Only if a person will answer it within a few minutes, during every hour the widget says you are online. If that is not true, the bubble in the corner is not a feature. It is a promise you are about to break in front of a customer who was ready to talk.
The question is not whether chat is good. It is whether you can staff it.
What does a chat bubble actually promise?
A reply, now.
That is the whole reason it exists. A phone number says "call us". A form says "we will get back to you". A chat bubble says "someone is here". The visitor who clicks it has chosen the fastest path you offered, and they are holding their question in their head, waiting for the little dots.
Everything that follows depends on whether the dots appear.
What happens when nobody answers?
The visitor waits, then leaves, and they leave with a worse opinion of you than if the bubble had never been there.
Think about what they saw. A green dot that said online. A greeting that used their first name if the widget could get it. Then silence. Then, sometimes, a form asking for their email so you can "follow up", which is the thing they were trying to avoid by using chat in the first place.
You do not see any of this. Chat dashboards show you the conversations that happened. They are much quieter about the ones that opened and stalled. The customer you lost does not send a complaint. They call the next business on the list.
Who can actually keep the promise?
Businesses where somebody is already sitting at a screen during opening hours, with the authority to answer.
A shop with staff at a counter between customers. A clinic or a firm with a receptionist who already handles the phone. A B2B supplier with a sales desk that is at a keyboard all day anyway. An online store where the questions before checkout are about sizing, stock and delivery, and one person can answer those from a list.
Notice who is not on that list. A trades business where everyone is on site. A restaurant during service. A one-person consultancy where the person is in meetings. A salon where every pair of hands is busy. For these businesses a live-chat widget is a bet that the owner will glance at their phone in time, and the owner loses that bet on the busiest days, which are exactly the days the enquiries come in.
The test is simple. Name the person who will answer chat at 2 p.m. on a Tuesday, and again at 4:45 p.m. on a Friday. If you cannot name them, do not install it.
Is a chatbot the way around that?
Sometimes, for a narrow job. Not as a stand-in for a person.
A bot that answers the same handful of questions every visitor asks is useful. Opening hours, whether you serve their town, whether you take insurance, what a first visit involves. Those answers are already written down somewhere, and a bot that hands them over at 11 p.m. is doing something a form cannot.
What it cannot do is quote a job, judge a case, tell someone whether their symptom is urgent, or make a decision. The newer AI assistants are more fluent than the old button-menu bots, and that fluency is the problem. Ask one something it does not know and it will answer anyway, in confident sentences, and the visitor will treat that answer as yours. A wrong price from your chatbot is a price you now have to argue about.
If you use one, fence it. Give it your written answers and nothing else. Make it say plainly when a question is outside what it knows, and hand over to a real contact path at that point instead of guessing. And read its transcripts every week, because they are the most honest record you will ever get of what customers actually want to know.
There is a quieter lesson in those transcripts too. If the bot answers the same question all day, that question belongs on the page as plain text, where a search engine can find it and nobody has to type anything. That is the argument in does your business need a blog, and it applies here word for word.
What does the widget cost, beyond the subscription?
Weight and attention.
The chat script is a third-party file loaded on every page from a server you do not control. It sits alongside the analytics, the pixel and the review embed, and together they can be the reason a site that should be quick feels slow on a phone. That trade is covered in why your website is slow, and a chat widget is not a small passenger.
Attention is the other cost. On a phone the bubble covers part of the screen, and on some sites it covers the button that mattered. It pops open a greeting a few seconds after arrival, which interrupts reading. And it pulls at the owner too, because every ping on their phone is either a question worth answering or a spam bot testing the widget, and they cannot tell which without looking.
What should you offer instead?
Contact paths you can keep, with the reply time written next to each one.
A phone number that is a real link, tapping straight into a line somebody answers. If nobody answers the phone either, that is a different problem, and a chat widget will not solve it.
A text or WhatsApp link with an honest promise beside it. "Message us, we reply within the hour on weekdays" is a promise a busy owner can keep from a van, and a chat widget is not. The visitor gets the thing they wanted from chat, which is a conversation on their phone, without you pretending to be at a desk.
A short form that says what happens next. A few fields, and a line above the button: "You will hear from us by the next business day." Then keep to it. If your form already exists and enquiries feel thin, check that it is actually delivering before you add anything else. The test is in contact form emails not arriving.
Answers on the page. Prices, or at least ranges. Service area. What a first appointment involves. Who to talk to. A visitor who finds the answer does not need to chat at all, and a visitor who still wants to talk is now much closer to buying.
If what people really want is a time to speak with you, that is a booking problem rather than a chat problem, and it has its own trade-offs. They are in online booking on your website.
If you do add it, how do you set it up so it works?
Make the widget tell the truth, and route it to a phone that someone carries.
- Set the hours to when a person is genuinely watching, not to your opening hours. If that is 10 to 3 on weekdays, say 10 to 3.
- Outside those hours, show the offline state honestly and turn the widget into a message form with a stated reply time. Do not show a green dot at midnight.
- Send notifications to one named person's phone, and make answering part of that person's job. Notifications that go to a shared inbox go to no one.
- Turn off the automatic greeting that pops open on arrival. Let the visitor start it.
- Once a week, message your own site from a phone that has never contacted you and count the minutes until a human replies. That count is the only honest measure of whether the widget is helping.
If the weekly test keeps coming back with no reply, take the widget down. A site with a phone number and a form is not missing anything. A site with a chat bubble nobody answers is.
What does this cost to fix?
Not much, if the site is otherwise working. Removing a widget, rewriting the contact section with honest reply times, and adding a WhatsApp or text link is a short job, and the contact paths above are standard parts of any build we do.
If the site itself is the reason enquiries stall, our marketing sites and landing pages run from $1,200 to $8,000 and store and B2B builds start at $2,900. The ranges are on pricing and how a build gets scoped is on web design.
Whether visitors are contacting you at all, and by which path, is the first thing to look at before deciding on chat. The checklist for that is in traffic but no leads.
If you want to know what your contact paths are actually promising and whether any of them are losing people, send us the URL for a free teardown. We will tell you which path a first-time visitor is likely to pick, where it goes quiet, and what to put there instead, whether or not you work with us.