How to Migrate From One Chatbot to Another (Without Losing Data)

First, figure out what you're actually moving

Before you touch anything, make a list. This is the part most people skip, and it's the reason migrations go sideways.

Open your current chatbot and write down four things: your saved replies and canned answers, your chatbot flows (the scripted conversations), your conversation history, and your integrations (things like Shopify, WordPress, or your email tool).

Then ask yourself a simple question: which of these do I actually need to bring with me?

Start with a list, not a feeling. Most of what you're dreading turns out to be either exportable or rebuildable. A typical small business needs its saved replies, its main chatbot flows, and a way to keep old conversations for the record. Everything else is usually clutter — and clutter is the reason a two-hour job feels like a two-month project.

Export your data before you cancel anything

The rule that matters here: never cancel your old chatbot until the new one is live and tested.

Most tools will let you export your conversation history as a CSV or similar file. Do this first and save it somewhere safe — a folder on your computer, or better, cloud storage. Even if the new tool can't import it directly, you'll have a record.

For your chatbot flows, take screenshots. Lots of them. The flow builder in your old tool probably won't translate directly to the new one, so you'll be rebuilding those flows by hand. Screenshots are your blueprint.

One warning: some tools delete your data soon after you cancel. Check the vendor's site for their data retention policy before you pull the plug. Assume nothing about how long your history survives — the export is the only copy you actually control.

Pick a new tool that matches how you actually work

This is where the migration gets personal. The right chatbot for a support team is different from the right one for a Shopify store.

**Chatwoot — best if you want to own your data.** It's open source, which means you can self-host the full feature set for free if you have someone technical on hand. The catch is real, though — self-hosting means running your own servers, and that's not for everyone. The cloud version starts at $19 per agent per month billed yearly, and the free Hacker plan covers 2 agents and 500 conversations a month. It suits teams that want one inbox for live chat, email, WhatsApp, and social messages.

**LiveChat — best if reliability and reporting matter most.** It has been around since 2002 and it shows in the agent tooling: rock solid, with excellent reporting. Its Text Intelligence AI drafts replies and summarizes chats. There's no free plan, just a 14-day trial, and pricing grows fast as you add agents — $19 per agent per month billed yearly, or $25 billed monthly. For a small support team that lives in chat all day, that's a fair trade.

**Crisp — best if budget is tight and you want a lot bundled together.** It's flat per-workspace pricing at €45 per workspace per month, which gets cheap fast for bigger teams. You get a shared inbox, chatbot, campaigns, and video chat in one place. The free plan includes 2 seats but no AI credits, and the AI features are lighter than Tidio or Botpress. It's a good fit for startups that want one affordable tool instead of three.

Rebuild your flows, don't just copy them

Here's the part people underestimate: chatbot flows almost never transfer cleanly between tools. Each platform has its own builder, its own logic, and its own way of handling things like handoffs to a human.

So treat the rebuild as a chance to improve. Look at your old flows and ask: which ones actually worked? Which ones did customers ignore? A flow that seemed clever six months ago might be dead weight now.

Rebuild only the flows that earned their keep. Start with the three or four that handle the most common questions — order status, business hours, returns, and pricing. Get those working before you add anything fancy.

If you're moving to a tool with AI training features, this is also the moment to feed it your best content. Upload your FAQ page, your return policy, and your product docs. The AI will use those to answer questions, and cleaner input means better answers.

Test side by side before you switch over

Don't flip the switch on a Friday afternoon. Run both chatbots at the same time for at least a week if you can.

The easiest way is to install the new chatbot on a low-traffic page first — your contact page or a single product page. Then ask friends, staff, or a few real customers to try it and tell you what breaks.

Watch for three things: Does the bot answer common questions correctly? Does the handoff to a human work? Do conversations show up in the right inbox? If you sell on Shopify, add one more: does an order-related question actually pull the right order, or does the bot guess?

This is exactly why the overlap week exists. A broken Shopify sync on day two is a five-minute fix while the old bot is still catching traffic. The same bug discovered after you've cancelled is a week of lost orders.

Only after the new bot passes your tests should you remove the old one. And keep the old account active for a few weeks after that, just in case.

Update everything that points to your old chatbot

This step is boring and easy to forget, and it's where migrations quietly fail.

Your chatbot lives in more places than you think. Check your website footer, your email signatures, your social media bios, your help center, and any printed materials or PDFs that mention it.

If your old chatbot had a direct link customers used, update it or set up a redirect. If you had a widget embedded on specific pages, remove the old code before adding the new one — two chat widgets on one page is a confusing mess for customers.

Also tell your team. A one-page note explaining what changed, where to find conversations now, and who to ask for help saves weeks of confusion.