Why Website Traffic Is Almost Dead Today

What It Covers

Neil Patel and Eric Siu analyze the collapse of website traffic across major publications — Digital Trends down 97%, ZDNet down 90% — and examine how AI is forcing businesses to pursue creative destruction, using Intercom's pivot to Fin as a case study in rebuilding from near-zero growth to 400M ARR.

Key Questions Answered

• •Traffic collapse by content type: Informational content (how-to guides, definitional articles) is experiencing the steepest traffic declines — Digital Trends lost 97% of visits between March 2024 and January 2026. Transactional and navigational content remains viable because AI tools cannot complete purchases on behalf of users, meaning bottom-funnel pages still drive measurable revenue.
• •Revenue over traffic as the primary metric: Tracking website visitors as a success metric has been misleading for years. Even before AI disruption, LTV, profitability, and revenue were the metrics that mattered. If an AI agent completes a transaction on your behalf without a site visit occurring, the revenue outcome is identical — the visit itself carries no inherent value.
• •Topic authority still requires comprehensive coverage: Even when informational content generates zero direct traffic, publishing thorough topical coverage on your website remains necessary. AI systems and Google use it to establish authority and source citations. The strategy shifts from targeting high-volume informational keywords to building credibility that earns AI-generated referrals.
• •Intercom's creative destruction playbook: Intercom recovered from 4% growth to 37% growth at 400M ARR by redirecting 80% of R&D to Fin while it represented single-digit revenue, replacing the entire board with startup founders, launching a separate brand at fin.ai, and routing 100% of paid media to the new product — all before the new line proved itself financially.
• •AI team structure for agencies and service businesses: The emerging operational model pairs marketing strategists with AI automation engineers who build agent workflows using tools like Claude and open-source models. Running local models on hardware such as Mac Studio with 512GB unified RAM reduces API costs — one team's Anthropic spend reached several thousand dollars monthly before exploring on-premise alternatives.