Where Data Tells the Story
© Voronoi 2026. All rights reserved.
AI chatbots already tend to provide a much shorter list of recommendations to buyers than traditional search engines do. But just like on a search engine's results page, being recommended first leads to considerably more clicks in AI search.
According to a new analysis of AI chatbot smartphone recommendations by Searchable, the AI visibility platform, users in the US market are most likely to be recommended Apple devices in response to general and 'all-rounder' questions about which phone is best to buy.
However, for specific buyer categories, Apple fails to come top. Ask about a budget, a camera or a battery instead, and it drops out of first place entirely.
First-place recommendations were concentrated in roughly a quarter of all devices that surfaced. The chatbots named 1,159 different devices in total as part of their recommendations, but only 295 of them ever came recommended in first place.
Which phone lands at the top of the answer can also depend heavily on which chatbot you ask. Across the eight categories the gap between the highest and lowest engine, measured on that category's own leading phone, runs to 57 percentage points at the median. The Google Pixel 10a is named first in 87.6% of Gemini's $300–$700 answers and just 7.8% of ChatGPT's.
The chatbots tested were ChatGPT, Google Gemini and Google AI Mode. Between them, they cited 331,480 sources from 2,584 domains, and just over four in ten of those citations came from only ten websites. TechRadar and Tom's Guide were cited in 9% and 7% of answers respectively.
Searchable prompted ChatGPT, Google Gemini and Google AI Mode over 30,000 times with 184 unique phone-buying questions, which covered: under $300, $300–$700, camera, battery, flagship, foldable, gaming, and general all-rounder questions. The number one placed recommendation was then counted and ranked by model and by brand. Data collected July 2026.