Live chat and chatbots: worth it for a small business site?
The chat bubble in the bottom right corner has become a default, which is exactly why it deserves a second look. On some sites it produces a steady stream of enquiries that would otherwise never have happened. On others it sits there unanswered for three days, quietly telling every visitor that nobody is home. The difference is not the software.
What chat actually changes
Chat lowers the cost of asking a question. A phone call is a commitment and a contact form feels like paperwork, while a chat message costs the visitor almost nothing.
That matters most for people who are early in their decision and have one small blocking question. Do you work in my area, do you handle this brand, can it be done this week. These are people who would not have filled in a form, and that is where the additional enquiries come from.
It also shortens the gap between interest and contact. Someone reading your pricing page at nine in the evening can ask right then, while the interest is live.
There is a second, less obvious benefit. The questions people ask in chat are a free list of everything your website fails to explain, and reading a month of transcripts is one of the cheapest content audits available.
What chat does not do is fix a weak offer or replace a proper contact route. It is an additional door, not a substitute for the ones you already have.
And it only works under one condition, which the next section is entirely about.
The response time problem
Chat sets an expectation of immediacy simply by existing. A visitor who sees a chat bubble assumes someone is there, and an unanswered message does more damage than no chat at all.
Be honest about which mode you can sustain. Live chat means a real person answers within a couple of minutes during stated hours. Asynchronous chat means messages arrive like email and you reply later. Both work. Pretending the second is the first does not.
State your hours in the chat window itself. Outside them, switch the widget to a message form that asks for an email address and says clearly when you will reply.
If you cannot answer within a few minutes during working hours, do not run live chat. A message left waiting until tomorrow afternoon costs you the customer and tells them something about how you handle everything else.
For a one person business, the practical setup is a mobile app with notifications and honest hours, for example weekdays from nine to five, plus an out of hours form. That is entirely credible and takes very little effort to maintain.
The one habit that makes all of this work is capturing an email address early in the conversation, because a visitor who closes the tab is gone forever, while a visitor whose email you have is still a lead.
Bot or human
A bot has one genuine strength: it answers instantly at three in the morning. Everything else about it is a compromise, and the compromise is worth making only in specific situations.
Bots work well for repetitive factual questions where the answer never changes: opening hours, location and parking, delivery times, which brands or areas you cover, how to track an order.
They work badly for anything requiring judgement, and worse still for anything emotional. A frustrated customer looping through a menu of options they did not want is a customer you are actively losing.
The pattern that works is a bot that answers a handful of common questions and hands over to a human quickly and visibly, with an obvious way to reach a person at every step.
Language models have made bots considerably better at understanding what was asked, but they introduce a specific risk for a small business: a bot that confidently invents a price or promises a delivery date is worse than no bot. Restrict it to your own documented information and test it with awkward questions before it goes live.
For most small businesses the honest answer is that a bot is unnecessary. The volume of questions does not justify it, and a human answering within ten minutes during working hours beats an instant automated reply on every measure that matters.
Where to put it and where not
Placement is conventionally bottom right, and there is no reason to be creative about it. Visitors know where to look, and moving it only adds friction.
- Keep it small until it is used. A widget that expands automatically and covers the content is a common cause of people leaving a page.
- Never let it cover the phone number, the main call to action or the mobile navigation, which is a frequent and expensive layout bug.
- Check it on a phone, where the same bubble takes a far larger share of the screen than it does on a desktop.
- Make it dismissible, and remember the dismissal so it does not reappear on every page.
Proactive messages, which open automatically after a delay, are worth using sparingly and only where they make sense: on a pricing page, on a long service page, or on a checkout step where people abandon. Firing one on every page after five seconds is the digital equivalent of being followed around a shop.
And watch the weight. Some chat widgets load several hundred kilobytes of scripts and measurably slow the page, which trades a small number of conversations for a slower site for everyone. Check the impact in PageSpeed Insights before and after installing.
Cost, privacy and measuring the result
Free tiers exist and are genuinely usable for a small business, typically covering one or two agents with basic history. Paid plans start around fifteen to thirty euros per agent per month, and the features that justify the upgrade are usually saved replies, mobile notifications and integration with whatever you use to track clients.
The real cost is not the subscription, it is attention. Chat only works if someone is actually watching it, and that time has to come from somewhere.
Privacy matters here more than people expect, because chat transcripts contain personal data and often quite sensitive detail. Choose a provider that stores data in the European Union or has a proper transfer basis, mention chat in your privacy policy, and set a retention period rather than keeping transcripts forever.
If the widget sets cookies or loads before consent, it belongs behind your consent banner along with everything else that does.
Measure four things after a month: how many conversations started, how many turned into real enquiries, your median response time, and what share of conversations happened outside working hours. That last number tells you whether an out of hours form or a bot would actually earn its place.
Then make the decision honestly. If chat produced a handful of conversations and none became customers, remove it. A missing chat widget costs nothing, while an abandoned one damages trust on every single page of your site.
FAQ
Do I need live chat if I already have a contact form and a phone number?+
Not necessarily, and the deciding factor is whether you can answer quickly, because chat adds value only by being faster and lighter than the routes you already offer, and it removes value the moment it is slower. It genuinely helps in two situations: when your service involves small blocking questions people want answered before committing, such as coverage area, brand compatibility or availability this week, and when a meaningful share of your visitors browse in the evening, after your phone line has closed. If you cannot commit to answering within a few minutes during stated working hours, a well designed contact form with a clear promise about response time will serve you better, because it sets an expectation you can actually meet.
Should I use an AI chatbot instead of answering myself?+
For most small businesses the honest answer is no, because the volume of questions rarely justifies the setup and the maintenance, and a human replying within ten minutes during working hours outperforms an instant automated reply on every measure that matters to a customer. A bot earns its place in two specific cases: when you receive the same handful of factual questions repeatedly, such as opening hours, location or delivery times, and when a significant share of your traffic arrives outside working hours and would otherwise get nothing at all. If you do use one, restrict it strictly to information you have documented yourself and test it with awkward questions before launch, because a bot that confidently invents a price or a delivery date creates a problem considerably more expensive than the one it was meant to solve.
Does a chat widget slow down my website?+
It can, and the difference between providers is large enough to be worth checking rather than assuming, because some widgets load several hundred kilobytes of scripts on every page while others are a fraction of that. Measure it directly: run PageSpeed Insights on a key page before installing and again afterwards, and pay particular attention to the mobile score, since that is where the impact is most visible and where most of your visitors are. If the difference is significant, there are two straightforward fixes, namely loading the widget only after the page has finished rendering, which most providers support as a setting, or loading it only on the pages where it actually earns conversations, such as pricing and service pages, rather than on every article in your blog.