaries above the search results, to everyone in the US in May 2024, and began rolling out AI Mode, a conversational way to search, to everyone in the US in May 2025. Many people also ask ChatGPT, Gemini or Perplexity directly. Part of the journey that used to start with a click on a search result now starts, and sometimes ends, in an AI answer.
For marketers this means two jobs instead of one. Content still has to rank, and it also has to be the kind of source an AI answer quotes: clear answers to real questions, visible expertise, facts that can be checked and pages that crawlers can read. We described the early stage of this shift in how AI affects SEO.
Website testing
A company’s website is the most important place where it serves its customers. By optimizing website design, marketers can improve the site and ultimately make conversion easier. To find the best design, companies test many versions and choose the one that converts best. The most popular traditional test method is A / B testing. It is good for testing a single hypothesis you want to confirm, such as a change to the navigation structure, but it is limited when it comes to finding the combinations of ideas that improve the experience the most.
AI-driven testing goes further in two ways. Multi-armed bandit algorithms shift traffic toward the better-performing version while the test is still running, so fewer visitors see the losing one. Personalization engines show each visitor the version most likely to work for them, instead of one winner for everyone. The tools changed too: Google shut down Google Optimize, its free testing tool, on 30 September 2023, so teams moved to other testing platforms or to the testing built into their analytics, CMS or customer data platform.
Personalized emails
Email has existed in internet marketing from the very beginning, and it is still one of the most effective channels. So marketers want as much insight as possible into the kinds of emails their audience responds to best. AI analyzes the strongest email campaigns: which designs worked, which subject lines inspired the most and much more. With these observations, artificial intelligence helps marketers understand what recipients react to and generate emails tailored to each group of recipients. Emails are already written to feel like a personal conversation, and artificial intelligence lets you connect with potential customers at a deeper level than ever before.
Today that means send-time optimization, which delivers each message when that person usually reads email; subject lines and copy generated and tested in many variants; and messages triggered by what a customer just did on the site, which is the core of behavioral email marketing. The model should optimize for clicks and sales, not opens alone, because a subject line that gets opened but disappoints costs you trust.
Is it legal to use AI in marketing?
Yes, but the usual rules apply to it, and some new ones were written with it in mind. In the US, the Federal Trade Commission’s final rule announced on 14 August 2024 bans fake reviews and testimonials, explicitly including AI-generated ones that claim to come from people who do not exist. In the EU, the transparency rules in Article 50 of the AI Act have applied since 2 August 2026: deepfakes must be disclosed, and systems that generate images, audio, video or text must mark their output in a machine-readable way. Personal data used for targeting and personalization still falls under privacy law such as the GDPR.
The ethical risks are just as practical. Models trained on biased data can leave groups out of offers, hyper-targeting can feel like surveillance, and a confident AI-written claim about a product is still the company’s claim. Good practice is simple to state: check facts before publishing, tell people when they are talking to AI, collect data with consent, and keep a named person accountable for every campaign.
What will the future bring?
AI-powered applications are transforming the ways that marketers analyze data, conduct tasks and even create content. In 2019 we expected enhanced analytics to shift the marketing mindset from reactive to proactive as businesses plan, define and execute campaigns. That shift is well under way, and the next one has started: AI agents that run parts of a campaign on their own, while people set the goals and the limits.
We’ve entered the age of the agile marketing team. AI-powered technologies increase teams’ ability to react to change in the moment, without lengthy campaign planning and drawn-out strategy. Greater responsiveness means the end of irrelevant communication, as brands can finally reach the right audience with messaging that resonates at the right time.
As personalization becomes even more sophisticated, mass marketing will be replaced by intimate one-to-one marketing across all customer touchpoints. After all, who wants to spend dollars on people who are not likely to become customers?
Will AI replace marketing jobs? It replaces tasks first: first drafts, report building, bid management, resizing creative. Roles change around what is left, which is strategy, judgment, the brand and checking what the machines produce. Marketers need a growing capacity to analyze data, because their work increasingly resembles that of data analysts, and they need to brief and review AI as carefully as they would a junior colleague.
Despite the hype and some confusion in the market, investment in AI remains high, and the technology still moves faster than most plans. The good news is that new insights, better forecasting and more creative organizations are worth having. AI already helps marketers understand their customers and makes the experience easier for those customers, from predicting what someone will want to helping them through the purchase and making sure they are satisfied afterward.
Any questions? Let's talk!
Write to us,