You write content, you optimise your pages for specific keywords, and yet your traffic curves stagnate. Quite often the cause lies in a site structure designed page by page, keyword by keyword, while Google now tries to understand what the person wants to accomplish.
The days when stuffing a page with keywords was enough to climb the search results are over. Today, algorithms no longer simply try to match terms, they try to understand what the user really means. This is the shift from keyword to search intent.
This paradigm shift forces companies to rethink their website architecture entirely. AI is not only there to generate text. It becomes the invisible architect that connects your content to the real needs of your prospects.
A structure built around intent is, in practice, a site where each page answers a single need: understanding a topic, finding a specific piece of information or taking action (requesting a quote, buying, booking). Pages that serve the same topic are grouped around a pillar page and linked to each other, so that the visitor moves naturally from the question they are asking to the page that answers it, then to the one that lets them act. The keyword does not disappear: it is used to name the page and to check that it matches a real search, but it is intent that decides its existence, its format and its place in the site tree. In this work, AI is an analysis tool: it helps sort hundreds of queries by intent and spot the topics the site does not cover, with the decision remaining human. Here is how we approach this transformation, step by step.
The limits of keyword-based SEO
For years, the strategy was simple: one keyword, one page. That mechanical approach led to well-known excesses. You ended up with bloated websites, filled with duplicate or low-quality content, written for robots rather than for humans.
The “Keyword Stuffing” problem
Over-optimisation, or “keyword stuffing”, was long the technique of choice. The result? A disastrous user experience. Imagine landing on a page where every sentence artificially contains the term “best digital agency Lausanne”. Not only is it unreadable, it is now penalised by search engines that favour fluency and relevance.
A degraded user experience
This focus on the exact keyword often obscured what mattered: answering the searcher’s question. A visitor looking for “roof renovation price” does not have the same expectation as one typing “how to repair a roof tile”.
If your site structure does not distinguish these nuances because it focuses only on the generic keyword “roofing”, you lose conversion opportunities. The visitor leaves without having found an answer, and the page fulfils neither its commercial role nor its editorial role.
FURTHER READING: SEO and AI: should you still optimise every page individually?
Understanding user intent
To structure a modern site, you first need to understand the motivations behind each search. Google generally sorts intentions into three broad categories, and your site must have distinct sections to answer each of them.
| Type of intent | Description | Search example | Ideal content format |
|---|---|---|---|
| Informational | The user is looking for an answer or wants to learn. | “How to improve your SEO” | Blog articles, guides, tutorials |
| Navigational | The user is looking for a specific site or page. | “LinkedIn login” | Home page, “About” page |
| Transactional | The user is ready to buy or to act. | “Roof renovation quote Lausanne” | Product pages, landing pages, quotes |
Identify and address every need
The classic mistake is to mix these intents on the same page. A product page (transactional) should not be buried under 2000 words of historical explanation (informational). Conversely, an educational blog article should not be an aggressive sales pitch from the very first line.
This is where AI comes into play: it makes it possible to analyse thousands of queries in order to classify these intentions precisely and structure your site accordingly.
How AI is redefining site structure
Artificial intelligence does not simply read words; it analyses context, behaviour and search patterns. It makes it possible to move from a linear structure to a semantic one, which is far more powerful.
Predictive behavioural analysis
Analytics tools, increasingly assisted by AI, show how visitors navigate your site, and SEO tools reveal the queries your competitors cover. They detect friction points: where do people click? Where do they stop reading? What questions do they ask next?
This data makes it possible to create smooth user journeys. If the analysis shows that a good share of visitors reading an article on “thermal insulation” then look for “cantonal subsidies”, your site structure should reflect that logical link through optimised internal linking.
Creating topic clusters
This is probably AI’s most significant contribution. Instead of isolated pages, AI suggests creating “Topic Clusters”.
The principle is simple:
- A pillar page covers a broad subject in depth (e.g. “digital marketing”).
- Satellite pages (cluster content) cover specific sub-topics (e.g. “SEO”, “Email marketing”, “Social advertising”).
- Hyperlinks systematically connect these pages to one another.
AI can analyse your existing content and suggest which pages to group to strengthen the coherence of a topic. It also helps spot the “holes” in your content (content gaps) to fill in order to cover a theme completely. We detail this logic of pillar page and satellite pages in our article on the topic cluster.
Examples of AI-driven strategies
Take the example of a SaaS software company. Thanks to AI, it can analyse conversations on social media and forums to detect an emerging intent not covered by its competitors.
If the tool detects a rise in questions about “integrating AI into CRMs”, the company can quickly deploy a content cluster dedicated to that specific topic before the keyword becomes saturated. That is a proactive strategy, no longer a reactive one.
Implementing intent-based SEO
In practical terms, how do you transform your current site? Here is a four-step method for aligning your structure with user intent.
Step 1: AI-assisted semantic audit
Use tools such as SurferSEO, SEMrush or an agency’s own solutions to scan your site. The point is not to look at keyword rankings, but at thematic coherence.
- Do your pages answer a single intent clearly?
- Are the pages on the same subject linked to each other?
Step 2: Restructuring into silos
Group your content. If you sell sports shoes, do not mix training advice with product specification sheets in the same URL category. Create clear silos:
- /shop/running/hoka-shoes (Transactional)
- /blog/running-advice/how-to-choose-your-shoes (Informational)
AI can help you map these URLs; the final validation rests with those who know your customers and your offering.
Step 3: Optimisation of existing content
Ask an AI (such as ChatGPT or Claude) to analyse your most important pages. Give it the content and ask: “What is the main intent of this text? Is it missing crucial information for a user looking to [insert the goal]?” You will be surprised by the gaps it identifies.
Step 4: Dynamic internal linking
This is often the poor relation of SEO. Use algorithms to identify the missing links between your pages. Good internal linking guides the user towards conversion naturally. If a reader is on a comparison guide, the AI should suggest inserting a link to the winning product of the comparison, because the intent shifts from informational to transactional.
Two typical scenarios: betting on intent
To make the method concrete, here are two typical scenarios. The companies are imaginary, the approach is the one we apply.
Scenario 1: an electronics shop in Lausanne
Take a Lausanne online shop specialising in electronics, whose traffic is stagnating. Looking at the queries that lead to the site, we find that the product pages are very technical and speak only to connoisseurs, while many potential buyers are still hesitating between two models.
The action: create a set of simple buying guides (informational intent), each linked to the relevant product pages, and review the product pages so that they answer the questions of a buyer ready to order. What we observe next: in Search Console, we check whether the guides appear on comparison queries that the product pages did not reach, and in the statistics, whether readers of the guides move on to the products. It is this transition we follow, more than the volume of visits.
Scenario 2: an HR software publisher in Fribourg
Imagine a Fribourg publisher of payroll software that struggles to rank on a highly contested query such as “payroll software”.
The action: rather than fighting over this term, the team uses AI to group long-tail questions, such as “how to automate payslips for a Swiss SME”, and builds a complete cluster around payroll in SMEs, with a pillar page and articles that each answer one question. What we observe next: fewer visitors than with a generic term, but whose questions match the product. The follow-up looks at the demo requests coming from these pages.
The future of SEO: towards hyper-personalisation
We are only at the start of this revolution. The future of SEO is already being played out in Google’s AI Overviews (formerly SGE), rolled out in the United States from May 2024, which display direct, synthetic answers above the traditional results.
Your site will no longer be visited for raw information, but for unique expertise and an in-depth experience. AI will keep refining its understanding of human nuance. The sites that win will be those that have managed to build a flexible architecture, able to adapt in real time to micro-changes in search intent.
Key takeaways
Structuring a site by intent is no longer reserved for big players, and AI makes the analysis accessible to an SME. Moving from a keyword logic to an intent logic takes the work beyond SEO: it forces you to clarify what you offer, and to whom.
You align your offer with your clients’ real demand. You create a smooth experience that reassures, informs and converts.
The first step remains an assessment of your current structure: which pages answer which intent, which compete with each other, which are missing. If you would like to carry it out with us, let’s talk about it in a first conversation.




