Live chat

Live chat is the fastest synchronous support channel short of a phone call. A customer clicks a bubble in the corner of your site and expects a reply in seconds to minutes rather than hours, which is what makes it effective at the purchase moment, when the question standing between someone and their credit card is usually small and the alternative to an answer is closing the tab.

The cost of live chat is not the number of chats, it is the interruptibility, and that is the calculation small teams get wrong. Suppose you get 60 chats a month. Spread over 22 working days that is under three a day, and at ten minutes of attention each it is barely half an hour of work. But to answer them in seconds someone has to stay reachable for the whole eight hours, which means the real price is one person losing deep focus every day to handle thirty minutes of chat. For a team of two to ten where the same people also build the product, that trade is often bad, and it is invisible until you have already promised it.

This is why almost every serious chat tool is really hybrid, or async chat. When nobody is available, the conversation is captured and continues by email instead of sitting unanswered in a widget the visitor has already closed. Run it that way deliberately: put your real hours on the widget, say plainly what happens outside them, and answer in one thread rather than starting a new one by email. Customers are fine with "we will email you back this afternoon." What they are not fine with is a chat box that implies someone is there when nobody is.

The other thing that makes chat affordable at small scale is answering the easy part without a person. A widget that searches your published help center and offers the matching article resolves the "what are your shipping times" class of question at the moment it is asked, which is exactly the class that would otherwise interrupt you. That only works if the articles are good, so chat is downstream of your knowledge base rather than a substitute for it.

Where chat clearly pays for itself is on high-intent pages. Pricing, checkout, and signup are where a thirty-second answer changes an outcome, and where a four-hour email reply is worth nothing because the visitor is gone. Where it pays least is on documentation pages and in your app for existing customers, who will generally accept email and often prefer it, because they want a written record. A reasonable starting configuration is chat on the pages that decide a purchase and email everywhere else.

AcornReply includes a live chat widget that installs as one script tag on your site. It is genuine two-way chat with live delivery, attachments and screenshots, read receipts, and unread counts, not a contact form dressed as a bubble. Every chat is an ordinary conversation in the same shared inbox as your email and form messages from the moment it starts, so there is no separate chat console to sit in and nothing to merge afterwards.

You choose how much the AI does on that channel: nothing, offer matching help center articles only, reply whenever it can, or reply only when nobody on the team is available, which is driven by your own availability schedule and can also show an online or offline indicator to visitors. Only curated, publishable knowledge is ever shown to a customer, and answers captured from past conversations stay private to drafting. After an AI answer the widget asks whether it helped, and a thumbs down or a request for a person reopens the conversation for a human and stops the AI re-engaging. If the visitor has closed the widget by the time you reply, the reply is delivered by email instead, so the thread continues either way.