Every few months, someone declares SEO dead. Then traffic reports come in, and most sites still depend on organic search. What changed is not whether SEO matters. It is where visibility shows up, what gets clicked, and what kind of content earns trust from both Google and the AI systems quoting the web. If you still optimize like it is 2019, you will keep publishing pages that technically rank but never get seen. ## Search results stopped being a list of links For years, SEO meant one goal: rank high enough that people click your result. That model is incomplete now. A single query can produce: - Classic organic results - AI Overviews (Google's AI-generated summary at the top) - Featured snippets and People Also Ask boxes - Video carousels, local packs, and shopping units - Answers inside ChatGPT, Perplexity, Claude, and Copilot Users often get their answer without visiting any site. [SparkToro and Similarweb research](https://sparktoro.com/blog/2024-zero-click-searches-up-organic-clicks-down/) has tracked zero-click searches climbing for years. AI summaries accelerated that trend. **The shift:** Ranking #1 and getting traffic are no longer the same problem. You can rank well and still lose the click to an AI answer above you. ## Two discovery layers now exist Think of modern search visibility as two overlapping systems: | Layer | What it optimizes for | Where you show up | | --- | --- | --- | | **Classic SEO** | Crawlability, relevance, authority, UX | Google/Bing results, maps, video | | **Answer visibility** | Clear facts, citations, brand mentions, structured content | AI Overviews, chatbots, voice assistants | Classic SEO is still the foundation. If Google cannot crawl, index, and understand your pages, AI systems will not cite them either. They pull from the same open web. But answer visibility adds new requirements: content that machines can quote accurately, entities they can recognize, and sources they can trust. You do not pick one. You need both. ## AI Overviews rewrote the top of the SERP Google's AI Overviews (formerly Search Generative Experience) summarize answers and link to a small set of sources. Early data showed they appear on a meaningful share of queries, especially informational ones. What this means in practice: 1. **The top of the page is crowded.** Even a #1 organic result sits below summaries, ads, and widgets. 2. **Being cited matters as much as ranking.** AI Overviews link to sources they used. Getting into that citation set is a new form of winning. 3. **Thin content loses twice.** Pages that barely answer the query get skipped by both users and AI summaries. Pages that win citations tend to have direct answers near the top, clear headings, original examples, and enough depth that a summary cannot replace the full page. ## Chatbots became a referral channel (sort of) ChatGPT, Perplexity, and similar tools do not work like Google. They synthesize answers and sometimes link to sources. Traffic from these tools is still small compared to traditional search for most sites, but it is growing fast on some publishers. What seems to help content get cited by LLMs: - **Named entities and clear definitions.** If you explain a concept cleanly, you become easy to quote. - **Original data and first-hand experience.** Surveys, benchmarks, case studies, and "we tried this" write-ups stand out against generic rewrites. - **Consistent terminology.** Pick one term for a concept and use it throughout. Ambiguity makes models less likely to attribute you. - **Author and brand signals.** Pages tied to a real person or company with a track record get treated differently than anonymous content farms. Do not optimize for chatbots by stuffing pages with FAQ blocks written for robots. The same quality bar applies: helpful, specific, human-readable content wins in both systems. ## E-E-A-T went from buzzword to baseline Google's quality guidelines have long emphasized Experience, Expertise, Authoritativeness, and Trustworthiness. AI-generated content at scale made that bar more visible. Search engines and AI systems both lean on signals that a source knows what it is talking about: - Author pages with credentials and history - About pages that explain who runs the site and why - Citations to primary sources - Content that reflects real experience, not recycled summaries - Accurate, maintained information (dates, version numbers, pricing) A generic "ultimate guide" with no point of view is easier to produce than ever. It is also easier to ignore. If you cannot answer "why should anyone trust this page over the other ten," AI summaries will not save you. ## Keyword strategy got more specific Broad head terms still matter for brand awareness, but AI summaries often satisfy those queries directly. The opportunity moved toward: - **Long-tail and question-led queries.** "How do I fix X when Y happens" beats "X tool." - **Comparison and decision content.** Users still click when they need nuance AI glosses over. - **Local and transactional intent.** "Near me," pricing, availability, and product pages still drive clicks. - **Topic clusters.** One strong pillar page plus supporting articles signals depth better than isolated posts. Search intent matters more than keyword density ever did. Match what the user is trying to accomplish, not just the string they typed. ## Technical SEO did not go away If anything, technical quality matters more when machines parse your content. Still non-negotiable: - **Crawlability:** Clean robots.txt, valid sitemap, no accidental noindex tags - **Core Web Vitals:** Slow pages lose users and ranking signals - **Structured data:** JSON-LD that matches visible page content (see next section) - **Canonical URLs:** Duplicate and parameterized URLs still dilute signals - **Internal linking:** Orphan pages rarely rank or get cited - **Mobile experience:** Mobile-first indexing is the default A beautiful content strategy on a broken site still fails. Fix the plumbing first. ## Structured data: code you can actually ship Structured data is not a ranking hack. It is a label on your page that tells crawlers and AI systems: "this is an article," "this person wrote it," "these are the steps," "this is the price." When the JSON-LD matches what users see on the page, you get richer search results and clearer signals for citation. When it lies (fake reviews, hidden FAQ text, wrong dates), you risk manual actions and lost trust. Here is practical code for the schemas most sites need. ### 1. Reusable JSON-LD component (Next.js) Drop this in `components/json-ld.tsx` and reuse it on every page type: ```tsx type JsonLdProps = { data: Record | Record[]; }; export function JsonLd({ data }: JsonLdProps) { return (

Templates, ready to clone.

PERFUMA — Fragrance Ecommerce