Implementation & How-To· 5 min read· By The SpideyChat Team

What Your Team Needs to See When a Chat Is Handed Over

A context checklist for the person receiving a chatbot handover: who the visitor is, what they asked, what the bot already said, and how urgent it is.


When a chat is handed to a person, they need six things before typing a word: who the visitor is and how to reach them, what they want in their own words, what the bot already told them, anything that was quoted or promised, how urgent it is, and any evidence attached. Most bad handovers come from skipping the third item and replying to the last message only. A good handover is less about the tool passing context and more about the receiving person reading it.

Our post on designing a human handoff your customers won't hate covers the visitor's side. This one is for the team member on the receiving end.

The receiving checklist

Use this list every time a conversation reaches you, whether live or as a ticket.

1. Who is it, and how do you reach them? A name if they gave one, and an email or phone number. If you have neither on a live chat, ask once, early, before the visitor leaves.

2. What do they want, in their own words? Their first message usually states the original question, and their latest message states where things stand now. Read both. "Do you deliver to Ireland?" followed by "my order hasn't arrived" are different problems.

3. What has the bot already told them? This is the one most often skipped. The visitor has read the bot's answers and may be acting on them. Scan every bot reply before writing yours.

4. Has anything been quoted or promised? Prices, delivery times, return windows, availability. If the bot quoted a figure from your pages, your reply should match it or correct it clearly.

5. How urgent is it? Is someone waiting in the chat right now, or did they leave a ticket overnight? Is it a question, a complaint, or something broken that costs them money?

6. Is there evidence? Screenshots, an order number in the text, the exact error message.

Where each item lives in SpideyChat

The tools show most of this, in two places.

In the Live inbox, for conversations happening now, each row shows the visitor's name if given (or "Visitor"), their email if given, the latest message, the message count, a short reference such as CNV-3f2a9c1d, and a Waiting badge if they asked for a person and nobody has joined. Opening a conversation shows the full thread, including the bot's replies.

On a ticket, created when a visitor uses Report an issue or accepts the bot's offer of a ticket, you get the ticket number, a subject taken from their description, their name, email and phone as given, up to three attached screenshots, a priority (tickets from the chat widget start as high), and the transcript of the conversation that led to it. If the visitor reports again in the same conversation, no second ticket is opened; they get the same number, and any new screenshots are added to it. The ticket inbox, with status, priority and notes, is on Pro; the ticket email, with the same number, goes to the account owner on every plan.

What the tools cannot tell you is what the visitor was thinking. That is why item 3 matters: the transcript is the closest thing you have.

A 60-second reading routine

Before replying to any handover:

  1. Read from the top. Read the first visitor message, then every bot reply, then the latest message.
  2. Mark the bot's claims. Note any fact the bot stated, especially numbers.
  3. Check those claims. Are they right? If the bot answered from an outdated page, you have found a content fix as well as a customer to help.
  4. Write down what is still open. Usually one or two questions.
  5. Reply to those questions, confirming or correcting the bot where needed.

A reply built this way sounds like this: "Hi Sam, I've read through your chat. The assistant was right that standard delivery to Ireland is 3 to 5 working days. For your order that hasn't arrived, could you send me the order number? I'll check with the courier today."

Compare that with "Hi, how can I help?", which makes the visitor start again.

Mistakes that undo a handover

Asking for details already given. If their email is in the conversation, do not ask for it. If the issue is described in the ticket, do not ask them to describe it.

Contradicting the bot silently. If the bot was wrong, say so: "Sorry, the assistant gave you our old price. It is now £85." Then fix the source page so it does not happen again.

Ignoring the language. Transcripts show exactly what was said, in the visitor's language, and the dashboard does not translate them. If they wrote in Italian, paste the transcript into a translation tool before replying, and reply in Italian where you can.

Leaving no trace. On a ticket, write down what you did and what is pending. Notes stay on the ticket, and you can mark one as internal, so the next teammate can pick up where you left off.

Forgetting the reference. Quote the ticket number in every email about it. The visitor has it in their confirmation email, and it keeps the thread findable on both sides.

Make the bot collect context before handover

You get better handovers when the visitor has already said what matters. Two things help.

Clear content. The bot answers from your pages, and when it cannot, it says so and offers your team. The fewer questions it misses, the more the handovers you receive are genuinely ones that need a person.

A flow that asks first. If you use the visual flow builder, a flow that ends in a handoff step opens a ticket automatically and fills it with what the flow collected, such as name, email, phone and a description of the issue. The post on how AI chatbots collect the right info before a human steps in covers what to ask, and handing complex tickets to a person with context covers the harder cases.

See the handover screens yourself

The quickest way to know what your team will see is to play the visitor. Open your chatbot's preview, ask a question, press Report an issue, and then open the ticket from the dashboard. The features page lists the live inbox, ticket and transcript tools, and pricing shows which plan includes the ticket inbox.

Frequently asked questions

What should a team member check before replying to a handed-over chat?
Who the visitor is and how to reach them, what they asked in their own words, what the bot already told them, anything that was promised or quoted, how urgent it is, and any screenshots they attached. Reading the whole transcript first avoids asking them to repeat themselves.
Does the team see the full chatbot conversation?
Yes. In SpideyChat the Live inbox shows the full thread of a conversation you open, and a ticket created from the chat widget has the conversation's transcript attached on its detail page.
Why does it matter what the bot already said?
Because the visitor read it. If the bot quoted a delivery time or a price and your reply says something different without explanation, the visitor stops trusting both. Confirm or correct the bot's answer explicitly.
What if the visitor wrote in another language?
The transcript shows the conversation exactly as it happened, in the visitor's language, and the dashboard does not translate it. Paste it into a translation tool before replying.

Want a 24/7 AI Employee for Your Website?

See how SpideyChat can answer customer questions, capture leads and handle enquiries while your team is offline.

No obligation. We will review your website and show you how SpideyChat could work for your business.

Keep reading