AI in marketing, and where a business should start

The channels stay the same: search, paid social, email, your website, outbound. What changes is how much good work one skilled person can get through in a week, and how quickly the numbers tell them what to change. A business that treats AI as a new channel to add usually ends up with a few subscriptions and no change in its results.

Most businesses have started in some way. Nearly three in ten UK businesses said they used at least one AI technology in June 2026, rising to almost half of those with 250 or more staff. In the US the Census Bureau put the rate at about one business in five in early May 2026, and 37% among firms with 250 or more employees.^[1][2]

Marketing is where much of that use lands. In a Census Bureau study of US firms that use AI, sales and marketing was the most common business function for it, named by 52% of them, ahead of strategy and business development at 45% and IT at 41%. The use is still shallow. Among UK businesses with ten or more staff that use AI, the ONS found only 10% using it extensively.^[3][4]

Search interest follows the same pattern. Far more people look up what AI marketing is than look up how to use it, and the strategy question barely registers beside the headline terms.

At the top of the funnel the gain is volume with judgement: a person can put far more ad angles into a test than they could write by hand, then keep the ones the numbers back. On ToastPal, a B2C startup that helps people write wedding speeches with AI, Jakub used AI to produce ad copy variants for weekly Google Ads tests across the US, UK, Australia and Canada, before he founded the company. The weekly testing, pausing losers and scaling winners, is what turned the volume into revenue.^[6][7]

Further down, AI earns its place in speed and consistency. Enquiries get researched and routed the moment they arrive, abandoned baskets get an email written for what was left behind, and the Monday report arrives already summarised. Each of those is a small, measurable job, which is why they are the right place to start.

Start with these:

And leave these until the basics work:

Most strategies that fail skip the second item. They list tools and use cases without saying which leak they are meant to close, so nobody can tell after three months whether anything improved. Written the other way round, the strategy is short: the goal, the leak, the jobs, the checks and the scoreboard. Our AI marketing strategy framework goes through each part in more depth.

The shop is invented to show the order of work, and it carries no results: what those weeks deliver depends on the traffic, the product and the offer. The order is the point. Measure first, fix the most expensive leak, and only then spend more on attention.

Whichever you choose, ask the same questions. Can they show results with their context, not only the headline number? Who reviews what the AI produces before a customer sees it? Which numbers will they report every week? And if you stop, do the accounts, data and automations stay with you?

AI for Marketing is the fourth kind. We run marketing for ecommerce, SaaS and B2B service businesses on an AI system Jakub Cambor built, and Jakub reviews the work before the client approves it. He has worked in marketing since 2018, and several of the results on this page come from his freelance work before the company was incorporated in 2025. You can read how our AI marketing agency works, or see what it looks like for ecommerce, SaaS and B2B services.^[7]