If you’ve been doing SEO for more than a few years or working with an agency that has, you’ve probably built strategies around practices that felt rigorous and data-driven at the time. Topic clusters. Intent mapping. Content refreshes. Featured snippet optimization.
None of these were bad ideas. But a lot of them have quietly stopped doing what they used to do, and in some cases they’re actively pulling attention away from what actually works right now.
What Changed, and When
The shift didn’t happen overnight. But two things accelerated it: the rollout of AI Overviews in Google Search, and Google’s release of an official guide to optimizing for generative AI features in May 2026.
That guide is worth reading in full if you haven’t. What makes it notable is it’s what it explicitly tells you to stop doing. Google calls out llms.txt files, content chunking, long-tail keyword page proliferation, and inauthentic mention-building as things you can ignore. Those are practices a significant chunk of the SEO industry has been selling as “AEO strategy.”
But before we get to the AI era, there’s a layer of pre-AI practices that also deserve a closer look.
Featured Snippet Optimization as a Standalone Strategy
For several years, there was an entire discipline around capturing position zero. Practitioners were studying snippet formats, reverse-engineering word counts, restructuring pages around the 40-to-60-word paragraph answer. It worked often enough to become standard practice.
The problem: AI Overviews now absorb much of the query space that featured snippets used to serve. The click value of a snippet has dropped significantly as AI-generated summaries take up more of the page. Optimizing specifically for snippet format rather than for genuine helpfulness was always a fragile strategy. Now it’s even more fragile.
Topic Clusters Built for Architecture, Not Audience
The pillar-cluster model was a genuinely useful framework when it was introduced. Build a broad authoritative pillar page, surround it with cluster content on related subtopics, connect everything with deliberate internal links.
What happened in practice: teams started building clusters to satisfy a content architecture diagram rather than because their audience needed all those pages. A lot of cluster content was thin, overlapping, and existed primarily to signal topical authority through volume. Google’s guidance on this is now explicit. A high quantity of pages doesn’t make a website higher quality or more relevant to users, and creating pages primarily to manipulate rankings violates the scaled content abuse spam policy.
The framework isn’t dead. But clusters built around what your audience genuinely needs are a different thing entirely.
Search Intent as a Checklist Rather Than a Question
Intent classification became almost ritualistic. Every content brief started with: is this informational, navigational, commercial, or transactional? Match the format to the intent type, write to the prescribed length, add a FAQ section, publish.
The classification itself isn’t wrong. The problem was that it became a box-ticking exercise. Intent matching tells you what kind of page to build. A lot of content passed the intent test and failed the “is this genuinely useful” test. Google’s systems, including the AI systems, have gotten significantly better at distinguishing the two.
Content Length Targets
Correlational research showing that top-ranking pages averaged 1,800 or 2,500 words got absorbed into SEO culture as causal guidance. Word count became a KPI. Teams were padding posts with redundant H2s, expanding FAQ sections not because the questions were useful but because they added length, writing longer introductions to get to the threshold faster.
Word count was never a ranking signal. It was a proxy for comprehensiveness that got misapplied as a target in itself. Google’s AI guide makes this clear from a different angle: there’s no ideal page length. Make pages for your audience, not for a word count tool.
Cosmetic Content Refreshes
“Update your content annually to keep it fresh” became a standard recommendation. In practice it often meant re-publishing posts, bumping the date, adding a paragraph at the top, and moving on. Google’s freshness signals care about substantive updates; new data, corrected information, meaningfully expanded coverage. They don’t respond to cosmetic changes, and AI systems absolutely don’t.
What the AI Overviews Era Actually Asks For
Google’s AI optimization guide is worth reading not as a list of new tactics but as a clarification of what has always mattered and why shortcuts to signal those things no longer work.
The central ask is non-commodity content. Google defines this explicitly: commodity content is based on common knowledge, could come from anyone, and adds little unique insight. Non-commodity content provides expert or experienced perspectives that go beyond what’s already widely available, including what a generative AI model could produce itself.
That last part is the real shift. If your content isn’t more authoritative, more specific, or more experiential than what an LLM could write in 30 seconds, you’re competing against AI for your own citation. That’s a competition you won’t win on volume or format alone.
What wins: first-hand experience, proprietary data, local specificity, genuine expertise, and a distinct point of view. These things can’t be scaled cheaply, which is precisely why they’re what the systems now reward.
The other signal worth paying attention to: spam policies now explicitly apply to generative AI responses in Google Search. Google confirmed this in May 2026. The tactics that manipulated traditional results; scaled AI content, inauthentic mentions, thin pages dressed up as topical coverage face the same enforcement in AI Overviews that they face everywhere else in Search.
What This Means if You’re Running a Business
You don’t need to rebuild everything. The foundational work: crawlable site structure, clear page metadata, legitimate link acquisition, well-organized content, still matters and hasn’t changed.
What’s worth examining is whether your content strategy is built around signal-chasing or substance. Some honest questions to work through:
- Does your content reflect something you actually know, or does it restate what’s already widely available? AI Overviews synthesize the widely available stuff. What earns a citation is what stands apart from it.
- Are your pages built around what your customers actually need, or around filling out a content calendar? Volume without genuine usefulness is exactly what Google’s scaled content policies are designed to catch.
The good news: businesses that have actual expertise, real customer relationships, and genuine local or industry knowledge are exactly the kind of sources Google’s AI systems are trying to surface. The opportunity is real but it just requires building something worth citing rather than optimizing something that looks like it should be.
If you’re not sure where your current strategy stands, we’re happy to take a look. Newfound Marketing works with businesses across North America on SEO and AI visibility strategy.
