Google AI Overviews Stole Your Traffic. Here's What You Can Actually Do About It.

In early 2025, Google began rolling out AI Overviews to the majority of English-language searches. By mid-2026, they appear on over 47% of all queries. For many content teams, organic impressions stayed flat while clicks dropped between 15% and 64%. The traffic is not gone. It is being consumed by a single answer box that sits above everything you worked to rank for.

There are two coherent responses: accept the new reality and optimize to be cited inside the Overview, or shift investment toward query types the Overview cannot easily answer. This guide walks through both paths and gives you a diagnostic framework to identify which queries in your portfolio belong in each bucket.

What actually happened to your traffic

AI Overviews synthesize an answer at the top of the search result page. When users get a complete answer there, they have less reason to click through to any individual source. The effect is not uniform. Informational queries where the Overview can fully satisfy the question (definitions, how-to basics, simple comparisons) take the worst hit. Transactional queries and queries requiring recent information see smaller impacts.

A common mistake is treating all traffic decline as an AI Overviews problem. Pull your Google Search Console data and filter for queries where your average position is 1–5. If clicks-per-impression dropped sharply for those queries without a position change, AI Overviews are the most likely cause. If position dropped simultaneously, that is a standard ranking problem unrelated to Overviews.

Which query types are most affected

Query type AI Overview frequency Avg click-rate impact

• →Definition / what-is
• →Very high (80%+)
• →-52 to -64%
• →How-to (basic)
• →High (65%+)
• →-38 to -48%
• →Comparison (generic)
• →High (60%+)
• →-30 to -40%
• →Best X for Y
• →Medium (40%)
• →-18 to -25%
• →Commercial / transactional
• →Low (15%)
• →-6 to -12%
• →Branded queries
• →Very low (5%)
• →-2 to -5%
• →News / time-sensitive
• →Very low (8%)
• →-3 to -6%

The two viable strategies

Strategy 1: Get cited inside the Overview. If a query generates an AI Overview for your most important content, optimize to be one of the 3–5 sources cited inside it. Being cited still drives qualified traffic — users who click through from an Overview source link are self-selecting for deeper engagement. This requires applying GEO signals: concise direct answers, FAQ Schema, and structured data that makes your content easy to extract.

Strategy 2: Shift to queries the Overview cannot answer. Overviews struggle with: proprietary data, original research, strong first-person opinions, pricing for specific configurations, and comparison tables with unusual variables. Invest in content that requires something the LLM cannot synthesize from generic training data.

How to get cited inside AI Overviews

Google's Overview system is a RAG (retrieval-augmented generation) pipeline. It retrieves candidate sources and synthesizes from the best-structured ones. The structural signals that predict Overview citations are identical to the signals that predict ChatGPT and Perplexity citations.

• →Place the direct answer to the query in the first 50 words of the page
• →Use FAQ Schema for every question your article addresses
• →Keep paragraphs under 150 words with a single answerable claim per paragraph
• →Add a structured comparison table if your content is comparative
• →Include the exact query phrasing in at least one H2 heading

Google gives preference to sources it already trusts based on entity authority signals. The Organization, Person, and BreadcrumbList schema types help Google verify your site as a known entity. Implement them if you have not already.

Content that lives beyond the Overview

Some content formats are structurally resistant to being absorbed by an AI Overview because they require interaction or provide value the Overview cannot replicate. These include:

• →Interactive tools and calculators — users must visit the page to use them
• →Original research with proprietary data — cannot be synthesized from public training data
• →Community and forum content — individual experience answers that AI cannot fabricate credibly
• →Product and service detail pages — transactional intent users want current pricing and configuration, not a summary
• →Visual-first content — diagrams, step-by-step screenshots, embedded video walkthroughs

The 5-step diagnostic process

• →
  1. →Identify the drop
• →In Search Console, filter for date ranges before and after May 2025. Sort by impression-to-click ratio change for pages with stable or improved position.
• →
  1. →Categorize affected queries
• →For each high-impact query, manually search it. Note whether an AI Overview appears and whether you are cited inside it.
• →
  1. →Prioritize by intent
• →Separate informational from transactional queries. Apply Strategy 1 (get cited) to informational, Strategy 2 (query shift) to transactional where possible.
• →
  1. →Apply GEO signals to informational pages
• →Run the pages through RankAsAnswer to identify the specific structural and schema gaps blocking Overview citation.
• →
  1. →Track citation rate, not just traffic
• →Measure whether your content appears inside Overviews for target queries. This is your leading indicator before traffic recovers.

Related Google AI Overviews: A Complete Guide for Content Marketers Related Zero-Click Search in 2025: Threat or Opportunity?