How SEO has changed: from keywords to AI answers

SEO has moved from tricking search engines with repeated keywords to earning genuine trust. In the early days, you could rank by stuffing a page with the phrase people typed. Search engines then learned to judge links, quality and meaning. Now AI answers summarise the web and recommend a handful of businesses, so being known and trusted matters more than ever.

What was SEO like in the early days?

In the early days of web search, SEO was mostly about keywords. Search engines matched the words in a query to the words on a page, so people repeated phrases over and over, hid text in the same colour as the background and filled their meta tags (short pieces of code that describe a page to search engines) with long lists of terms.

Google changed things by also counting links. A link from another website was treated like a vote of confidence, and links from respected sites counted for more. That idea, known as PageRank, made results much better, but it soon created a new shortcut: buying, swapping and manufacturing links in bulk.

Why did Google crack down on SEO shortcuts?

Google cracked down because the shortcuts made results worse for the people searching. Around 2011 and 2012 it released two well-known algorithm updates (changes to the rules it uses to rank pages): Panda, which targeted thin, low-quality and copied content, and Penguin, which targeted manipulative links.

Sites built on tricks lost much of their traffic, sometimes almost overnight. The lesson has repeated itself ever since: tactics that work around the rules tend to stop working, while pages that are genuinely useful tend to survive each change.

How did search learn to understand meaning?

Over the 2010s, search moved from matching words to understanding what people mean. Updates such as Hummingbird, and later machine learning systems such as RankBrain and BERT, helped Google interpret whole questions, synonyms and context rather than hunting for an exact phrase.

At the same time, searching moved onto phones. Google began favouring sites that work well on mobile, and local results with maps and reviews became central for any business serving a particular area. Quality also became more about the source: Google's guidelines for the people who assess its results talk about experience, expertise, authoritativeness and trust, often shortened to E-E-A-T.

The practical result was simple. Writing for an exact keyword mattered less, and answering the real question well mattered more.

How are AI answers changing SEO now?

Search is shifting from a list of ten blue links to a written answer. Google's AI Overviews put a summary at the top of many results, and assistants such as ChatGPT, Gemini and Perplexity answer questions directly, often naming a few businesses and citing a few sources.

Making your business easy for these systems to find, understand and recommend is often called AEO, or answer engine optimisation. It overlaps heavily with good SEO, because AI answers frequently draw on pages that search engines already trust. But it adds a layer: these systems also pick up what reviews, news coverage and other websites say about you.

Two things follow. Some people now get what they need without clicking, so traffic alone is a weaker measure of success. And there are far fewer places in an AI answer than on a page of results, so being one of the businesses mentioned is worth more.

There is a twist, too. AI has made it cheap to produce pages of text, and the web is filling up with similar, generic content. It is the keyword era again in a new form: volume is easy to fake, reputation is not.

What does the evolution of SEO mean for your business today?

It means the fundamentals matter more than ever, and shortcuts matter even less. In my experience, the businesses that do well in both search results and AI answers tend to do the same handful of things consistently:

• Be specific. Say clearly what you do, who it is for and where you work. Vague pages are hard for search engines and AI to recommend.
• Answer real questions. Write about what customers actually ask, in plain language, drawing on your own experience. Copy written and edited by people stands out precisely because so much now sounds the same.
• Keep the foundations sound. A fast, mobile-friendly site that search engines can read easily still underpins everything else.
• Build reputation beyond your website. Reviews, mentions in local or trade press and consistent business details across directories all feed what AI systems learn about you.
• Check what is being said. Ask the AI assistants about your business and your services, and see whether you appear and whether the description is accurate.

If you would rather not do that last step by hand, ShiftCode offers a free check of what reviews, news, ChatGPT and Gemini say about your business. It is a useful starting point before deciding where to put your effort.

SEO is not dead, whatever you may have read. The job is the same one it has always been underneath the tactics: being worth recommending. What has changed is where the recommendation happens, and how hard it has become to fake. That is why a good SEO and AEO approach today looks less like a bag of tricks and more like building a reputation that holds up wherever people search.