Answer Engine Optimization

Anyone who used to want to know something typed it into Google and got a list of links. Today, many people instead ask their question to a chat program like ChatGPT and receive a ready-made answer directly. Such systems are called answer engines. For companies, this fundamentally changes the situation: it’s no longer enough to rank high in a list of links. You have to appear in the formulated answer itself. Answer Engine Optimization is the work aimed precisely at that.

Why companies are fighting for a place in the answer

A results list has ten spots, an answer often has only two or three sources. This makes competition harder, not easier. Anyone not mentioned is practically nonexistent to the user. There is no second page to browse through.

On top of that comes an economic problem known in the industry under the buzzword “zero-click.” The user gets their answer and no longer clicks through to the website the information came from. Publishers and online shops lose visitors and advertising revenue as a result, even though their content is being used. For many media companies, this is a serious threat to their business model.

That’s why a dedicated services market has emerged around AEO within just a few years. Agencies sell analyses that measure how often a brand appears in AI answers. Such figures are now reported in boardroom meetings, similar to Google rankings in the past. Whether these methods actually work reliably is disputed among experts.

What you do to get the AI to name a source

Many answer engines work in two stages. They first search the web for suitable passages of text and then write an answer based on these findings. This process is technically called Retrieval Augmented Generation, or RAG for short. Anyone wanting to be mentioned there must first make it into this preliminary selection.

In practice, this means texts are constructed so that individual sections remain understandable even when taken out of context. A clear question as a heading, directly followed by a concise answer in two to three sentences. Facts, figures, and dates are spelled out rather than merely hinted at. In addition, operators embed invisible structural information in the page’s source code, so-called schema markup, which tells machines what constitutes a price, an author, or opening hours.

A second lever is being mentioned elsewhere. Language models rely heavily on pages that many others cite, such as Wikipedia, specialist forums, or review sites. Anyone appearing there is picked up more often in answers. This is what distinguishes AEO from classic SEO, where links to one’s own site mattered most. And it has a downside: some try to flood forums deliberately with praise for their own products.

AEO in products and headlines

You encounter the result of this daily without hearing the term. Google’s AI Overviews above the search results, the answers from Perplexity, or Copilot in Windows are all answer engines. Every source citation within them is a spot companies compete for.

In business news, AEO usually comes up in two contexts. First, in figures from publishers reporting plummeting search traffic. Second, in legal disputes and contracts in which media companies demand payment for having their articles appear in AI answers. Both describe the same shift: attention is no longer gathering on websites, but in the answer that precedes them.

A common misconception is that AEO is a reliable process with fixed rules. Unlike with search engines, there is no verifiable ranking, and the models don’t always answer the same question the same way. Much of it is therefore based on experience and can be rendered obsolete by the next model update. Anyone practicing AEO works with probabilities, not guarantees.