TLDR: AI didn’t kill blogging—it raised the bar. Search (and the new wave of “answer engines”) still rely on high-quality, linkable, attributable content. Blogs are the best vehicle you control to build topical authority, earn links, feed AI answers with your brand’s perspective, and convert the visitors you do get. Below I’ll unpack how modern SEO strategies works, what’s actually happening to clicks, and the specific reasons blogs remain essential—plus how to write posts that win in the AI era.
What AI search really does under the hood (and why it still needs blogs)
Most modern AI systems are built on transformers, a neural-net architecture that learns patterns across tokens via attention. That’s the “self-attention” you hear about, and it’s the core of GPT-style models behind today’s generative answers.
There are two distinct phases to understand:
- Pretraining: Models learn general language patterns from huge text corpora (web pages, books, Wikipedia, etc.). A large share of those corpora come from open web crawls like Common Crawl (plus licensed sources). In other words, the model’s baseline language ability is literally shaped by what’s published on the web. If the web goes thin, models get blander; if the web is rich with original, first-hand material, models get more useful.
- Retrieval / the “RAG” era: For fresh, specific answers, many systems (including AI Overviews and chat search features) pair the model with retrieval-augmented generation (RAG)—the model embeds your query, fetches relevant documents from an index (vectors + traditional signals), and then generates an answer conditioned on those sources. RAG improves factuality and gives provenance. That pipeline still depends on indexable, high-quality pages—i.e., your blog posts.
It’s also why AI companies have begun licensing premium content (from major newsrooms, communities, and publishers): authoritative source material matters. If your brand publishes original, useful blog content, you’re building the raw material AI and search systems prefer to surface and cite.
Are clicks really disappearing?
Clicks are changing, yes—but that’s context, not a death sentence.
- AI summaries and zero-click growth: Independent analyses show zero-click behavior has increased since AI summaries rolled out; several data sets and media studies indicate users click through less often when an AI answer appears at the top. That means you must earn the fewer clicks available with pages that demonstrate unique value.
- Publisher traffic pressures: Trade groups and publishers have reported noticeable declines in Google referrals tied to AI summaries, though the impact varies by site and category.
- Google’s counterpoint: Google has said overall referrals are “relatively stable” in aggregate—so the mix of who wins/loses is shifting more than the total. Translation: thin content loses; authoritative, helpful content can still win.
- AI referrals are growing: Traffic from AI platforms is rising fast (though still smaller than classic search), which again rewards content that’s clear, citable, and authoritative.
Bottom line: There are fewer “free” clicks. Strong blogs still earn them—and now earn visibility inside AI answers.
Also checkout our free bulk keyword rank checker if you’re interested in checking where your blogs are ranking.
“Is ChatGPT just sourcing from Google?”
The plumbing is evolving, but here are the facts to keep in mind:
- Officially, ChatGPT Search relies on a web index (notably Bing’s) plus other sources to fetch real-time results and provide citations.
- In practice, external tests in 2025 have shown scenarios where ChatGPT returned answers seemingly derived from content indexed by Google, suggesting it may sometimes access Google-sourced data directly or indirectly via partners. These are experiments—not formal disclosures—but they illustrate that LLM answers are still tethered to web-published content and the search ecosystems indexing it.
- Strategically, reporting indicates OpenAI has sought broader search access and content licensing. The signal is consistent: LLMs depend on robust web indexes—and on the content those indexes crawl.
Whether the path is Bing, Google, or a hybrid, the through-line remains: your blog is the primary way your brand’s expertise enters the knowledge graph that AI systems draw from.
10 reasons blogs still matter for SEO in 2025+
- They build E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness). Google guidance and rater docs emphasize helpful, people-first content created by identifiable experts. Blogs are the best format to show first-hand experience, cite sources, and demonstrate depth.
- They’re the backbone of topical authority. A focused cluster of posts signals breadth + depth on a subject, supporting rankings across head, mid-tail, and long-tail queries—still crucial as AI Overviews summarize from those trusted sources. Since 2024, Google has also folded “helpful content” signals into core ranking systems, explicitly cracking down on thin, scaled output.
- They feed AI answer engines. RAG pipelines, AI Overviews, and chat-search cite and condense crawlable, well-structured posts. If your perspective isn’t published, it can’t be surfaced or cited.
- They win complex and long-tail intent. AI summaries often handle “obvious” facts; users still click for nuance, methodology, comparisons, pricing context, and fresh data—the exact things great blogs provide. As zero-click rises, the remaining clicks skew toward posts with truly differentiated value.
- They future-proof your first-party RAG. Many brands are deploying internal search/chat using their own vector DBs. Your blog becomes a curated knowledge base you own, ready to power support, sales enablement, and product discovery via your own AI.
- They earn links and mentions. Editorial backlinks remain one of the strongest signals; original research, data visuals, and “how we did it” posts attract citations that compound authority.
- They unlock structured data and richer SERP features. Clear headings, Q&A sections, tables, and schema improve eligibility for enhanced results and make your content easier to retrieve and quote in AI contexts.
- They support internal linking and crawlability. Topical clusters and hub-and-spoke architectures improve discovery and distribute PageRank—important as updates since 2024 continue to demote thin, scaled content and reward helpfulness.
- They convert and capture community. Blogs nurture email signups, remarketing audiences, and product education—value that persists even as pure SEO traffic fluctuates.
- They are licensable assets. As platforms license “trusted” archives, durable, high-quality posts become both monetizable and more likely to be included in attributed AI answers.
How to write blogs that win in the AI era
- Lead with first-hand experience + unique insight. Demonstrate work, data, customers, failures, samples—anything an LLM can’t fabricate. Tie every insight to a named author with real credentials. This maps directly to E-E-A-T.
- Be “answer-engine ready.” Open with a crisp summary (TL;DR), include a structured table of contents, short Q&A blocks, and clear subheadings. These are easy for retrieval systems to chunk and cite.
- Cite sources—and be citable. Use clear outbound citations and link to your own related posts. Add original charts/tables; include plain-English captions. Retrieval systems gravitate to well-structured, easily chunkable content.
- Publish clusters, not one-offs. Map a topic and ship a logical series. Interlink to a canonical hub. This boosts discovery and topical authority.
- Avoid scaled, low-value content. Recent updates have targeted “scaled content abuse,” “expired domain abuse,” and “site reputation abuse.” Stick to quality, originality, and usefulness.
- Use schema where relevant. Article, FAQ, and HowTo (when appropriate), plus clean meta data, help both search and AI retrieval route the right snippet to the right question. Keep it people-first.
- Measure beyond last-click. Track assisted conversions, newsletter growth, and qualitative signals (mentions, links, citations in AI answers). Traffic from AI surfaces is growing—configure analytics to recognize it.
The takeaway
AI changed the distribution of clicks—but didn’t change the need for credible, attributable, original expertise on the open web. Blogs remain your most flexible, ownable, and index-friendly format to build authority, earn links, power RAG (yours and theirs), and inject your perspective into the answers people see. In 2025 and beyond, the winners won’t publish more posts; they’ll publish better ones—deeper research, first-hand experience, clearer structure, and tighter topic focus.
Sources
- Google Search Central – Creating Helpful, People-First Content
https://developers.google.com/search/docs/fundamentals/creating-helpful-content - Google Search Central – SEO Starter Guide (Search Essentials)
https://developers.google.com/search/docs/fundamentals/seo-starter-guide - WhitePress – Mastering Google’s Helpful Content Guidelines in 2025
https://www.whitepress.com/en/knowledge-base/2227/google-helpful-content - Vaswani et al. (2017) – Attention Is All You Need
https://arxiv.org/abs/1706.03762 - Wikipedia – Attention Is All You Need (contextual overview)
https://en.wikipedia.org/wiki/Attention_Is_All_You_Need - Lewis et al. (2020) – Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
https://arxiv.org/abs/2005.11401 - Wikipedia – Retrieval-Augmented Generation (definition & context)
https://en.wikipedia.org/wiki/Retrieval-augmented_generation - NVIDIA Blog – What Is Retrieval-Augmented Generation?
https://blogs.nvidia.com/blog/what-is-retrieval-augmented-generation/ - Wall Street Journal – Vector Databases & RAG (background)
https://www.wsj.com/articles/how-a-decades-old-technology-and-a-paper-from-meta-created-an-ai-industry-standard-354a810e - Time – Profile of Patrick Lewis & RAG
https://time.com/7012883/patrick-lewis/ - Wired – The Transformer Revolution (“Attention Is All You Need”)
https://www.wired.com/story/eight-google-employees-invented-modern-ai-transformers-paper
