Chatbots have been around for years, but the arrival of large language models has transformed what AI chatbots can do. Instead of following rigid scripts, modern AI assistants can understand everyday questions and answer in natural language. Many businesses are wondering whether they should add one to their website.
The honest answer is: it depends. Here is how to tell whether a chatbot will help your customers or simply get in their way.
Where AI chatbots work well
- Answering repetitive questions. Opening hours, delivery areas, returns, pricing guidance and “do you offer…?” questions can be answered instantly, day or night.
- Helping visitors find information on large websites with lots of pages, guides or products.
- Collecting enquiry details, such as the type of job, location and timescale, before passing them to your team.
- Out-of-hours cover, so evening and weekend visitors get a response rather than silence.
- Internal use, helping staff search policies, procedures and product information.
Where they cause problems
- Making things up. AI models can produce confident but incorrect answers, known as hallucinations. A chatbot that invents a price, a policy or a promise can cause real trouble, and businesses are generally responsible for what their chatbot tells customers.
- Blocking access to people. Few things frustrate customers more than being trapped in a conversation with a bot when they need a human.
- Complex or sensitive conversations. Complaints, bespoke quotes, legal or medical issues and emotionally charged situations need a person.
- Slowing the site down. Some chat widgets load a lot of code on every page, affecting speed and Core Web Vitals.
Photo by Charanjeet Dhiman on Unsplash
How to introduce one responsibly
- Ground it in your own content. The safest chatbots answer only from information you provide, such as your website pages, FAQs and policies, rather than from the AI model’s general knowledge. If the answer is not in your content, it should say so.
- Always offer a human route. Make it easy to request a call back, send a message to the team or see your phone number at any point.
- Be upfront. Tell visitors they are talking to an AI assistant.
- Protect personal data. Check where conversations are stored, whether they are used to train the provider’s models and how that fits with UK GDPR and your privacy notice. Avoid collecting sensitive information through the chatbot.
- Review conversations regularly. Read a sample each week to catch wrong answers and spot gaps in your website content.
- Start small. Launch on a few pages or for a limited set of questions, then expand once it is proving useful.
Is it worth it for your business?
A chatbot tends to pay off when you receive a high volume of similar questions, when your team’s time is stretched or when visitors frequently arrive outside office hours. If you get a handful of enquiries a week and each one needs a tailored conversation, a clear contact page and a quick reply from a real person may serve you better.
It is also worth asking whether the questions people put to a chatbot reveal something about your website. If the same question keeps coming up, the answer probably deserves a clear place on the relevant page.
Beyond the chat window
Some of the most useful applications of AI are behind the scenes rather than in a chat bubble: summarising enquiries, drafting replies for staff to check, sorting incoming requests or extracting information from documents. These can often be added to existing systems as part of wider operational automation.
If you are considering AI for your website or business processes and want an honest view of what will work, let’s talk.
