Conversational search
Conversational search is searching by talking: asking a full question, then follow-ups, in one ongoing chat with an AI assistant.
Conversational search is searching by having a conversation. Someone asks an AI assistant a full question, reads the answer, and asks a follow-up, and then another, in the same chat. Each new question builds on what came before.
It changes what being found means. With a search box, you needed to show up for one query. In a conversation, you need to show up in the first answer and still be there after the follow-ups have narrowed things down.
How a conversation narrows
For example, someone asks: "What are good coffee subscriptions in the UK?" The answer names five brands, Fernhill among them. They follow up: "Which of those deliver weekly?" Three are left. Then: "Which is best for an office of 20?" The answer now recommends Kestrel.
Fernhill was in the first answer and dropped out by the third. That's where the work is: the details that decide the later turns, like delivery options and office plans, need to be clear wherever the platform reads about you.
What's different about the questions
Questions in a conversation are longer and more specific than search queries. People write the way they talk, and they add details as they go. Later questions also lean on earlier ones: "Which of those deliver weekly?" only makes sense after the first answer.
In AnswerOrbit
We ask each of your prompts once a day on each platform, as someone who isn't signed in, so each prompt is the start of a conversation. To follow the later turns, write prompts that carry the details someone would add in a follow-up, like "best coffee subscription for an office of 20 with weekly delivery." See How we count and Topics and prompts.