Organic search clicks have dropped sharply since AI Overviews began answering informational queries inside Google Search: publishers now compete for a smaller pool of clicks while AI-generated and scaled content flood the results. That shift forces a simple question for content teams: which top-of-funnel (TOFU) formats still earn organic traffic, and where should you stop publishing?

What the recent publisher data shows

Search Engine Land compared Semrush organic traffic estimates for August 2026 vs. August 2025 across 30 major publishers in nine industries. The pattern is clear: TOFU traffic fell most heavily in YMYL categories (finance, healthcare, legal), while certain B2B tech verticals — notably cybersecurity, cloud infrastructure and marketing/sales software — saw TOFU gains.

Importantly, middle-of-funnel (MOFU) pages — product comparisons and other “vs.” or “best” queries — moved in parallel with TOFU within the same industries. In short: funnel stage mattered less than industry context; both top- and middle-funnel visibility rose or fell together for a given vertical.

Why some industries lost clicks and others didn’t

Google’s AI summaries favor fast, authoritative answers for queries where users seek an immediate outcome. That behavior explains steep TOFU declines in YMYL topics: when users ask about symptoms, taxes or legal definitions, the search interface increasingly provides a concise summary that reduces the incentive to click through. Even large, authoritative publishers in those verticals lost significant traffic — a worrying signal for smaller sites.

B2B tech topics behave differently because they reward depth, recentness and operational detail. Buyers researching migrations, architectures or vendor tradeoffs need granular, experience-based content an LLM trained on older data may not supply — which preserves demand for authored TOFU content in those areas.

TOFU formats that still perform in 2027

The dataset and related analyses point to several practical TOFU formats that typically attract sustainable visibility and clicks:

  • Local-first TOFU: AI Overviews and LLMs are less reliable with location signals; AI summaries also appear less frequently for local queries. That leaves organic room for genuinely local content — not templated pages, but guides tailored to the needs and constraints of specific service areas.
  • Interactive tools and calculators: Ungated, domain-specific tools (calculators, screeners, checkers) produce unique outputs that AI summaries cannot reproduce. Studies of AI-referred traffic show tool pages frequently pick up citations and referrals, making tools a high-leverage TOFU format.
  • SME-led content and original data: Content informed by subject-matter experts — with firsthand experience, tradeoff analysis or original data — resists the flattening effect of LLM summaries. Specialist insight, case-based nuance and opinionated analysis make pages more likely to be cited and clicked.
  • Prudent AI-assisted production: Fully automated, high-volume AI copy remains risky. Improvements in long-form generation have slowed, model watermarking and detection are increasing, and search quality updates target low-value scaled content. AI assistance with rigorous human editing is the safer path.

These approaches are complementary: a location-aware tool built and explained by internal SMEs combines multiple protective features that make pages harder for AI summaries to replace.

How to act now

Start by mapping your industry to these patterns. If you operate in a YMYL vertical, accept that TOFU returns may be structurally lower and reset expectations. If you serve B2B tech buyers, continue TOFU investment but prioritize depth, timeliness and practitioner insight.

Next, audit TOFU pages with a new filter: does this page deliver unique utility an AI summary cannot supply? Practical criteria include original data, interactive outputs, local specificity or clear SME value. Consolidate or remove pages that fail the test and reallocate effort toward tools, localized content and hard-to-replicate expertise.

Finally, track AI-driven search behavior for your topics: monitor where AI Overviews appear, which domains they cite, and whether your pages are referenced by AI or picked up as referrals. Those signals show which content types are still earning visibility and should inform ongoing investment decisions.

The 2027 SERP is narrower and more automated than the one most teams planned for. That raises the bar: not every TOFU page will attract traffic, but material that AI cannot reproduce — location-aware content, interactive tools and true SME insight — still earns clicks and influence.