The old world of search is not coming back: SEO for the synthesis economy

Let’s be clear about what is happening online right now. The way people find information, the way information is presented and the way it is consumed has changed. All three parts of the market:

• How supply is created: websites, pages.
• How demand is communicated: prompts and chat queues instead of queries.
• How the two are matched: synthesis instead of listings.

The old world is not coming back.

Understanding and accepting this shift is crucial for businesses. Nostalgia for the old ways won’t help.

Remember the old search experience? ~2000 to ~2025

For a quarter of a century, finding information online required effort. You typed queries. You received lists of links, images, maps … many of them requiring in-depth evaluation. It was a long, sometimes tiresome process of sifting, comparing and choosing.

SEO focused on making a specific listing show up, then stand out in that comparison, and then offer enough value that the user interacted with it. It relied on users doing the filtering work.

The new reality: synthesised answers

Contrast that with today. People pose questions to AI tools in natural language:

Plan a budget-friendly family trip to Prague next month, including weather expectations. I am in Vienna.

Almost instantly they receive a synthesised answer: an itinerary, cost estimates, advice.

This immediate conclusion appeals to our preference for ease and efficiency. The friction of sorting through listings disappears. Demand is expressed as a conversational prompt, and the response is a ready-made synthesis, not a list of options.

Google is a marketplace, AI is a personal shopper

Think of traditional search results as a marketplace. Website owners set up their stalls (websites), arranged their goods (targeted pages) and used signals (content, usage values, brand) to attract visitors browsing the aisles.

AI synthesis changes this model. An intermediary now assesses the stalls, selects components and presents a single, consolidated answer to the user. The original sources, your pages, often remain unseen. You have become part of a supply chain feeding a synthesiser, instead of engaging the end user directly in the first discovery phase.

The result? Choice compression. Instead of evaluating multiple options, users are presented with one. And because it is so easy, this single answer is often accepted without scrutiny.

The challenge of unseen sources

Convenient, yes. But the shift introduces a problem. When users mostly interact with synthesised answers, the signals of source credibility become less visible: website, UX, design, domain reputation, author expertise, visible context.

The evaluation shifts from the user assessing multiple sources to the user trusting or questioning the AI’s synthesis. The underlying process isn’t transparent. (At Google Search it wasn’t either.)

SEO: Synthesis Engine Optimization

Strategies focused only on rankings and CTR within lists of links need rethinking. The business is no longer just about appearing in the list. It is about influencing the synthesis, and providing value beyond what the AI summarises.

This change is unlikely to reverse. Once users adopt a more efficient way of getting information, they rarely go back to a more effortful one. The expectation of synthesised answers is here to stay.

Adapting for relevance: what to do now

Focusing on past methods is unproductive. The path forward requires adaptation. Here is where to direct your execution:

• Prioritise structured clarity. Make your information digestible for machines. Structure data logically. Provide clear, factual answers that can be accurately incorporated into a synthesis. Communicate your prices, your use cases, your value. Clearly.
• Create reasons to click through. If your content is referenced in a synthesised answer, that snippet should only offer partial value. Your full page must provide nuance, data, analysis, interactive elements, a perspective the AI summary cannot fully capture. Give users a reason to seek out the original source.
• Accept that there are no established best practices. Yet. Generative AI, LLMs and GPTs are changing so fast that the best practices of six months ago are already outdated. As of May 2025: AI-powered search was a joke at the end of 2024, now it is a powerful tool. AI-generated answers in Google were shallow a few months ago, now they are the reason users are Googling again.

Optimising for these new channels means using them, learning them, trying them, working them. You can do that randomly, or systematically. I recommend systematically.

In short

The search landscape has shifted from lists to answers, mediated by AI. Success now depends less on being listed and more on being valuable enough to inform the synthesis, and compelling enough to draw users beyond the summary to your original content.

The internet has evolved. For better or worse doesn’t matter. As businesses that want to work these channels, we must adapt. We can wish for the world to be different, but we have to work the world that exists.