How to Write Chatbot Fallback Messages That Keep Customers

By Marcus Reed · Updated 2026-11-21 · reviewed before publication

Every chatbot hits a wall eventually. A customer types something the bot doesn't understand, and the bot has to say something. That "something" is called a fallback message, and most small businesses treat it like an afterthought.

That's a mistake. The fallback is where you either keep a customer or lose them. A bad one sounds like a robot shrugging. A good one feels like a helpful employee who knows when to get a human.

This guide walks you through how to write chatbot fallback messages that actually work. No technical background needed. We'll cover the different types of fallbacks, give you templates you can copy, and point to the tools that make this easy: Tidio, Chatbase, and Crisp.

One note on sourcing: every price and limit below comes from the vendors' own public pricing pages and documentation, not from paid testing.

Our top picks for this guide
TD
Tidio ★★★★★
from $24.17/mo billed yearly (Starter, 100 billable conversations) · Live chat + Lyro AI chatbot for small businesses
Review
CB
Chatbase ★★★★★
from $40/mo (Hobby, 700 message credits) · Train a ChatGPT-style bot on your own content
Review
CR
Crisp ★★★★★
from $45/workspace/mo (Mini, 4 seats) · Affordable all-in-one inbox: chat, chatbot, emails
Review

What a fallback message actually is (and why it matters)

A fallback message is what your chatbot says when it can't match what the customer typed to anything it knows. Think of it as the bot's "I'm not sure what you mean" reply.

Here's why it matters more than you'd think. A customer who gets a confusing reply doesn't usually try again. They leave, and they rarely tell you why. The fallback moment is where a lot of conversations quietly die.

A good fallback does three things: it admits the bot didn't understand, it offers a clear next step, and it keeps the tone human. A bad one does the opposite — it repeats the same generic line, traps the customer in a loop, or pretends to understand when it doesn't.

The three types of fallback (you need all three)

Not every fallback is the same. Good chatbot setups use a tiered approach, and the tiers get more helpful as the bot struggles.

  • **First fallback:** The bot didn't catch the question once. Stay light and offer a rephrase or a menu of common topics.
  • **Second fallback:** The bot missed again. Now get more direct — offer a human, a phone number, or a link to your help page.
  • **Handoff fallback:** The bot has failed enough. Stop guessing and pass the customer to a real person or a contact form.

Picture a bot stuck on one line: "Sorry, I didn't get that," repeated forever. Visitors read that as a broken site and stop replying. Add a second-tier message with a "text us here" button and the conversation at least has somewhere to go.

The lesson: never let your bot loop. Each fallback should escalate, not repeat.

Write the first fallback like a helpful person, not a robot

The first fallback is your most common one, so get it right. Keep it short. Two sentences max.

Bad: "Invalid input. Please try again."

Better: "I didn't quite catch that. You can ask about orders, returns, or store hours — or type 'human' to reach our team."

See the difference? The second one gives the customer a map. It tells them what the bot *can* do instead of just admitting failure.

A few rules worth keeping:

  • Never blame the customer ("You entered that wrong").
  • Always offer at least two concrete options.
  • Use plain words. Skip "I apologize for the inconvenience" — it sounds like a form letter.
  • Match your brand voice. If your store is friendly, the bot should be friendly too.

Make the second fallback a real exit ramp

If the first fallback didn't help, the customer is now frustrated. This is not the time for another cheerful "Try asking me something else!" This is the time to get them to a human.

Your second fallback should include a direct handoff. Something like: "I'm still not getting it — sorry about that. Let me connect you with someone who can help. Tap below and a team member will reply shortly."

This is where tool choice matters. Chatbase, for example, is great at training a bot on your own content, but its built-in Helpdesk with human handover requires the Standard plan and up — so on cheaper tiers you'll need to route people to email or a contact form instead. Tidio builds live chat in, so the handoff happens in the same widget.

Pick a tool that matches how you actually want to hand off. If you have staff answering chat, use a live-chat-first tool. If you're solo, a contact form fallback is fine.

Test your fallbacks with real, messy questions

Here's the step most businesses skip: actually testing the fallback. Don't test with clean questions. Test with the weird stuff real customers type.

Type things like "where my stuff at," "do u take paypal??", "i want refund now," and a few typos. Watch what the bot does. If it falls back on a question it should obviously know, your training data has a gap — not your fallback.

Run this test monthly. Customer language shifts, and a fallback that worked in January can feel stale by summer.

One tip: keep a running list of the questions that triggered fallbacks. That list is gold. Every entry is either a missing answer you should add or a sign your fallback wording needs work.

Pick a tool that lets you edit fallbacks easily

You can write the perfect fallback, but if your tool buries the setting three menus deep, you'll never update it. Ease of editing matters.

Tidio is the pick for most small stores. The vendor's own setup claim is a live chat widget working in under ten minutes — that is their wording, not a stopwatch reading — and Lyro AI answers questions from your FAQ automatically. The catch: Lyro AI conversation credits run out quickly on cheaper plans, so watch your volume. Pricing starts at $24.17/mo billed yearly, and there's a free plan with 50 conversations/mo to test the waters.

Chatbase is the choice if your bot lives on docs or a website knowledge base. You upload content and get a ChatGPT-style bot in minutes, with a simple embed on any site. Just know the human-handoff Helpdesk needs the Standard tier ($150/mo), and its free tier was withdrawn in October 2026. Plans start at $40/mo.

Crisp is the budget-friendly all-in-one: shared inbox, chatbot, and campaigns in one place, with flat per-workspace pricing at €45/workspace/mo. The tradeoff is lighter AI than Tidio — the free plan has zero AI credits — so it suits teams that mostly want a solid inbox with a simple bot attached.

Whichever you choose, check the vendor's site for current pricing before you commit. Plans change.

Common fallback mistakes to avoid

The same errors show up again and again. Here's the short list:

  • **The infinite loop.** The bot repeats the same fallback. Always escalate.
  • **The dead end.** "I don't know" with no next step. Always offer a path forward.
  • **The over-apology.** Three sentences of "so sorry" before any help. Get to the point.
  • **The false confidence.** The bot guesses wrong instead of falling back. A clean "I'm not sure" beats a wrong answer.
  • **The forgotten fallback.** You wrote it once and never updated it. Review quarterly.

Fix these and your bot will feel dramatically more competent — even if the underlying AI hasn't changed at all.

Frequently asked questions

How many fallback messages should a chatbot have?

At least three: a first fallback that offers options, a second that escalates, and a handoff that reaches a human or contact form. More tiers help, but three covers the basics for most small businesses.

What should a fallback message say?

Keep it short and give the customer a next step. Admit the bot didn't understand, then offer two or three concrete options — common topics, a human handoff, or a link to your help page. Never just say "try again."

Can I edit fallback messages in Tidio, Chatbase, or Crisp?

Yes — all three let you customize bot replies, including fallbacks. Tidio and Crisp are the easiest to edit for non-technical users, while Chatbase gives you more control over how the AI responds to unknown questions.

Why does my chatbot keep repeating the same fallback?

It's usually because you only have one fallback set. Add a second and third tier that escalate to options and a human handoff, so the bot never loops on the same line.

Ready to pick one?

TD
Tidio ★★★★★
from $24.17/mo billed yearly (Starter, 100 billable conversations) · Live chat + Lyro AI chatbot for small businesses
Review
CB
Chatbase ★★★★★
from $40/mo (Hobby, 700 message credits) · Train a ChatGPT-style bot on your own content
Review
CR
Crisp ★★★★★
from $45/workspace/mo (Mini, 4 seats) · Affordable all-in-one inbox: chat, chatbot, emails
Review
MR
About the author

Marcus Reed — Marcus Reed is the senior editor at ChatbotRadar. He tracks live chat and AI chatbot tools for small teams, verifying pricing on vendor sites and trialling tools on their free plans. How he tests · Report an error

Disclosure: this article contains affiliate links. If you buy through them we may earn a commission at no extra cost to you. Pricing and features are re-checked regularly but can change — always confirm on the vendor's site.