Practical guides
How to launch a client app on a custom domain without breaking email
A custom-domain launch checklist for agencies: preserve DNS and email, verify the intended app and HTTPS, prepare rollback, and give the client a maintainable handoff.
Connecting a client application to app.example.com can look like a one-record job. The domain may also support the client's email, main website, sign-in callbacks, and verification records. A successful web launch needs to preserve those existing services.
This guide gives agencies and consultants a proposed preflight, launch, and handoff method. The goal is a smaller, reviewable change with evidence of the result. It cannot guarantee an outage-free migration. Follow the current instructions for the provider and project you actually use.
Separate the responsibilities before editing DNS
Write down who controls the domain registration, authoritative DNS, application hosting, email service, and HTTPS certificate management. These responsibilities may sit in different accounts even when one company sells several of them.
For the handoff, the client should know the registrar, renewal owner, and recovery route. ICANN's registrant guidance covers registration rights and responsibilities; it does not make an agency the appropriate permanent owner of the client's account. Agree ownership and authorized access before the launch.
This distinction also helps answer a practical question: “If the agency is unavailable tomorrow, who can renew the domain and recover the application?”
Choose the hostname in the accepted brief
For a new client portal, consider a dedicated hostname such as portal.example.com or app.example.com when the client's architecture permits it. This is our design recommendation: keeping the existing main website in place can make the intended change easier to identify and review.
It is not a universal deployment rule. The right hostname depends on branding, authentication, existing systems, and hosting support. Record that decision before implementation. Do not replace the client's main website simply because a tutorial used the root domain.
Confirm how the exact hostname will appear in sign-in redirects, approved callbacks, customer links, and any integrations included in scope. Those checks belong to the application's acceptance plan.
Inventory the existing services
Before changing records, keep a private copy of the current DNS zone and identify which records support which service. Note the original value of every record the launch might change. A screenshot of one visible row may miss records farther down the zone.
Include:
- Existing website and application hostnames.
- Incoming-mail MX records and the current mail provider.
- Mail-authentication records, including SPF, DKIM, and DMARC.
- Ownership and service-verification records.
- The authorized people who can make and reverse the change.
Google Workspace explains that MX records direct incoming email delivery. Its guidance also says working legacy Workspace MX configurations do not need to be changed merely to adopt its newer setup. Connecting a web app is not a reason to replace a functioning mail configuration.
Treat the complete zone inventory as private operational information. A public article or client-facing screenshot should not expose recovery details, tokens, or unnecessary infrastructure data.
Get the exact DNS target from the application host
Add the intended hostname in the hosting platform and obtain the record values for the actual project. Do not guess a target or copy another customer's value.
Provider requirements differ. Vercel documents different setup paths for apex domains and subdomains, including project-specific CNAME targets for subdomains and additional verification when applicable. The useful general lesson is to compare the authoritative DNS records with the values shown for your project.
If the required change is broader than the approved plan, pause and revise the plan. Moving nameservers requires a review of the whole zone and its services. It should not become an improvised extra step during a one-hostname launch.
Validate the hostname and HTTPS separately
Domain ownership, DNS routing, and certificate readiness are related checks, but one positive status does not establish the others.
Cloudflare for SaaS explicitly separates hostname ownership validation from certificate validation. Its production requirements include an active hostname, active certificate status, and DNS pointing to the intended SaaS target. Other platforms expose their own status names and processes.
Follow the sequence for the selected provider. Where supported, complete the available pre-validation before switching traffic. Do not promise a specific propagation time or zero downtime from a generic checklist.
Keep the platform's current instructions and the observed statuses in the private release record so another operator can understand what was checked.
Make one approved change and verify the application
Identify the person authorized to perform the cutover. State the exact record being changed, its previous value, the intended value, the rollback trigger, and the person available to reverse it.
After cutover, collect evidence for the application itself:
- The intended hostname resolves to the expected destination.
- The hosting platform reports the required readiness states.
- HTTPS opens without a certificate warning.
- The response is the accepted application release, rather than an old or unrelated deployment.
- An ordinary user in a clean browser session can complete the critical workflow.
- The expected login redirects and in-scope callbacks still work.
A homepage response alone is a limited observation. It does not demonstrate a booking, quote, file upload, or customer account workflow. Reuse the client-app handoff checklist to test the actual business job.
Check what was supposed to stay unchanged
Repeat the baseline checks for the existing main website and relevant subdomains. With the client's authorization, use a non-sensitive test message to check incoming and outgoing email. Verify the existing mail configuration and any other services covered by the accepted scope.
Record a result for each check. Avoid treating “we did not intentionally change it” as evidence that it still works.
If a check fails, stop unrelated changes. Compare the observed failure with the original inventory and the exact cutover. Diagnose or roll back through the agreed route before calling the launch complete.
Define rollback before the launch
A useful rollback plan identifies the previous record value, the authorized operator, and the condition that triggers restoration. For example, the team might agree to reverse the cutover if the intended release is unavailable at the agreed checkpoint or a critical existing service fails.
Choose thresholds for the particular project rather than copying a deadline from this article. Consider how caching and provider behavior affect both a change and its reversal.
After restoring the previous configuration, verify service recovery. “We reverted DNS” records an action. “The original site and agreed email checks pass again” records an outcome.
Give the client a handoff they can use
Include these items in the private handoff packet:
- Ownership: registrar, DNS provider, hosting owner, renewal responsibility, and client administrators.
- Change record: hostname, previous and current values, purpose, and approved operator.
- Release evidence: deployment identity, acceptance date, tested workflows, and remaining limits.
- Email and HTTPS: providers, management responsibilities, and observed checks.
- Operations: monitoring owner, support route, recurring costs, and rollback instructions.
- Access: which temporary agency permissions should end, and when.
Use the approved account-invitation or secret-management process for credentials. Do not paste passwords, API secrets, or recovery codes into the handoff.
Copy this launch brief into your project
Launch the accepted application at the agreed client-owned hostname. Inventory existing DNS and preserve unrelated website, email, and verification records. Use the exact record values supplied for the selected hosting project. Verify hostname readiness, HTTPS, the intended release, the critical user workflow, and existing services. Keep the previous configuration and a named rollback owner. Deliver ownership, access, costs, support, and acceptance evidence to the client.
You can use this brief when evaluating or working in Admas. The article supplies a planning method; it does not establish that Admas automatically completes every registrar step or that its custom-domain flow has passed an end-to-end acceptance test for your provider. Confirm the current supported flow and verify the result before promising it to a client.
Method and evidence
An original proposed launch and handoff method, source-checked on November 10, 2026 against current Cloudflare, Vercel, Google Workspace and ICANN documentation. Provider examples describe their documented behavior; use the actual project's current settings. No customer outcome, zero-downtime result, or Admas custom-domain end-to-end test is claimed. Admas owns this publication; no sponsorship or affiliate arrangement is asserted.
Put the idea to work
Bring your next useful problem.
Describe a workflow, inspect the result, and keep improving it in Admas.
Start a project