Collect visitor details

What it is

Collect info is a Call Action type that lets the chatbot gather contact details during a normal conversation and save them to your Customers list. In the type picker it reads "Ask the visitor for details and save them."

There is no form. The chatbot asks for the details conversationally, at the moment it makes sense, and when it has them the action saves the contact. The visitor experience is a chat, not a lead-gen popup.

Collect info or the pre-chat form?

Agentency has two ways to capture a contact and they solve different problems.

	Collect info action	Lead capture fields
When it asks	Mid-conversation, when relevant	Before the chatbot's first reply
Feels like	A conversation	A form
Blocks the chat	No	Can be set to
Configured on	The Actions tab	The chatbot's lead capture settings

Use lead capture when you must have an email before you spend a reply on someone. Use collect info when you would rather answer the question first and ask for details once the visitor is interested. See Lead capture fields.

They coexist happily. Both write to the same Customers list.

When you would use it

• Someone asks for a quote, a callback, or a brochure.
• A visitor wants to be told when something is back in stock.
• A conversation is going well and you want the follow-up details before they close the tab.
• A support case needs an email address so you can reply later.

Where to find it

Open Dashboard → Chatbots → your chatbot → Actions, select New action, then choose Collect info under Start from scratch.

Steps

1. Open the Actions tab and select New action.
2. Choose Collect info.
3. On Details, set a Name such as Save the visitor's contact details.
4. Write When should the AI use this? — for example, "When the visitor asks for a quote, a callback or a brochure, and has given at least their email address." Naming the precondition ("has given") stops the chatbot firing before it has anything to save.
5. On Configure, add one parameter per detail you want. Each row has a Name, a Type, a Description (helps the AI fill it), and a Required switch.
6. Write the description of each field as if you were explaining it to a new colleague: "The person's work email address" fills far more reliably than "email".
7. Save, run the Test tab with a sample message to see which arguments the model would extract, then turn on Enabled.

If you add no parameters at all, Agentency falls back to a sensible default of name (optional) and email (required), so a minimal action still works.

Choosing your fields

Keep the list short. Every extra required field is another thing the chatbot must extract before it may save anything, and a long list turns a conversation into an interrogation.

Field types available: string, integer, number, and boolean. Most contact fields are strings. You can also give a field a comma-separated list of allowed values, which forces the chatbot to pick one of them — useful for things like an enquiry category.

A realistic set for a quote request:

Name	Type	Required	Description
name	string	No	The person's full name.
email	string	Yes	The person's email address.
company	string	No	The company they work for, if mentioned.
enquiry_type	string	No	What they need a quote for. Allowed values: installation, maintenance, parts.

Mark only the fields you genuinely cannot follow up without. Everything else stays optional so the chatbot can save what it has instead of nagging.

What happens to the details

The collected details go straight into your Customers records — they never travel back to the visitor's browser with the chat reply. The saving happens on the server after the reply is generated, reading only what the chatbot actually extracted.

Where to find them afterwards:

• Customers → the contact's profile shows the details and the conversation that produced them. See Customer profile.
• Where leads come from explains how a contact captured by an action is attributed. See Where leads come from.
• Records can be exported or anonymised like any other contact. See Export customers and Anonymize a customer.

If the same person comes back later and gives the same email, they are matched to the existing record rather than duplicated.

What you will see

The visitor sees the chatbot ask for the details in plain conversation and thank them once it has them. No form, no separate screen.

In the dashboard you see a normal action card with run count and success rate. In the Run log, a successful run's Arguments panel shows which fields were filled — so the run log is where you look when you want to know why a contact came out incomplete. See Call Action history.

Privacy considerations

You are asking a visitor for personal data, so treat it as such:

• Only ask for what you will actually use. A phone number you never call is a liability, not a lead.
• Say what you will do with it. An instruction item like "When asking for contact details, say we will only use them to answer this enquiry" makes the chatbot state it every time. See Write instructions.
• Collected details land in Customers, which supports export and anonymisation, so you can honour a deletion request without touching the transcript.

Limits and plan notes

• Collect info counts as one enabled Call Action: Free 0, Starter 3, Standard 8, Pro 12, Agency 20 per chatbot. See Plans and what you get.
• You need permission to manage actions on that chatbot. See Roles and permissions.
• The action makes no outbound call and stores no credential, so there is nothing to authenticate and nothing to break when a third-party API changes.
• Requires confirmation is available but rarely useful here — the visitor already consented by typing the details.
• The chatbot can only save what the visitor actually said. It does not read cookies, form fields, or anything else on your page.

Common problems

Contacts are saved with the email missing.

The field's description is too thin, or it is marked optional. Mark it required and describe it explicitly: "The person's email address, as they typed it."

The chatbot asks for details far too early.

Rewrite the description to state the precondition — "only after the visitor has asked for a quote or a callback" — instead of describing the fields.

Nothing appears in Customers.

Check the Run log first. No runs at all means the action never fired: confirm it is Enabled and that the chatbot's model supports tool calling (see Chatbot AI settings). Runs that succeeded but produced no contact usually mean every parameter was optional and none was filled — make the email required.

A test run did not create a contact.

The Test tab is a sandbox. It shows you what the model would extract but writes nothing. Use Test Chatbot for an end-to-end check. See Try the test chat.

I want the details in my CRM, not just in Customers.

Add a Custom API action alongside it using the HubSpot, Airtable, Notion, or Zapier template. See the Call action preset reference.