Shift 01Search Became a Conversation
There's a simple test you can run right now. Look at your site's Google Search Console and filter by query length. If your top-performing queries are still two or three words — "web design agency", "B2B website cost" — you're looking at traffic that's about to shrink. Not because you'll rank lower, but because fewer people are searching that way.
Average search query length has grown from 2.3 words in 2016 to over 4.8 words in 2025. On mobile and voice, it's even higher. The shift is structural: AI assistants trained an entire generation of users to express intent in full sentences, not compressed keywords.
The user who used to type "buy iPhone Berlin" now types "where can I find the best price on iPhone 16 Pro in Berlin with same-day pickup". The intent is clearer. The purchase signal is stronger. And the content that ranks for it looks completely different.
What this means for your website
Short keyword-stuffed headings no longer work as primary SEO signals. Google BERT and MUM — the language models powering search since 2019 and 2021 respectively — rank by intent and context, not keyword density. The sites that win the long-tail are those that write the way their customers think.
Practical shift: Rewrite your H2 and H3 headings as questions. Instead of "Web Design Services", write "How much does a custom website cost for a B2B company in Europe?" Instead of "Brand Identity", write "What does a brand identity project include and how long does it take?" These aren't blog tricks — they're how your buyers actually phrase their problem.
FAQ sections — long dismissed as filler content — are now one of the highest-leverage SEO investments. Marked up with FAQPage schema, they surface directly in the SERP as rich results, occupying twice the visual real estate of a standard listing. More importantly, they're the format AI Overviews pull from when generating answers.
The shift also changes how landing page copy should be written. Conversational, specific, question-answering copy outperforms generic benefit statements. The visitor who arrived via a four-word query is much closer to a decision than the one who typed a brand category. Your page needs to confirm their intent immediately — not make them hunt for the answer.
Shift 02Websites Stopped Showing Everyone the Same Page
For most of the web's history, personalisation meant showing your name in an email subject line. Then it meant recommending products based on purchase history. Now it means the website you see is meaningfully different from the one the next visitor sees — from the hero headline to the case studies in the sidebar to the CTA copy.
This isn't magic. It's the result of a structural shift in how companies collect and act on behavioural data. The technology behind it is the Customer Data Platform — a system that unifies every touchpoint a user has with your brand into a single profile, then feeds that profile back into your marketing and product experience in real time.
"The average e-commerce site using a CDP with behavioural personalisation sees 15–25% lift in revenue per visitor within the first six months. That's not optimisation — that's a different product."
CDP adoption tripled between 2022 and 2025 across mid-market companies. What was previously enterprise-only infrastructure — Segment, Salesforce Data Cloud, Adobe Real-Time CDP — is now accessible at the scale of a 50-person company. The cost came down. The complexity didn't fully, but the vendors absorbed most of it.
unified user profile
From segments to individuals
The old model of personalisation was segment-based: visitors from Germany get German-language content, returning visitors get a different banner, paying customers see an upsell. These are static rules applied to categories of people.
The new model is behavioural and predictive. A CDP tracks not just who you are, but what you're about to do. It ingests session data, scroll depth, time on page, previous visit patterns, email open history, and CRM status — then scores the likelihood of each next action. Is this visitor about to bounce? About to convert? Are they in a research phase or an evaluation phase?
The CJM is no longer a diagram on a wall
Customer journey mapping used to be a workshop output — a beautiful diagram showing idealised touchpoints, framed and hung in the product team's office. With CDP infrastructure, the CJM becomes a live data model. You can see in real time where people drop off, which paths lead to conversion, and where the journey deviates from what you designed.
Companies that have rebuilt their CJM around live behavioural data — rather than research-based assumptions — are identifying friction points in days instead of quarters. The output isn't a prettier diagram. It's a fundamentally faster feedback loop between what users do and what the product does in response.
The practical implication: If you're still running basic GA4 traffic reporting with audience segments as your primary analytics, you're operating with about 20% of the data available to you. The investment isn't in a bigger dashboard — it's in connecting your CRM, your site behaviour layer, and your outbound touchpoints into a single system that can act on what it sees.
Shift 03Third-Party Data Died. First-Party Data Became the Moat.
In January 2020, Google announced it would phase out third-party cookies in Chrome by 2022. Then 2023. Then 2024. In July 2024, Google reversed the deprecation decision entirely — but the market had already moved. Safari and Firefox blocked third-party cookies years ago, GDPR enforcement is increasingly aggressive, and most sophisticated advertisers had already started building alternative infrastructure.
The result: retargeting as it existed in 2019 is gone for a meaningful portion of the web. Cross-site tracking, third-party audience segments, lookalike audiences built on external data — all degraded. Campaigns that relied on this infrastructure saw CPMs rise and attribution collapse simultaneously.
What first-party data strategy actually looks like
First-party data is any information a user voluntarily gives you — directly or through their on-site behaviour. Email address, preferences, purchase history, session data with consent. The difference from third-party data isn't just legal: it's accuracy. A user who opted into your list is more precisely modelled than one inferred from cross-site tracking.
The strategic shift this requires isn't technical — it's about value exchange at every touchpoint. You can't capture first-party data without giving something in return. That something has to be genuinely worth it:
- Content that requires registration (guides, research, tools) — not just a PDF behind a form
- Loyalty programmes that give back measurable value, not just points
- Personalised calculators or configurators that require light sign-in to save results
- Email sequences that are genuinely useful — not just nurture drips with the same case study recycled four times
The companies building durable first-party data assets are treating every digital touchpoint as a negotiation: what can we offer that makes this user want to identify themselves to us? The companies that haven't started yet are increasingly dependent on paid channels they can't accurately measure.
Shift 04You Can Rank #1 and Get Zero Clicks
Google's AI Overviews — the generated summaries that appear above organic results — now appear on approximately 15–20% of all queries in the US, with broader rollout across Europe ongoing through 2025 and 2026. The effect on organic traffic is measurable and significant: click-through rates for top-ranked pages drop 30–45% when an AI Overview is present.
The model is straightforward. A user asks "what's the best way to structure B2B pricing on a SaaS website". Google generates a 300-word answer from multiple sources. The user gets what they need. They do not click. The site that ranked first for that query still gets the attribution in Search Console — as an impression — but the visit never happened.
"Ranking number one used to mean traffic. Now it sometimes means being the source an AI quotes to users who will never visit your site. The goal has shifted from rank to citation."
The same dynamic is accelerating on Perplexity, ChatGPT, and Gemini — AI-native tools that are now primary research interfaces for a growing share of knowledge workers and technical buyers. These tools cite sources, but they don't send traffic the way a ranked link does. You can be the authoritative source on a topic and receive a fraction of the visitors you would have in 2021.
What being "AI-citable" actually requires
The criteria AI systems use to select citations aren't fully public, but the patterns are clear from analysis of which content gets cited consistently:
- Structured, direct answers — AI models surface content that answers a specific question clearly, without burying the answer in preamble
- Schema markup — FAQPage, HowTo, Article schema make content machine-readable in the format AI crawlers prefer
- Authoritative domain signals — brand mentions, backlink quality, E-E-A-T (Experience, Expertise, Authoritativeness, Trust) still matter — now as citation quality filters
- Original data and named sources — AI tools prefer citing specific statistics, named studies, and original research over generic assertions
- Consistent entity presence — having your brand mentioned across multiple authoritative sources, not just ranking on your own domain
The channel implication: Zero-click search accelerates the case for owned channels — email lists, community, direct relationships — that don't depend on a click from the SERP. Brands that built direct subscriber bases over the last five years are significantly more resilient to AI Overview cannibalisation than those whose entire traffic model depends on organic clicks.
| Objective | Old playbook | 2026 playbook |
|---|---|---|
| Visibility | Rank position 1–3 | Be cited by AI + rank |
| Traffic | Organic clicks from SERP | Owned channels + direct |
| Content format | Keyword-dense articles | Structured Q&A + schema |
| Authority signals | Backlinks & domain rating | Entity mentions + E-E-A-T |
| Success metric | Impressions & clicks | Conversions + brand recall |
Where This All Goes: The Intelligent Website
Look at all four shifts together and a single pattern emerges. They're not independent trends — they're four expressions of the same underlying change.
Search got smarter, so users started talking to it like an intelligent assistant. Personalisation infrastructure got cheaper, so websites started responding to users as individuals. Data collection got restricted, so the relationship between brand and user became the asset, not the cookie. AI took over information delivery, so ranking became a means to citation rather than an end in itself.
The brands that understood this early didn't adapt to each trend separately. They rebuilt around a single premise: a website is no longer a brochure that waits to be found. It's an adaptive system that knows who's visiting, anticipates what they need, answers their question before they finish asking it, and captures enough context to make the next interaction better than the last.
The gap between companies that have absorbed these shifts and those still running 2020-era digital playbooks is widening. Not because the laggards are doing digital wrong — but because the baseline moved under them while they were optimising for the old metrics.
Is your digital setup built for how the web works now?
We work with companies who want to close the gap — in search strategy, personalisation infrastructure, and content architecture. Start with a conversation.