Zammad vs Chatwoot for customer support

Compare Zammad and Chatwoot using an email-to-chat support journey, with routing, permissions, service targets, exports and operating costs.

Last updated

In short

  • Compare the whole support journey across channels, not inbox screenshots.
  • Evaluate ticket controls and conversation handling against your own escalation rules.
  • Keep channel providers, AI services and edition costs in the deployment boundary.
On this page (7)

Product comparisons · Part of Zammad

Choose around the unit of work

Zammad is a sensible first candidate when a support request becomes an owned ticket with a lifecycle, escalation rules and service targets. Chatwoot is a sensible first candidate when the working context is a conversation arriving through a customer messaging channel. Both require evaluation against the actual journey; this distinction does not imply that one has no chat or the other has no service-management capabilities.

The Zammad administration manual documents groups, roles, calendars, SLAs, triggers and channels. The Chatwoot help center covers shared conversations, website chat, channel connections and automation. The exact edition, external service and configuration determine what a deployed instance can do.

A workflow comparison

Requirement Question for Zammad Question for Chatwoot
Email intake Does threading preserve requester and ticket context? Does the email conversation preserve the same context?
Website chat Can an agent continue the case after the visitor leaves? Can the live conversation become an owned follow-up?
Escalation Can calendars and ticket conditions express the target? Can the chosen edition and workflow express the target?
Confidential cases Can groups and roles isolate the request? Can teams and permissions isolate the conversation?
Reporting Can the required ticket lifecycle metric be reconstructed? Can the required conversation metric be reconstructed?
Exit What ticket history and attachments can be extracted? What messages, contacts and attachments can be extracted?

A first response can be an automated acknowledgement, a human answer or a useful resolution. Define which one matters before comparing dashboards. Otherwise two apparently similar reports can measure different events.

Run the same support journey

The following is a reference exercise, not a completed product test. Use a fictitious customer, two support agents, a team lead and a billing specialist. Record versions, editions, channel configuration and business hours. Use a test mail domain and a nonproduction website widget.

The customer emails about an invoice, adds a PDF, then opens website chat about the same problem. The first agent asks for information. The customer replies after business hours. A specialist receives an internal note and resolves the issue the next morning. The customer reopens it a week later.

Evaluate whether the agent can associate the messages without exposing unrelated contacts or losing original timestamps. Check the behavior when the customer uses a different email address in chat. Automatic merging may be convenient but should never be assumed safe merely because names match.

Send a duplicate inbound message, an oversized attachment and an email with an unusual subject prefix. Disconnect the outbound mail provider briefly, then reconnect it. Record whether the system retries, exposes the failure to an agent and avoids duplicate replies. These outcomes matter more than a list of supported channel logos.

Routing and authority

Give the billing specialist access only to the intended scope. Attempt access through search, direct record URLs, exported reports and attachment links. Remove the specialist and repeat the checks. Also test whether agents can change assignment rules, export the complete customer database or invite another user without approval.

Define who owns routing logic. An infrastructure operator can keep a server available while a newly edited automation sends every case to the wrong group. Keep a small set of representative requests as a regression checklist whenever business hours, routing or templates change.

If AI assistance is enabled, trace the provider, transmitted message content, retention and human handoff. Evaluate incorrect suggestions with non-sensitive examples. Buying or self-hosting the inbox does not itself make an external model endpoint private.

Reporting and information retention

Create a manual timeline for the sample case: received, assigned, first human response, waiting on customer, resolved and reopened. Compare it with each product’s report. Decide how waiting periods and reopened cases should affect the metric. Keep the exported raw events necessary to explain any difference.

Set retention separately for messages, attachments, contacts, audit logs and backups. A deletion action in the interface does not by itself prove removal from object storage, provider logs or older backup generations. Establish the intended lifecycle and verify the relevant documented mechanisms.

Migration and operating effort

Import a small set of historical conversations before deciding whether to move all history. Preserve original IDs in a searchable field and compare author attribution, private notes, attachment counts and timestamps. Knowledge articles, custom reports, channel approvals and automations may need separate reconstruction. Do not assume a ticket CSV recreates the entire support operation.

At cutover, nominate one system to send customer replies. Keep the previous system available for reference with tightly controlled access, and document how new messages are reconciled if the pilot fails. An inbox rollback is a message reconciliation task as well as a configuration change.

Include channel-provider charges, mail delivery, licences, storage, AI consumption and agent administration in the comparison. Run a restore covering messages and attachments before accepting the service. The help desk must remain usable when the person who originally configured it is absent.

Which should enter your pilot?

Start with Zammad if calendar-based service targets and an auditable ticket lifecycle dominate the brief. Start with Chatwoot if website and messaging conversations dominate it. If both patterns are essential, evaluate both against the same fixture and consider retaining the current platform if neither meets the mandatory controls.

Pilae offers scoped Zammad and Chatwoot deployments. Bring the sample support journey and failed acceptance checks to the discussion; they make the operating scope concrete.

Sources and further reading

Use the linked documentation to confirm the version, edition and contract that apply to your project.

Design our support workflow

Define a support journey that exposes channel, routing, permissions and reporting requirements.