Discovery is
changing shape.
Public research shows AI assistants moving into everyday use and sending more people to websites. These charts provide market context. They do not describe BGraph coverage, customers, or proprietary datasets.
AI-assisted discovery is growing quickly, especially in commerce and travel, but traditional search remains much larger. The useful question is no longer whether behavior is changing. It is where the change appears first and what happens after the AI interaction.
Three public views
of the same shift.
Each chart uses the source's published definition and period. Values from different sources should not be combined into one market-size estimate.
Use roughly doubled from 2023 to 2025. The 2026 survey expanded the question to include other chatbots, so that point is marked separately.
Source: Pew Research Center ↗Year-over-year change during the U.S. 2025 holiday season. Adobe reports that its retail analysis covers more than 1 trillion visits.
Source: Adobe Digital Insights ↗+357% YoYAI referrals in June 2025
AI referrals were about 0.6% of Google referral volume in this estimate. The signal is growing fast, but classic search still sets the scale.
Source: Similarweb ↗Use the numbers,
inspect the method.
Updated 8 August 2026. Every value comes from a cited public source or is marked as a BGraph calculation. No proprietary BGraph data is included.
Machine-readable source values and identified calculations.
Download ↓02 / MDMethodologyScope, source treatment, transformations, and reproducibility notes.
Download ↓03 / PNGChart image
Publication-ready chart image with source credit.
Download ↓04 / TXTCitationPreferred citation and original source credit.
Download ↓BGraph Research. (2026). AI search market signals. Updated 8 August 2026. https://bgraph.io/research/ai-search-market-signals. Public-source compilation using Pew Research Center, Adobe Digital Insights, and Similarweb.
- The three publishers use different populations, periods, devices, and definitions. Values must remain source-separated.
- Survey adoption does not measure query volume, commercial intent, or referral traffic.
- Referral growth starts from a smaller base and is not equivalent to total AI answer exposure or market share.
- Vendor traffic figures are estimates. They are not BGraph coverage figures or customer results.
What these signals can and cannot show
Adoption is not query volume
A survey can show how many people report using a tool. It does not reveal how often they use it, what they ask, or which commercial decisions follow.
Referral growth is not market share
A high growth rate can start from a small base. Referral visits measure clicks leaving AI platforms, not every answer viewed inside them.
Behavior connects the gap
To understand market impact, teams need the sequence around discovery: earlier research, channel movement, downstream destinations, repeat behavior, and aggregated outcomes.
Move from public trends
to a focused signal.
Define the category, geography, platforms, and outcomes that matter to your team. BGraph can shape delivery around the question and permitted data scope.
Discuss a dataset ↗Evidence you can inspect
Use the source link
AI search intelligence from adoption to downstream behavior
A practical guide to AI search intelligence, GEO measurement, citations, referrals, adoption, downstream journeys, product boundaries, and public evidence.
Continue ↗ProductAI search signals without personal conversation claims
Measure AI platform adoption, category movement, channel shifts, and downstream behavior using aggregated signals.
Continue ↗Use caseAI search intelligence built from aggregated behavior
Measure AI search adoption, category discovery, and search journey patterns using aggregated, privacy-secured behavioral signals.
Continue ↗IndustryPlan around the journeys audiences actually take
Use observed market journeys to improve channel planning, publisher research, content strategy, and competitive analysis.
Continue ↗