Key takeaways
- The strongest chatbots do one narrow, high-volume job: order status, FAQs, lead capture or routing. The wide ones are rare.
- Every example keeps a human handoff. Klarna is the clearest case: it automated most chats and then made humans easy to reach again.
- Support bots report resolution rates of 73% to 90%. Lead generation bots report engagement and lead counts instead, so compare each one on its own metric.
- A bot is only as good as its sources. Gecko Hospitality’s support manager reads Lyro’s answers every day, and the bot has held up.
- Results are company-reported. Read “resolution” as the share of chats closed without a person, not as customer satisfaction.
Burker, a watch brand with about 20 employees, now handles 10,000 chats a month with an AI chatbot. A Mexican discount retailer got more than half of its website visitors to talk to a bot. Klarna’s assistant took two-thirds of its support chats in its first month, and then Klarna brought human agents back for the hard cases. A good chatbot does one job well, and these 12 websites show which jobs work.
Below are 12 chatbot examples from real websites. Nine run on Tidio, across e-commerce, real estate, recruiting, finance and retail. Three are large brands (Klarna, Bank of America and Alaska Airlines) that show the same patterns at scale. Each example says what the bot does, the result the company reports and one idea you can copy.
Jump to: Summary table · Support chatbots · Sales and lead generation chatbots · Flow chatbots for FAQs and routing · Chatbot types · What the best ones share · Which chatbot for which job · Build your own · FAQ
Build your own chatbot with Lyro and Flows, free to start
Websites that use chatbots at a glance
| # | Brand | Industry | What the chatbot does | Result (company-reported) | Built with |
|---|---|---|---|---|---|
| 1 | Burker | E-commerce (watches, jewelry) | Answers order and product questions in the customer’s language | About 80% resolution, 10,000 chats a month | Tidio Lyro |
| 2 | MattressNextDay | E-commerce (mattresses) | Answers routine questions on the website, email, Instagram and Facebook | 73% of conversations automated | Tidio Lyro |
| 3 | Bella Santé | Med spa | Answers service questions and suggests treatments | 75% of chat questions automated | Tidio Lyro |
| 4 | Klarna | Fintech | Handles refund, return and payment chats | Two-thirds of chats in month one | OpenAI |
| 5 | Bank of America | Banking | Erica answers account questions in the app | 3.2 billion interactions since 2018 | Built in-house |
| 6 | Gecko Hospitality | Recruiting | Answers visitor questions and collects candidate details | 90% of conversations handled, 257 leads in six months | Tidio Lyro and Flows |
| 7 | Endeksa | Real estate data | Routes chats and collects visitor details | 138% more leads | Tidio Flows |
| 8 | ADT Security | Home security | Qualifies leads and routes chats by department | 17% higher lead-to-sale conversion | Tidio Flows |
| 9 | Sound Royalties | Music financing | Answers FAQs around the clock and captures leads | 54% bot engagement, 3.2% of traffic to leads | Tidio Flows |
| 10 | Alaska Airlines | Travel | Suggests destinations from a plain-language request | 75% less planning time | Azure OpenAI |
| 11 | Praktiker Hellas | Retail (home improvement) | Answers common questions in Greek | 8.99% engagement with one bot, 9,400+ runs in 30 days | Tidio Flows |
| 12 | Waldo’s Dólar Mart | Retail (discount stores) | Runs sales flows and tags complaints by department | 50.87% bot engagement | Tidio Flows |
Support chatbots
Support is where chatbots earn the most, because the questions repeat. Customers ask where the order is, how returns work and whether an item fits, and the answers already sit in a help center. These five bots answer them without a queue.
1. Burker: a multilingual chatbot for a global watch brand
Burker sells watches and jewelry from a Shopify store to customers in dozens of countries, with two people on social media support and an outsourced team on email. Around Black Friday and Valentine’s Day, messages jumped from 3,000 to 5,000 a month to as many as 17,000, and the website chat had to be switched off. Lyro now answers “where is my order?” and “which watch is right for me?” in the customer’s own language.
Result: Lyro resolved 75% of conversations in the first month and about 80% in April, when it handled 10,000 chats. Outsourced email traffic fell by about half.
What works: order status, shipping and sizing need no human judgment, so the bot takes all of them and the people keep the complex cases.
2. MattressNextDay: one chatbot across four channels
MattressNextDay, a UK online mattress retailer, answers customers on its website, by email, on Instagram and on Facebook. It trained Lyro on its product catalog and policies, so the same bot replies about delivery, order tracking and returns wherever the question arrives.
Result: Lyro resolves 73% of conversations, and the team saves more than 400 hours a month.
What works: the bot covers every channel the customers already use, so there is no separate inbox to check. Its customer services manager points to how few chats the bot transfers to an agent as proof that it works.
3. Bella Santé: a chatbot on top of a call center
Bella Santé, a day spa and med spa near Boston, did not replace its phone team. It put Lyro in front of it. Guests who only want a quick answer type a question, and Lyro responds from the spa’s FAQ and the team’s own question-and-answer pairs. A pre-chat survey asks for a name and email before the chat starts.
Result: Lyro handles about 75% of chat questions. The spa reports more than $66,000 in sales that Lyro assisted and more than 450 leads in six months.
What works: a pre-chat survey turns support chats into a lead list at no extra cost. The call center stays for the cases that need a person.
4. Klarna: a chatbot that handled most chats, then added humans back
Klarna’s assistant, built on OpenAI’s models, handles refunds, returns, payment questions and cancellations inside the Klarna app, in more than 35 languages. In its first month in 2024 it had 2.3 million conversations, about two-thirds of Klarna’s service chats.[1]
Result: Klarna said the assistant did the work of 700 full-time agents, and customers resolved their issues in under 2 minutes instead of 11.[1] By late 2025 the figure was more than 853 agents.[2]
What works: the lesson is in the second act. In 2025 Klarna’s CEO said that cutting cost had made quality drop, and the company began hiring people so that customers could always reach a human.[3] Plan the handoff before the launch.
5. Bank of America: a rule-based assistant at huge scale
Erica lives in the Bank of America app. It answers questions about balances, shows spending insights and sends alerts, and it can schedule a conversation with a specialist. Bank of America reports that 20.6 million people used it about 700 million times in 2025, and that total interactions have passed 3.2 billion since 2018.[4] It is not a free-text generative bot: it answers from a library of more than 700 responses.[5]
Result: the bank says more than 98% of users find the information they need, which reduces call center volume.[5]
What works: Erica stays inside banking tasks and hands off the rest. A narrow, predictable bot can be the right choice when a wrong answer is expensive.
Use live chat to support your customers in real-time
Sales and lead generation chatbots
A chatbot can also start the sale. These examples ask the visitor a few questions, collect contact details and pass the chat to the right person.
6. Gecko Hospitality: a recruiting chatbot that sorts candidates
Gecko Hospitality recruits for hotels and restaurants. Candidates do not want to fill in a long application, so a chat flow asks two short questions, picks the best-suited recruiter and sends the CV in as little as 90 seconds. Lyro answers the rest: where to send a resume and how to use the job board.
Result: Lyro handles about 90% of conversations, and the flow collected 257 leads in six months.
What works: two questions decide who gets the lead, so nobody reviews each resume by hand. Gecko’s support manager audits Lyro’s answers every day, which is how a team keeps an AI bot honest.
7. Endeksa: a chatbot that routes chats and collects leads
Endeksa, a real estate data platform in Turkey, Spain and Portugal, splits chats into customer service and sales and routes each to the right team. A pre-chat survey collects each visitor’s name, phone number and email. A second bot asks users whether the valuation they received was helpful.
Result: waiting time fell by 59%, leads grew by 138% and the satisfaction bot reached an 88% helpfulness rate.
What works: one question up front sends each chat to the right team, so nobody has to forward it.
8. ADT Security: a rule-based chatbot for lead-to-sale conversion
ADT Security Service in Mexico uses Flows to qualify leads and to filter chats by department. The team gets about 30 new prospects a week, handles 45% more conversations and misses 74% fewer.
Result: lead-to-sale conversion rose by 17%, customer satisfaction by 30% and first response time dropped by 22%.
What works: the flows ask a qualifying question and then route the chat by department, so the right person picks up each lead.
9. Sound Royalties: FAQ bots that capture leads around the clock
Sound Royalties finances music creators in 18 countries, many of whom would rather chat than pay for an international call. Its FAQ bots answer at any hour, and missed messages go to email. Account questions and sensitive cases are transferred to a team member.
Result: 54% of visitors engage with the FAQ bots, and 3.2% of website traffic converts into leads.
What works: the FAQ bot is a buffer. It answers the simple questions and frees the team for the financing conversations.
10. Alaska Airlines: a chatbot that plans a trip from a sentence
Alaska Inspires lets travelers type or say what they want, such as a diving trip or live music, and suggests destinations that they can compare and book in one place. It runs on Azure OpenAI, takes voice input and supports more than 90 languages.[6]
Result: Microsoft’s customer story reports 75% less time spent on destination planning, a 7.16% conversion rate against a 5% standard and 90% guest satisfaction.[6]
What works: the bot replaces a form with a sentence. Travel is a clear case for chatbots that recommend, not only answer.
Flow chatbots for FAQs and routing
Not every chatbot needs AI. A flow is a set of buttons and rules that you build without code, and it is the cheapest way to start.
11. Praktiker Hellas: Greek-language FAQ bots for a home improvement retailer
Praktiker Hellas, a Greek home improvement and DIY retailer, runs Greek-language bots on praktiker.gr built with Flows. A “Solve Problems” bot answers common questions and routes complex issues to human operators.
Result: the bots ran more than 9,400 times in 30 days, and the “Solve Problems” bot reached 8.99% engagement.
What works: a flow in the shoppers’ own language beats a generic one, and the human handoff is built into it.
12. Waldo’s Dólar Mart: sales flows and complaint tags
Waldo’s Dólar Mart, a Mexican discount retail chain, uses one set of flows on its website chat, WhatsApp, Messenger and Instagram. Sales flows switch on and off with each promotion. A tagging flow records which department a complaint is about, so managers see where problems start.
Result: 50.87% of visitors engaged with the bots over 30 days, and the company reports millions in sales generated through chat automation.
What works: every button a customer presses is also data. Tags turn complaints into a list of what to fix.
Chatbot types: rule-based bots and conversational AI
The examples use two kinds of bots. A rule-based chatbot follows a script: it waits for a trigger word or a button press and returns a fixed answer. Bank of America’s Erica, the flows at Praktiker and the routing at ADT work this way. Conversational AI reads what a person means, remembers the last few messages and answers in full sentences. Klarna’s assistant, Alaska Inspires and Lyro work this way.
The fastest way to see the difference is a question the bot was never scripted for:
- Rule-based bot: “returns” triggers the returns policy. “Can I send this back?” may miss.
- Conversational AI: “can I send this back?”, “what’s your return window?” and “this didn’t fit” all resolve to the same answer, pulled from your help content.
Most teams end up with both: flows for routing and lead capture, and an AI agent for open questions. For more cases of the second kind, see Tidio’s list of conversational AI examples.
What the best chatbot examples have in common
Strip away the industries and the same five habits repeat:
- One job first. Burker started with order status, ADT with lead routing. Neither tried to be a general assistant.
- A real source. The AI answers from the help center, the FAQ and the catalog. If those are thin, so is the bot. More on that in Tidio’s chatbot best practices.
- A handoff that works. Lyro, Erica and the Flows bots all pass the conversation to a person with the history attached. Klarna had to add this back.
- Every channel the customers use. MattressNextDay and Waldo’s cover the website, Instagram and Facebook with one setup.
- A number to watch. Pick one metric before the launch: resolution rate for support, engagement or leads for a flow.
Which chatbot fits which job
| Your goal | Example to copy | Chatbot type | What to set up first |
|---|---|---|---|
| Answer repeat questions | Burker, MattressNextDay | AI agent | Connect your help center and product pages |
| Take pressure off the phone | Bella Santé | AI agent | Add the FAQ and a clear handoff to a person |
| Capture and route leads | Endeksa, ADT Security | Rule-based flow | One routing question and a pre-chat form |
| Qualify candidates or leads | Gecko Hospitality | AI agent and flow | Two short questions that pick the owner |
| Answer in the shopper’s language | Praktiker Hellas | Rule-based flow | Write the flow in the local language |
| Recommend what to buy or book | Alaska Airlines | AI agent | Connect the catalog and prices |
For tools that build bots like these, see Tidio’s guide to chatbot software and the best chatbot platforms.
Create your own successful chatbot
Copy the pattern, not the brand. Pick the one question your customers ask most, write the answer once and test it with ten real messages before you go live. Add a handoff to a person on day one.
Tidio puts the pieces in one place: live chat, the Lyro AI agent and no-code Flows. Start with an FAQ flow or order tracking, then add Lyro for open questions. If you sell on Shopify, the Shopify chatbot guide shows the setup, and how to create a chatbot for a website walks through the steps.
Frequently asked questions
Many do, from banks and airlines to small online stores. In this article: Klarna, Bank of America, Alaska Airlines, Burker, MattressNextDay, Bella Santé, Gecko Hospitality, Endeksa, ADT Security, Sound Royalties, Praktiker Hellas and Waldo’s Dólar Mart. They use chatbots to answer support questions, route chats, capture leads and plan trips.
Burker’s chatbot is one: it answers “where is my order?” and watch-sizing questions in the customer’s language, and it resolves about 80% of its conversations. Another is Bank of America’s Erica, which answers account questions in the app.
A rule-based chatbot waits for exact trigger words or button presses and follows a script. Conversational AI understands natural language, works out what a person means even when the phrasing is messy, and remembers earlier messages. Many bots combine both.
It depends on the job. For answers to repeat questions, pick a bot that learns from your own help content, works with your platform and hands off to a person. For lead capture, a flow with a pre-chat form is usually enough. Small teams get the most from a tool that combines live chat with an AI agent. Tidio’s chatbot for business guide compares the options.
One that can answer order status, shipping and returns questions, and that works with your platform. Burker and MattressNextDay both use an AI agent for exactly these questions. Tidio’s e-commerce live chat guide covers the setup.
Tidio is an AI customer support platform that more than 300,000 businesses use for chat, support and sales. Its AI agent, Lyro, learns from your website and help content, answers customers on your website and on channels such as Instagram and Messenger, and hands the conversation to your team when it cannot resolve it. Lyro can resolve up to 75% of support requests. Tidio also includes live chat, a help desk and no-code Flows for rule-based chatbots.
Boost customer satisfaction with a professional AI agent
