From Keywords to Conversations: How AI is Digital Marketing More Human

In May 2026, Google AI searches surpassed 1 billion users with AI search queries more than doubling each quarter of the last year. This milestone set the tone for the company during the Google I/O developer conference where Google announced their most significant changes to Search within the last 25 years.   

This new Google Search is faster with Gemini 3.5 Flash and can now generate outputs in any format (e.g text, video, audio and image). The refreshed platform also incorporates a larger “intelligent search box” designed for long -form questions and a new AI Mode merging the original AI Overviews into a single Search flow featuring deeper, multi-step conversational follow-ups with the AI agents.  

But what do these changes mean for the Search Engine Optimization and digital marketing practices we have used for the past quarter century? 

The Old World 

To understand the changing search landscape, we first have to recognize what makes AI models different from traditional search engines. Traditional search engines, such as Google Search in both its paid and organic forms, are based on a simple “one input, one output” system. For a long time, successful digital marketing was based on targeting the right keywords and out bidding competitors. Matching one key word to an ad was enough to drive results. Or, as Frederick Vallaeys former Google developer and the current CEO of Search Engine Land says, “Google made us lazy,” because the search algorithm was never tasked with figuring out a user’s intent or overarching goal. 

The Journey Onward 

While some users treat Large Language Models such as ChatGPT and Claude as fancy search engines, results tend to be poor, inaccurate or vague. A misconception or misunderstanding of AI is to think of it as just personalized search. Users Google “mattress” and quickly get a list of companies who sell mattresses, and maybe some consumer insight information about the best mattresses. AI can do that too. However, according to Vallaeys, this is a mistake. His advice is to give AI the real goal; don’t just ask for the best mattress – explain the context to the AI that you want to be healthier and contextualize sleep as part of the broader problem and let AI help work through the options.

To effectively use AI requires understanding that these tools can perform structured outputs. AI models are based on multi-step and muti-input processes from the user, and results come in the form of actionable steps or generative content such as a spreadsheet, video or graphics. Users are no longer limited to just asking for information; AI invites them to engage in conversations about the steps needed to solve their problems.  

The New Frontier  

AI forces both users and marketers to shift from an input/output model to conceptual thinking about solutions for specific problems. As users come to use AI more fluently, the challenge for digital advertisers is that we are no longer tasked with simply matching one keyword to one ad.  

For Celtic, this is shifting our approach to the tactical recommendations we make. We are placing greater emphasis on content that dives deeper and answers specific questions such as FAQs and “How To” landing pages. We are better leveraging the expertise of our clients' internal SMEs to enhance the authority of our sites. We continue to stress the importance of video to broaden the library of content types and formats for each client. 

At the end of the day, successful marketing tactics are about stimulating a sense of connection with a customer. According to McKinsey, roughly 71% of consumers expect companies to deliver personalized interactions. Brands are now expected to answer the “why” behind their solutions and not just the “what.” The move to a conceptual content strategy allows brands to go back to basics and explore who your customer is and how to help them.