tawk.to

When a conversation cannot finish live, an agent can create a ticket from chat. Choose an owner and a valid return email, then check that the visitor receives the follow-up in the expected thread.

Teams can assign chats manually or set routing by department, channel or availability. Test which agent receives a new message and what happens at shift handoff before relying on rules during busy periods.

The product page describes a help center on a custom domain, while current help associates that setup with Remove Branding. Confirm custom-domain entitlement before planning DNS work. Keep the default hosted help center option separate from paid branding.

For evaluation, install the widget on a test page, send a fictional chat, assign and transfer it, make one draft article and publish one safe sample, then create a ticket and verify the reply path. Confirm how agent roles and connected channels behave on the property; no real support outcome was tested here.

If you may need to move chat history, test an export while the tawk.to property is active and an Admin still has access. Its help describes an emailed dashboard link to a ZIP of JSON chats and says manual Inbox export is limited to 50 messages at a time; it points to the REST API for larger or automated exports. Keep the download link with trusted people and verify the files in the receiving workflow before migration.

Before using tawk.to triggers for proactive greetings, inspect the default Welcome message and test a harmless page visit with the widget online. The vendor says triggers work only when the widget is online, new accounts include that site-notification trigger, and a property allows up to 20 triggers. Trigger Type cannot be changed after saving, so choose the type and delay deliberately before using visitor messages.

If you plan to leave tawk.to, test tickets and contacts as separate exports alongside chat history. The ticket guide requires an Admin and describes a property-specific selection that emails a JSON download link valid for 30 days. Contacts > People instead emails a CSV of the property's current contacts and needs a browser or desktop app, not mobile. Have an authorized Admin sample all three exports on the intended property, inspect their contents and protect emailed links before treating the handoff as usable; the guides do not establish a complete archive.

Before putting internal instructions in tawk.to Knowledge Base, test a fictional article in Draft, Published Private and Public states on the intended property. The vendor says drafts remain hidden from visitors, private articles stay hidden even with a direct link and from search and widget content, while a category appears publicly once it contains a public article. When configured to use Knowledge Base, AI Assist draws only on published public articles, never private ones. Check the public link, widget and category, and an AI response if configured, before storing real internal material.

When an agent leaves a tawk.to team, remove them from each relevant property and separately review internal group chats and direct messages. The official Help says property removal leaves the agent appearing in existing group chats and DMs; for group chats, it advises asking the former member to leave or creating a new group. Recheck property membership and harmless-message visibility before treating the offboarding as complete.

Before relying on tawk.to Widget Scheduler for business hours, preview a harmless page after hours and during an in-hours period with no available agents. The scheduler keeps the widget offline outside configured hours even if an agent is logged in; within scheduled hours, the widget can remain visible without an available agent, but visitors cannot start a chat. Check the operating time zone, the optional offline-when-all-agents-offline setting, and what visitors see when no one can answer. An ongoing chat is not cut off when the schedule goes offline.

Before relying on tawk.to email alerts, configure Missed Chats and Offline Form notifications for the property and choose recipients for each: specific members, the visitor-selected department, or external email addresses. An enabled type with no recipients shows a warning. Emails for conversations assigned to an agent use that agent's separate setting, which is off by default. Test with synthetic submissions and confirm the intended inboxes before depending on the routing.

If spam or abusive chats are a concern, distinguish a tawk.to visitor ban from blocking access to your website. The vendor says a ban stops interaction with the chat widget while the website stays viewable. A ban without the IP option can be bypassed by clearing cookies; Ban IP applies to the same IP address, and staff can remove a ban from the Ban List. Rehearse the choice and reversal with a fictional visitor on the intended property before relying on it for moderation.

When transferring a live chat in tawk.to, verify the recipient joins before treating the handoff as complete. The desktop Help shows agent or department selection, an optional handoff note, a Cancel action before the recipient joins, and a visitor notification; the conversation stays open. A named agent must be online, and an unavailable agent cannot complete the transfer. Rehearse with a fictional chat at shift change, including an unavailable recipient and department fallback, and confirm who owns the reply.

During a tawk.to live chat, use Whisper next to Reply when agents need to confer without showing the message to the visitor. The composer turns orange in Whisper mode; tawk.to says only agents in that conversation see a Whisper, and its history remains visible to agents after the chat ends. Before relying on that separation, have authorized staff rehearse a fictional two-agent chat and inspect the visitor view and ended chat while checking the composer mode before each send. If transcript exports or record retention matter, verify those separately; this Help article does not describe export treatment.

When using tawk.to's website chat widget, its Consent Form can ask for permission before chat and, when enabled, holds widget-related cookies and local storage until the visitor accepts. The European form can target EU/EEA or all visitors; a separately configured U.S. form applies in selected supported states and takes precedence there. Check the chosen targeting and Accept/Decline flow in fresh browser sessions for the intended visitor regions, including Canada. The vendor says this is a consent-collection tool, not a complete compliance solution; review the wording and legal requirements separately.

Before relying on tawk.to ticket counts for support follow-up, distinguish Open, Pending and Closed. New tickets from chat, offline messages or manual creation start Open. An agent reply normally moves an Open ticket to Pending, but a visitor or agent reply to a Pending or Closed ticket returns it to Open. The standard flow never closes tickets automatically; an agent can choose a status while sending or change it without a message. In a fictional no-message desktop inbox review, compare an unanswered ticket with one already replied to, assign ownership of Pending follow-up, and check how any automation or integration updates status before treating Closed as resolved.

When evaluating tawk.to's Pre-Chat Form, decide which fields are genuinely required: visitors can skip the form if none are mandatory, while required fields must be completed before chat starts. Submitted custom fields linked to Contact Attributes persist on the contact record. If AI Assist is in use, Exclude from AI makes the AI wait for the visitor's first chat message, while agents can still see the form details. In a fictional no-chat walkthrough, map a skipped form, a required field and a stored answer to the team's handling expectations before enabling the form.

If your team promises tawk.to visitors a chat transcript soon after an agent finishes, check the session boundary first. Its desktop/laptop Help says a chat that ends while the visitor remains active can stay in Monitoring until the session buffer expires; it moves to Inbox after the session fully ends. A visitor-requested transcript and email notifications are sent only after that full closure. For a fictional chat with no customer information, compare the agent finish, Monitoring, Inbox and requested email steps before promising immediate delivery. The guide does not establish a guaranteed email-delivery interval for your property.