An owner asks me this every few months. Should we put chat on the website. Our competitor has it. The chat vendor's sales rep showed us a number about conversion rates.
My answer is the same every time, and it is a question rather than a recommendation. Who is going to answer it, and how fast?
Everything about whether chat works at a 10 to 50 person B2B company comes down to that. Get the answer right and chat is one of the better things you can add. Get it wrong and you have installed a device that publicly demonstrates that your company does not respond.
The one condition
Chat is a live channel. A visitor who types into it expects a person within a minute, because that is what every other chat window in their life has taught them.
So the condition is a named person, during stated hours, who is at a screen and has chat notifications turned on, and who can answer a real question about what you do and what it costs. That is a job assignment rather than a piece of software.
At a 30-person company that person is usually an office manager, an inside sales person, or an estimator. It works when the role already sits at a desk and already handles enquiries. It fails when the plan is everybody keeps an eye on it, because a shared responsibility with no name attached is nobody's responsibility, and it will be ignored inside three weeks.
If you cannot name the person and the hours right now, stop reading about chat and skip to the alternatives at the end of this article. They will serve you better.
What an unanswered chat actually does
Here is the sequence. A visitor with a real project types hello, do you do seismic upgrades on tilt-up warehouses. The widget shows typically replies in a few minutes. Nothing happens. They wait, they check back once, they leave.
That visitor has now learned something specific about your company, and they learned it from a live demonstration rather than from a claim. They asked a direct question and got silence. Before the chat window existed they would have read your service page and either called or not. Now they have a reason to call someone else.
The damage is bigger for the people who care most. A casual browser does not open a chat window. The person who opens it has a specific question, which means they have a live project, which means they were the enquiry you wanted.
I see abandoned chat widgets on B2B websites in Metro Vancouver constantly. Usually a tool installed with enthusiasm during a website build, live for two months, and now quietly collecting messages nobody reads. If that describes your site, the cheapest improvement available to you today is removing it.
Where a bot helps and where it does not
An automated chat window can do a small number of jobs well, and it fails badly outside them.
It does well at routing. Are you asking about a new project, an existing job, an invoice, or a job application. Four buttons, and each one leads to the right form or the right email address. This saves your office manager real time and it saves the visitor from sending a project enquiry to accounts.
It does well at answering fixed facts. Service areas, business hours, whether you are licensed in a particular jurisdiction, whether you handle residential work, what your minimum project size is. Facts that never change and that people ask constantly.
It does well at booking. Show a calendar, let them pick a slot, confirm by email. For a professional services firm this is often the whole value of having chat at all.
It does badly at anything that requires judgment, and it does badly at pretending to be a person. A bot that answers how much would a 12,000 square foot fit-out cost with a generic paragraph about your commitment to quality is worse than a form, because the visitor now knows they are talking to nothing. Have the bot say plainly that it can pass the question to a person, take their contact details, and give an honest time for a reply.
The AI assistants now built into these tools are better at understanding a typed question than the old keyword bots were. They still invent answers when the information is not in what you gave them, and at a B2B company the invented answer is usually about scope, price, or whether you can do something. That becomes a commercial problem in a way that a wrong answer about opening hours does not. Keep the automated part to facts you have written down, and hand off everything else.
The pages where chat earns its place
The default install puts the window on every page. That is the vendor's default because more coverage produces more chats, and more chats is their measure rather than yours.
Put it on three kinds of pages. Service pages, where someone is reading about the work and forming a specific question. Pricing or process pages, where the question is usually about fit and money. Landing pages behind ads, where you paid for the visit and a live answer protects the spend.
Keep it off your blog, your careers page, and your homepage. A blog reader is reading. The careers page will fill your sales chat with application questions. The homepage visitor has not yet formed a question, and a window that slides out over the page before they have read anything reads as an interruption. If you want to understand what else is quietly costing you enquiries on those pages, a page-by-page review of where visitors drop out is a better place to spend the effort.
Set the window to sit quietly in the corner rather than opening itself. Auto-opening with a message that pretends to be a person typing annoys the exact buyer profile you want, which at a B2B company is usually an engineer, a facilities manager, or a procurement lead.
Speed is the whole game
A chat answered in 30 seconds is a conversation. The visitor is still on the page, still holding the question, still in the frame of mind that made them type.
A chat answered in ten minutes is an email with extra steps. They have closed the tab. Your reply arrives in a notification they may or may not see, and the exchange becomes a slow back and forth over hours.
So the operational decision is about the hours you promise, not the tool. Better to state chat available 8am to 4pm Pacific, weekdays and honour it, than to leave the window up all the time and answer when convenient. Outside those hours, switch the widget to a form that says the team is offline, asks for the question and an email address, and commits to a reply time you will actually meet.
Track one number: how long the average first reply took. If it drifts past two minutes during your stated hours, either the person is too busy or the hours are too wide. Both are fixable and neither is fixed by changing the software.
The cost you cannot see
A chat widget loads a script from another company's servers on every page it appears on. On a site that is already slow, this is noticeable, and it is worst on a phone on a mobile connection at a job site, which is where a lot of B2B browsing now happens.
The widget also collects and stores the conversation, plus whatever personal details the visitor types, on the vendor's infrastructure. If you handle any customer information that carries obligations, your privacy policy needs to mention the tool and you need to know where that data lives. Canadian companies dealing with government or healthcare clients should check where the vendor stores it before installing, because that question tends to arrive during a procurement review at the worst moment.
And the widget needs permission in your website's security settings to load at all. This trips up more installations than anything else, and the symptom is a widget that works for the developer and appears for nobody else. Have whoever manages your site confirm it is loading for real visitors on real devices after install, rather than assuming.
Two alternatives that suit most companies better
A callback request beats chat at a lot of B2B companies. A small button that says request a callback, a form with a name, a number, and a one-line description, and a promise: we will call you within two business hours. It has no live staffing requirement, it produces a phone conversation rather than a typed one, and a phone conversation is where B2B deals actually progress. For trades and manufacturing especially, the buyer would rather talk than type.
A booking link beats chat for professional services. Accountants, consultants, engineers, recruiters, anyone whose next step is a scheduled conversation. Put a calendar on the service page, let them pick a 20 minute slot, send a confirmation. No live staffing, no response clock, and you get a qualified meeting instead of a chat transcript. If you add one, make sure the questions on the booking form do real work, which is the same thinking as a good set of discovery questions.
Both of these convert intent into a scheduled commitment. Chat converts intent into a conversation that has to be maintained. For a company without a person assigned to maintaining it, the commitment is worth more. The wider question is how a visit becomes an enquiry.
The takeaway
Install live chat when you can name the person answering it and the hours they are there, and when that person can answer a real question about scope and cost. Put it on service, pricing, and paid landing pages only, keep it out of the way until someone clicks it, and measure first reply time rather than chat volume.
If you cannot name that person today, use a callback request or a booking link instead. Both capture the same intent, neither requires anyone to be watching a screen, and neither can sit unanswered in front of a buyer who had a live project.
And if there is a chat widget on your site right now that nobody has read in months, take it down this week. An empty chat window is a promise your website keeps making on your behalf and failing to keep.