
How to Design a Pre-Chat Form That Doesn't Cost You Conversations
Quick answer: Write down what you will do with each answer, and drop every field that has no job. Keep required fields to what the chat cannot work without, usually the question and, when nobody can reply live, an email. Skip the form for visitors you already know, add a one-line privacy notice, and measure how many people who open the widget go on to send a message. Change one thing at a time.
Step 1: Decide what the form is for
There are three honest reasons to have a pre-chat form:
- Follow-up. You cannot always answer live, and you need a way to reply after the visitor leaves.
- Routing. Sales, support and billing questions go to different people, and getting it wrong costs time.
- Context. The agent answers faster if the question arrives with the chat.
If none applies, for example a one-person shop that answers every chat within a minute, you may not need a form. Intercom's article on requiring an email estimates that requiring one always reduces conversation volume by about 30% on average, and only outside office hours by about 5%. That is Intercom's own estimate with no published method, but the direction matches LiveChat's warning that more required fields make people less likely to chat.
Step 2: Give every field a job
Make a short table before you touch the settings. If a field has no "what we do with it", remove it.
| Field | What we do with it | Required? |
|---|---|---|
| Question | Agent or bot sees the topic | Yes |
| Reply if the visitor leaves or we are offline | Out of hours only | |
| Topic or department | Route to sales or support | Only if routing depends on it |
| Name | Greeting | No |
| Phone | Call-back | No, unless you really call back |
| Order number | Look up the order | No; ask in the chat when relevant |
LiveChat's page suggests that name, email and a short description work in most cases. Zendesk's Classic widget form has a greeting, name and email, phone, department and message; it shows the department field automatically when departments are set up, so check that you have not inherited a field you never chose.
For routing, use a short choice list rather than free text: LiveChat's group selection field sends the chat to the group tied to the chosen option. Keep the options few and in the visitor's words ("A question before buying", "Help with my order"), not your internal team names. How those answers then pick a team and an agent is covered in routing live chats to the right agent.
Step 3: Decide what is required, and when
Mark a field required only if the chat genuinely cannot proceed without it. The most useful pattern in the documentation read for this guide is time-based: Intercom lets you require an email "Always" or "Only outside of office hours", and recommends the second. During working hours an agent can ask in the chat; outside them, the email is the only way to reply. If you use an after-hours chatbot, the same logic applies to what the bot asks for before promising a reply.
If your platform has no time-based option, consider making email optional on the form and having the chatbot ask for it only when a follow-up is needed. That shifts the friction to the point where the visitor can see why you ask; see conversational form for the one-question-at-a-time approach and multi-turn form design for keeping answers in state.
Step 4: Write the greeting and labels
The form's greeting is the first thing a visitor reads, so it should say who answers and when. Templates to adapt:
Working hours: Ask us anything. We usually reply within a few minutes. Your email is optional and only used to reply to you.
Out of hours: We're offline right now. Leave your question and email and we'll reply tomorrow morning.
Say why you need each detail in plain words. Do not promise a reply time you cannot keep. Wording ideas for the first line are in chatbot welcome message design; test two versions with A/B testing chatbot messages.
Step 5: Skip the form when you already know the visitor, and plan for chats that bypass it
Known visitors. Asking a logged-in customer for their email is friction with no benefit. LiveChat documents pre-filling the form for logged-in customers through its API. Intercom does not ask a lead again once an email has been given. If your platform offers neither, check whether it can hide the form for identified users.
Chats that skip the form. Not every chat passes through it. Zendesk's proactive chat article says chats started by a proactive trigger skip the pre-chat form, and LiveChat's page says the form is not shown when an agent invites a visitor manually. LiveChat also lets admins decide whether the form shows after a campaign message. So your routing and follow-up cannot depend on the form alone: decide what the bot or agent asks first in those chats. See setting up proactive chat triggers and the handoff design guide.
Step 6: Add consent where you need it, then measure
Privacy. The form collects personal data before the conversation starts, so it is the natural place for a short notice and a link to your privacy policy. HubSpot's live chat settings include a consent option that can require visitors to click "I agree" before sending a message, or rely on legitimate interest instead. LiveChat offers sample consent clauses, including one for AI disclosure, and says they are general examples to adjust. What you need depends on where you and your visitors are; see the GDPR chatbot compliance guide and chatbot data privacy. This is not legal advice.
Measure. Count three things: widget opens, forms submitted and first messages sent. The drop between opening the widget and sending a message is the cost of the form. Also track how many chats end without a way to follow up. Compare a period before and after a change, and change one field at a time. See chatbot metrics and chatbot abandonment rate.
Check the arithmetic
An invented example, not a measurement. A small shop's chat widget is opened 500 times in a week. With a form that requires name, email and phone, 300 visitors send a first message: 300 ÷ 500 = 60%. The shop removes the phone field and makes the email required only out of hours. The next week, the widget is opened 520 times and 390 visitors send a message: 390 ÷ 520 = 75%. In the second week, 40 chats ended with no way to follow up, against 0 in the first. That is 390 − 300 = 90 more conversations, but the change is a gain only if those 40 chats were answered while the visitor was still there; read them before deciding. Two weeks with different traffic do not prove cause, so keep watching. These numbers are made up to show the method.
Common mistakes
- Fields with no job, kept because they were the default.
- Requiring email during working hours when an agent could simply ask.
- Internal team names in the routing dropdown.
- Asking known customers for details you already have.
- Assuming every chat went through the form and routing on data that proactive chats never collected.
- Measuring only submitted forms, not the visitors who opened the widget and left.
FAQ
What should a pre-chat form ask? Only what the chat needs: usually the question, an email if you may need to reply later, and a topic if you route by it. LiveChat suggests name, email and a short description for most cases.
How many fields should a pre-chat form have? No vendor page read for this guide gives a measured number. Start with two or three and remove any field without a clear use.
Should email be required before a chat? Consider requiring it only when nobody can answer live. Intercom offers that option and estimates a much smaller drop in conversations than requiring it always.
Is a pre-chat form better than letting the chatbot ask? A form routes from the first message; asking in the chat lets visitors start immediately. Many teams use a minimal form and let the bot ask for details when needed.
Why did some chats arrive without form data? Proactive and manually invited chats can skip the form, according to Zendesk's and LiveChat's documentation. Plan what the bot or agent asks in those chats.
Do I need a consent checkbox on the form? It depends on your location, your visitors and what you do with the data. HubSpot and LiveChat both offer consent options; check your own obligations.
Sources
- Intercom, Require leads to give their email to start a conversation — intercom.com/help/en/articles/2301967-require-leads-to-give-their-email-to-start-a-conversation, read 10 November 2026. Source of the "Always" and office-hours modes, the recommendation and the 30% and 5% estimates. No method published.
- LiveChat, Chat surveys — livechat.com/help/chat-surveys/, read 10 November 2026. Source of the field types, required fields, group routing, three-liner suggestion, pre-fill for logged-in customers, manual-invitation and campaign behavior, and sample consent clauses.
- Zendesk, How do I customize the pre-chat form? — support.zendesk.com/hc/en-us/articles/4408825438746-How-do-I-customize-the-pre-chat-form, read 10 November 2026. Source of the Classic widget elements and the automatic department field.
- Zendesk, Using proactive chat to target key visitors — support.zendesk.com/hc/en-us/articles/4408883473178-Using-proactive-chat-to-target-key-visitors, read 9 November 2026. Source of the note that proactive chats skip the form.
- HubSpot, Create a live chat — knowledge.hubspot.com/chatflows/create-a-live-chat, read 10 November 2026. Source of the email capture and consent options.
- Zoho SalesIQ, Pre-chat window fields — zoho.com/salesiq/help/configuration-fields-pre-chat.html, read 10 November 2026. Shows adding fields and marking them mandatory; consulted, not quoted.
- Chatbotscape evaluation methodology. /methodology (continuously updated).