Your inbox has the same email for the third time this week: "Do you do gluten-free cakes, and how much notice do you need?" The answer is on your website. It's the fourth question on the FAQ page. The customer didn't find it, or didn't look. So the AI chatbot vs FAQ page question isn't really about technology. It's about why people with a question don't use the answers you've already written, and what would get them an answer faster.
This post compares the two honestly, including where an FAQ page is the better choice, and shows how to run both without doubling your work.
What an FAQ page does well
An FAQ page is cheap, permanent and under your control. Nielsen Norman Group's article FAQs still deliver great value makes the case well. Good FAQs answer common questions so staff don't repeat themselves. They help people who are deciding whether to buy, because they show you're open about limits and policies. And they work around the clock.
FAQ pages are also good for:
- Policies that must be stated exactly. Cancellation terms, refund rules, age limits. You want the precise wording, every time.
- Scanning. Some people like to read every answer before they commit, especially for bigger purchases.
- Linking. You can send a customer a link to the exact answer.
- Search. A page that answers real questions in plain language is the kind of content Google's guidance on creating helpful, people-first content describes.
One thing has changed. In August 2023 Google announced changes to FAQ rich results: the expandable question boxes in search results are now shown only for well-known government and health websites. So adding FAQ markup no longer gets most small businesses extra space in Google. The page can still rank. It just doesn't get the special display.
Where FAQ pages fall down
The bakery email above is the common failure. FAQ pages have predictable weak spots.
- People don't find them. A visitor on a phone, landing on a product page from Instagram, may never see the link in your footer.
- Questions don't match. The customer asks "can I collect on Sunday?". Your FAQ says "What are your collection hours?". It's the same question in different words, and the customer gives up.
- Combined questions. Real questions come in pairs: "gluten-free, and how much notice?". An FAQ answers them separately, in different places.
- Answers that depend on details. "How much is a boiler service?" depends on the boiler. An FAQ can only give a range.
- They go stale. Prices and hours change, and the FAQ is the last page anyone updates.
- They can't act. Even when the answer is perfect, the customer still has to go and book, order or call.
What an AI chatbot does differently
An AI chatbot reads the question as the customer writes it and answers from the information you've given it. With Klugon's AI chatbot for websites, that information is your knowledge base: uploaded documents, pasted text such as your existing FAQ, imported web pages and a scan of your site's main pages, plus your services, products and opening hours.
The differences that matter:
- It handles the customer's wording. "Sunday collection" and "collection hours" are the same question to it.
- It answers combined questions. Both halves, in one reply.
- It asks back. For "how much is a service?", it can ask what kind of boiler before quoting your price list.
- It acts. In the same chat, it can check availability and book, or capture the person's details as a lead.
- It shows you what people ask. That's the part most owners don't expect.
A sample chat. The bakery is made up.
Notice what the assistant said about coeliacs. It said that because the bakery wrote it in its allergen policy. A chatbot is only as accurate as what you give it.
Where an AI chatbot falls down
Chatbots have their own weak spots, and you should weigh them as carefully.
- It can get things wrong. An AI answers from your knowledge, and hands over when it doesn't know, but no AI is guaranteed never to make a mistake. For anything with legal, medical or safety weight, keep the exact wording on a page and have the chatbot point to it.
- It needs feeding. If your knowledge base is thin, answers will be thin.
- It costs money. An FAQ page is free once written. A chatbot is a monthly subscription.
- Some people won't use it. Plenty of visitors prefer to read. An FAQ page serves them; a chat bubble doesn't.
- It must say what it is. In the EU, the AI Act's transparency rules for AI systems that talk to people apply from 2 August 2026 (see the Regulation on EUR-Lex). Make it clear in the welcome message that the visitor is talking to an AI assistant.
There's also a Klugon-specific limit. You can't reply to a website chat from Klugon's inbox. That works on WhatsApp, Instagram, Messenger and Telegram, but not on web chat. So when the assistant can't help, set it up to take a name and phone number or email, which it saves as a lead, and flags the conversation for your team. Your team then contacts the customer directly.
AI chatbot vs FAQ page compared
| FAQ page | AI chatbot | |
|---|---|---|
| Cost | Your time to write it | Monthly subscription plus setup time |
| Finding the answer | The customer searches or scrolls | The customer asks in their own words |
| Combined or detailed questions | Weak | Strong, if the knowledge is there |
| Exact policy wording | Strong | Best when it quotes or links to the page |
| Taking action (book, enquire) | Links out | In the same conversation |
| Search visibility | Can rank as a page | Chat content isn't indexed by Google |
| Upkeep | Edit the page | Update the knowledge base |
| Insight into what people ask | None, unless you add analytics | Every question is logged |
| Risk of a wrong answer | Low, if the page is current | Present; reduce with good knowledge and handover |
Use both: the FAQ feeds the chatbot, the chatbot improves the FAQ
You don't have to choose. The setup that works for most small businesses is both, each feeding the other.
Step 1: put your FAQ into the chatbot
Paste your existing FAQ into the knowledge base, or import the FAQ page by its URL. Klugon reads an imported page once and doesn't re-check it automatically, so add it again when you change the page. Our post on what to teach your AI receptionist covers what else to include. The same advice applies to chat.
Step 2: let the chatbot tell you what's missing
Klugon's Insights page lists "Questions without a good answer" and "Most asked questions, last 30 days". After a month, that's a list of real questions in your customers' own words, ranked by how often they come up. No FAQ page you write from memory will be as accurate.
Step 3: rewrite the FAQ from that list
Add the missing questions. Reword existing ones in the customer's language. Put the most-asked questions at the top. Then add the new answers to the knowledge base too.
Step 4: link between them
Have the chatbot link to the FAQ page for exact policies. Klugon has a links library for this, so the assistant can share your refund policy or booking page in the chat. And mention the chat on the FAQ page: "Can't find your answer? Ask our assistant."
Repeat every month or two. Each round, fewer questions go unanswered on either.
When an FAQ page alone is enough
Don't buy a chatbot you don't need. An FAQ page is probably enough if:
- you get only a few enquiries a week, and answering them is no burden
- your questions are simple and stable (hours, location, parking)
- most customers book or buy without asking anything
- your customers strongly prefer phone calls (in which case look at the phone, not the website)
When a chatbot earns its cost
A chatbot is usually worth it if:
- the same questions arrive by email, phone and social messages every day
- people ask outside opening hours and you lose them by morning
- answers depend on details (sizes, dates, which service), so an FAQ can only say "it depends"
- you want people to book or enquire in the same conversation, using AI appointment booking
A worked example
Assumptions: a dog groomer gets 15 website enquiries a week by email and contact form. Each takes 5 minutes to read and answer, including the back-and-forth. Half are questions already covered on the FAQ page.
- Time on enquiries: 15 × 5 = 75 minutes a week, about 5 hours a month.
- If a chatbot fully handles the half that are FAQ questions, that frees about 2.5 hours a month.
Whether that's worth it depends on your hourly value, and on how many of the after-hours enquiries would have gone elsewhere. Klugon's AI chatbot plans start at $49 a month for 100 conversations; see pricing. Uploading PDFs and documents needs the Business plan on the chat product; pasting text and importing pages works on every plan.
Adding a chatbot without rebuilding your site
Klugon's website chat is a script tag you paste before the closing body tag, so it works on any site that lets you add code, including WordPress and Shopify. There's no plugin and no app to install. You can set the colour and position. The script-tag chat is text; if you want visitors to be able to speak instead of type, embed or link to your hosted assistant page, which has a microphone button.
The same knowledge powers Klugon's other channels, so if customers also message on WhatsApp, the WhatsApp AI chatbot answers from the same information. If they phone, Klugon's AI receptionist can answer calls too, on the voice plans.
Frequently asked questions
Is an AI chatbot better than an FAQ page?
Neither is better in every case. A chatbot handles questions in the customer's own words and can act on them. An FAQ page is free, exact and readable. Most businesses do best with both.
Will a chatbot replace my FAQ page for SEO?
No. Chat conversations aren't indexed by search engines. Keep a well-written FAQ page for search and use the chatbot alongside it.
Do FAQ pages still help SEO after Google's 2023 change?
Yes, as ordinary pages that answer real questions. What changed is that FAQ rich results now show only for well-known government and health sites.
Can I train a chatbot on my existing FAQ page?
Yes. In Klugon you can paste the text or import the page by URL. Re-import it when you change the page, because imported pages aren't re-checked automatically.
What happens when the chatbot doesn't know the answer?
It should say so and hand over. In Klugon, it flags the conversation for your team and can save the visitor's contact details as a lead so someone can get back to them.
How do I know which questions to add to my FAQ?
Use the questions people actually ask. Klugon's Insights page shows the most asked questions and those without a good answer.
Does the chatbot need to say it's an AI?
It's good practice everywhere, and in the EU the AI Act's transparency rules now expect it. Say so in the welcome message.
If you want to see how your own FAQ performs as a chatbot, you can paste it into a Klugon assistant during the 7-day trial and ask it the questions your customers ask.