Conversational search

Conversational search is a way of searching in which people ask full questions in natural language and follow up in the same session, while the engine keeps the context of earlier turns. Google AI Mode, ChatGPT search, Gemini and Microsoft Copilot work this way, as Google's SGE experiment did in 2023.

Why conversational search matters

Conversational search changes the shape of the query. "Queries in AI Mode are three-times longer than traditional searches," Sundar Pichai said in February 2026, and a significant portion of them now lead to a follow-up question. Google has also made it possible, since January 2026, to ask a follow-up straight from an AI Overview and continue in Google AI Mode.

Longer questions also pull AI answers onto the page. In Pew Research Center's March 2025 data, 60% of Google searches that began with "who," "what," "when" or "why" produced an AI summary, against 8% of one- or two-word searches. Each turn of a conversation is a new answer, and a new chance for a brand to be included or dropped.

Keyword query vs conversation

	Keyword search	Conversational search
Typical input	A few words, such as "booking software clinic"	A full question with context and constraints
Memory	Each search starts from zero	Earlier turns carry over
Refining	Retype a new query	Ask a follow-up
Output	A list of links	A written answer with sources
What a brand competes for	A ranking on one page	A place in every answer of the thread

The memory row is the real difference. OpenAI's launch post for ChatGPT search puts it this way: "Go deeper with follow-up questions, and ChatGPT will consider the full context of your chat." Google's 2023 announcement of its SGE experiment already said: "Context will be carried over from question to question."

Example: one conversation, three turns

A physiotherapist chats with an AI assistant:

"Which online booking system suits a physiotherapy clinic with three practitioners?" Say the answer names five tools.
"Which of those send SMS reminders to patients?" The engine keeps "physiotherapy" and "three practitioners" and narrows the list.
"Which one is cheapest for three users?" It now looks for pricing pages of the remaining tools.

Because the follow-ups ask about "those" tools, a tool missing from the first answer is out of the thread, and a tool with no public pricing gives the third answer nothing to compare.

How to measure your brand in conversational search

Start with the first turn, where the shortlist is set. Write the prompts you track the way buyers talk, with their constraints: team size, budget, location, the tools they already use. Then check the follow-ups that matter most by hand. Hawkscan shows the full answers that ChatGPT, Gemini, Microsoft Copilot and Google AI Mode give to the prompts you track, every day, with the brands named and the pages cited. For how one question becomes many searches behind the scenes, see query fan-out.