A web app is ready for production when its critical risks are understood, tested, assigned, and recoverable—not merely when the feature list is complete. This production-readiness checklist gives you a repeatable way to review security, reliability, performance, accessibility, operations, and rollback before a first launch or major release.
Overview
Use this guide as an application release readiness checklist for a new web app, a significant version change, or a move to a new cloud deployment. It is designed for teams using conventional servers, serverless app hosting, backend as a service products, or a broader modern app stack.
The goal is not to achieve a perfect score. The goal is to identify release-blocking risks and make explicit decisions about the risks that remain. A production review should answer five questions:
- Can the application protect users, credentials, and business data?
- Can the team detect failures and determine who owns the response?
- Can the application meet its performance and accessibility requirements?
- Can a release be deployed without avoidable manual steps?
- Can the team restore service or roll back safely if the release fails?
For a broader cloud-specific checklist, see the Production Readiness Checklist for a Cloud-Deployed Web App. Use this article when you need a practical release decision process that can be repeated for each deployment.
How to estimate production readiness
Start by listing the checks that matter for your application. Mark each item as Ready, Needs work, or Not applicable. Do not treat “not tested” as “ready.” For a simple readiness estimate, calculate:
Readiness percentage = completed, passing checks ÷ applicable checks × 100
This percentage is a planning signal, not a safety guarantee. A single unresolved authentication flaw or untested database rollback can be more important than dozens of completed low-risk checks. Add a second measure: the number of open release blockers.
| Area | Review questions | Release evidence | Owner |
|---|---|---|---|
| Security | Are authentication, authorization, secrets, and security headers configured? | Test results, configuration review, dependency review | Security or engineering lead |
| Data protection | Are sensitive values encrypted, access-controlled, retained, and deleted appropriately? | Data map, access tests, backup settings | Application owner |
| Reliability | Are health checks, timeouts, retries, capacity limits, and failure paths tested? | Test scenarios and operational runbook | Platform owner |
| Performance | Does the app meet agreed page, API, and background-job budgets? | Repeatable test results and budget thresholds | Technical lead |
| Accessibility | Can users navigate, read, and operate important flows with supported assistive methods? | Manual and automated checks | Product and QA |
| Delivery | Are approvals, migrations, environment variables, and rollback steps documented? | Pipeline record and release checklist | Release owner |
| Operations | Will the team see incidents and know how to respond? | Alerts, dashboards, escalation path | Operations owner |
Set a minimum bar before review begins. For example, require all critical checks to pass, require a named owner for every remaining issue, and require a tested rollback for changes that can affect data or authentication. The exact threshold should reflect the application’s users, data, and business impact.
Inputs and assumptions
A useful production launch checklist depends on clear inputs. Record these before reviewing the release:
- Application scope: Identify the changed pages, APIs, workers, integrations, and database objects.
- User and data sensitivity: Note whether the app handles accounts, payments, health information, internal records, or other sensitive data. Avoid assuming that a hosted platform automatically satisfies your requirements.
- Supported environments: Document supported browsers, devices, operating systems, regions, and network conditions. Browser requirements should be reviewed when the framework, UI library, or authentication flow changes.
- Performance budgets: Define acceptable limits for key page loads, API response times, asset sizes, background jobs, and error rates. Use budgets that match real user journeys rather than a single synthetic score.
- Recovery targets: Decide how much data loss and downtime the product can tolerate. Then confirm that backup frequency, restoration time, and rollback design support those targets.
- Ownership: Assign an owner for the application, infrastructure, data, security findings, monitoring, and the release decision. A checklist without owners becomes a record of observations rather than a control.
Security and data checks
Verify that authentication protects every relevant route and that authorization is enforced on the server, not only hidden in the interface. Test account recovery, session expiry, logout, privilege changes, and direct requests to protected resources. Confirm that secrets are outside source control, production credentials are separated from non-production credentials, and logs do not expose tokens or sensitive personal data.
Review security headers and browser-facing settings appropriate to the application, including transport protection, content restrictions, cookie attributes, and framing behavior. Also check input validation, output encoding, file upload limits, rate controls, dependency updates, and error responses. For authentication design choices, the Auth0, Clerk, and Amazon Cognito comparison provides a related planning reference.
Reliability, performance, and operations checks
Test timeouts, retries, idempotency, queue behavior, third-party failures, partial outages, and resource exhaustion. Confirm that health checks measure useful dependencies without causing additional load. Establish dashboards for availability, latency, traffic, errors, saturation, and important business events. Alerts should identify an actionable condition and route to a person or team.
Test keyboard navigation, visible focus, labels, contrast, responsive layouts, useful error messages, and the main authenticated and unauthenticated journeys. Then verify deployment approvals, environment configuration, database migration order, smoke tests, and a documented rollback command or procedure. The safe database migration checklist is particularly useful when schema changes accompany the release.
Worked examples
Example 1: A marketing site with a contact form
Assume 32 applicable checks. The team has 28 passing checks, two items marked not applicable, and two open non-blocking tasks. The readiness estimate is 28 ÷ 30 × 100, or about 93%. That result is encouraging only if the remaining checks do not involve form abuse, secret exposure, broken monitoring, or an untested deployment path. The release owner should record the two accepted risks and assign due dates.
Example 2: A subscription dashboard with a database migration
Assume 45 applicable checks, with 40 passing and five incomplete. The estimate is about 89%. However, one incomplete item is a migration rollback test and another is an authorization test for account administrators. This release should not be approved simply because its percentage is high. The open items affect data integrity and access control, so they are blockers until tested or formally redesigned.
Example 3: A serverless API with a third-party dependency
The API passes functional tests, but the review finds no timeout or fallback behavior when the external service is slow. The correct action is to add bounded timeouts, define the user-visible failure response, capture useful telemetry, and test the degraded path. A successful normal-path test does not establish production readiness when a dependency can fail.
Keep evidence with the release record: test links, screenshots of key configuration, migration notes, approval names, known risks, and the exact rollback steps. For API-heavy applications, pair this review with the OpenAPI documentation and testing workflow.
When to recalculate
Revisit this production readiness checklist before every major release and whenever a meaningful risk input changes. That includes a new authentication provider, database engine, hosting platform, framework version, browser support policy, payment or identity integration, data classification, or deployment pipeline.
Review it at least every six months even when the application appears stable. Framework recommendations, browser behavior, cloud platform settings, dependency risk, and internal ownership can change without a feature release. Recalculate after an incident, a failed rollback, a major performance regression, or a change in recovery requirements. The cloud application incident response runbook can help turn incident findings into new checklist items.
Before approval, run this final sequence:
- Freeze the release scope and record the commit, build, or artifact being reviewed.
- Recheck critical security, data, migration, and rollback items.
- Run smoke tests in the production-like environment.
- Confirm dashboards, alerts, ownership, and escalation contacts.
- Record accepted risks with an owner and review date.
- Approve, postpone, or reduce the release scope based on evidence.
- After deployment, verify key user journeys and watch the agreed signals before closing the release.
A production launch checklist is most valuable when it changes decisions. Keep it short enough to use, specific enough to test, and current enough to reflect the application you actually operate.