Every study below is computed from the AI assistants running on live Chatonbo customer websites: what visitors ask, when they ask, how many become leads or orders. Each page states its data set, window and method, and is free to quote with attribution (CC BY 4.0).
Monthly, one row per completed month
Share of chats after hours, on phones, becoming leads, and the top topics, computed every month from the assistants running on live customer sites. Same definitions every month, with a JSON endpoint.
7.7% of conversations are in a language the site does not speak
The language of the first visitor message on live sites: the share not in English, how often a visitor writes in a language other than the site’s own, which sites see it, and whether those visitors leave contact details.
91% of store conversations include a product card
How often a store assistant shows a product inside the chat, how early the card appears, and how often a conversation with a product card ends in an order from the same visit.
39% of chats start outside business hours
Fourteen figures from live assistants and published customer results: what visitors ask, when, how many leave contact details, how fast an assistant goes live. Every number carries its data set and date.
12.9% of conversations end with an email or phone number
Lead rate by site type, conversation length, hour, weekday and device, and how many messages a visitor sends before leaving contact details. Service sites convert more chats than stores.
58% of chat-attributed orders are placed within an hour of the chat
What happens after a shopper talks to an AI assistant on a store: order rate, time from chat to checkout, what buyers asked about, and the share of orders placed from a phone or after hours.
44.7% of storefront chats happen after hours
The annual report on storefront conversations, March to September 2026: volume growth, devices, hours, lead rate by conversation depth, topics, and time from signup to a live assistant.
32.7% of conversations are product questions
The baseline study: 90 days of real website conversations classified by topic, first message, hour and length, with the method every later study reuses.
Conversations are taken from the live websites of Chatonbo customers only. Internal workspaces, test chats and app store review installs are excluded. Hours are the visitor's local time. A lead is a conversation in which the visitor left a name, email or phone number. Topics come from the assistant's own classification. Each study names its window and sample in its methodology section, and the Chat Index publishes completed months only.
Journalists and analysts: every page has a "Cite this" block with the line to quote and a formal citation. Store and site owners: enter your website and Chatonbo drafts an assistant from your content in about two minutes.
Try it on my site