Blog / AI & search /

AI and SEO in 2024: what actually changed

If you work in SEO, 2024 probably felt like one long AI headline. Search results changed, AI writing tools became routine, and everyone asked whether the old advice still applied.

The answer is less dramatic. Some interfaces changed fast. The work of understanding a customer and publishing something genuinely helpful stayed stubbornly important.

Search itself changed

Google introduced AI Overviews in the United States in May. That changed how some queries were answered: a summary could appear above the familiar links. The feature expanded through the year, and its presence depended on query, market and rollout. It didn’t mean every search became an AI answer, or that every site would lose traffic.

For a site owner, the lesson was to look at the actual result for important queries. Is Google showing an overview? Are the traditional results still visible? Does the query want a short fact, or a comparison and a decision? A search interface can change the click opportunity, but the content still needs to meet the need behind the query.

ChatGPT Search launched in October for eligible users, with broader access planned over time. It brought web sources and links into a conversational answer. Perplexity had already made source-linked answers familiar to many people. These products made it more natural to ask a system a follow-up question instead of opening several tabs at once. That was a meaningful change in how some people explored a topic, but not proof that ordinary search had vanished.

AI changed the production conversation

Writing tools got easier for brainstorming, summarising and drafting. Great for a blank page or a heap of notes. They also produce confident sentences that are shallow, repetitive or wrong. The tool does not know whether a business actually offers a service, whether a product detail is current or whether a claim is supported by the source.

Google’s guidance remained focused on the quality and usefulness of content, not on whether a human or a machine typed the first version. In March, its spam policies addressed scaled content abuse: producing many pages primarily to manipulate rankings can violate policy regardless of the production method. That distinction matters. AI is not a quality guarantee, and human authorship is not an automatic quality guarantee either.

I use AI as an assistant, but a person still has to decide what belongs in a draft. I check research notes, and I make sure examples are true and advice fits the reader. If a paragraph could be pasted onto a hundred sites unchanged, it needs more thought.

Measurement became more important, not less

When a new search surface appears, the temptation is to declare success or failure from a few screenshots. Keep a baseline instead. Search Console shows queries, pages, impressions and clicks; analytics shows what visitors do next. Neither tells the whole story or blames AI for every change.

A fall in clicks could have several causes: a ranking change, seasonality, a change in the search result, a site issue or a change in demand. Check the dates and affected pages before explaining it. Likewise, an increase in visits from an AI service is interesting, but it does not show how many people saw an answer and never clicked.

What did not change

People still need clear product information, trustworthy service pages, useful answers and working websites. Technical foundations still matter because a page that cannot be crawled, loads poorly or fails on mobile is harder to use. Search intent still matters because a guide cannot satisfy someone who needs to compare prices, and a product page cannot replace an explanation when the person is still learning.

What changed is the number of places where someone might meet an answer, and the need to watch how they evolve. Experiment with AI tools and review new features by all means. Abandoning useful content because a prediction says it’s finished is another matter.

A calmer way to plan for 2025

Keep a short list of the queries and pages that matter to the business. Review their current results and note any AI features you can actually see. Make the pages clearer and more complete for the person who does click. Measure the outcomes that matter to the business, and annotate significant site changes so you can interpret later movement.

Treat the year’s AI announcements as changes to observe, not a script for your whole strategy.

What I would keep in perspective

An announcement is not the same as a change to every result. AI search products can summarise information, but people still need to assess the source and complete tasks on the site itself. If your page is cited, the visit may or may not appear in analytics; if it is not cited, that alone does not establish a content problem. Track observable changes, keep pages accurate and avoid rewriting useful material to satisfy a theory about a system you can’t inspect.

Which search change did your own audience actually meet in 2024, and what evidence would help you decide whether to respond?