Important website work can fall between a developer, hosting company and business owner. A missing responsibility is a better starting explanation than assuming someone is hiding a problem.
These eight checks help make that responsibility explicit. They do not establish how common each problem is or predict a fixed loss of sales.
What a Public Scan Can and Cannot Establish
A website health check can provide a starting point for public observations. Treat its findings as scoped checks at a point in time. It cannot verify private backups, internal permissions, incident response, historic availability or ownership rights. Lab performance tests are not measurements from all your customers.
Problem #1: Certificate Renewal Has No Owner
A failed TLS certificate renewal can prevent visitors from connecting without a warning. Check the active certificate, renewal mechanism, failed-renewal alerts and who responds. Chrome’s connection guidance explains the connection indicators and warnings.
A certificate expiring soon is not automatically an emergency: many systems intentionally renew short-lived certificates close to expiry. What matters is whether renewal succeeds with enough time to fix failures. Do not depend on a remembered date alone.
Ask: “Can you show a successful renewal and what happens if the next renewal fails?”
Problem #2: Backups Have Not Been Restored
A backup job reporting success does not show that the data can be restored into a usable service. Test a recovery copy in an appropriate environment and check the application and data, not only the archive file.
CISA’s ransomware guidance recommends protected backups and regular recovery testing. Choose frequency and retention around changes, business impact and recovery objectives. There is no verified universal backup-failure or bankruptcy percentage in this article.
Ask: “When did we last restore, what was missing, and how long did it take?” The backup questionnaire can structure the discussion but cannot inspect a private backup.
Problem #3: Performance Is Judged From One Device
Images, scripts, network conditions, caching and application work can change performance. Establish whether a problem affects a particular page, device, region, interaction or traffic level.
Use real-user and lab measurements together. A PageSpeed score or a customer’s complaint is a reason to investigate, not evidence of a universal conversion-loss percentage. A low-traffic site may lack public field data and need its own monitoring.
Ask: “Which users are affected, and what measurement explains the delay?”
Problem #4: Security Updates Have No Risk-Based Process
Track supported versions, advisories and exposed components. Prioritize vulnerabilities according to exploitation, impact and exposure, with testing and recovery appropriate to the change. “Latest version” is not the only way a vendor delivers security fixes, because supported releases can receive backported patches.
For WordPress, official hardening guidance covers trusted software, updates and access. Neither “updated last month” nor a fixed plugin count is proof that a system is secure.
Ask: “Who tracks security fixes, and how do we know urgent fixes are applied?”
Problem #5: Nobody Owns Alerts or Recovery
Logging preserves events. Monitoring interprets signals and may trigger alerts. Neither guarantees an outage will be predicted or an alert will receive a response.
Monitor important customer actions, not just whether the homepage returns HTTP 200. Define who responds, their coverage hours and what they can do. Use the downtime scenario calculator with business-specific assumptions rather than treating an enterprise survey as your own hourly loss.
Ask: “Can we demonstrate a failed customer action producing an alert to the right person?”
Problem #6: Mobile Journeys Are Not Tested
Test contact, login and checkout on supported phones, including touch, autofill, the on-screen keyboard and error recovery. Desktop emulation is useful but does not cover every physical-device behavior.
Use your actual visitor and customer mix. Mobile traffic is not automatically lost revenue when a page has a defect, and traffic share is not the same as conversion share.
Ask: “Which key journeys were last completed on iOS and Android, and what failed?”
Problem #7: Hosting Costs Have No Explanation
Inventory resources, usage, renewal prices and support commitments. A large bill can be justified; a small bill can still include waste. There is no defensible monthly amount below which every small-business setup is appropriate.
Compare alternatives over the same period and requirements, including migration and operational time. The cloud cost guide and hosting comparison explain how to make assumptions explicit. They do not promise a universal percentage saving.
Ask: “What does each recurring charge support, and which changes have a measured net benefit?”
Problem #8: The Business Cannot Recover Its Accounts
Check business control of the domain, DNS, hosting, email, repository, analytics and integrations. Use named delegated users rather than sharing a contractor’s personal login. Verify recovery email, MFA and billing contacts.
Administrative access is distinct from copyright, contractual rights and a tested recovery copy. Follow the ownership transfer checklist to document each one.
Ask: “Can another authorized person access and operate these systems if our current contact is unavailable?”
Start With Evidence and Responsibility
Choose the unresolved issue with the greatest practical impact. Record the evidence, owner, next action and date, then verify the result. The 50-question checklist supports a wider review.
A responsible answer can include uncertainty and a need to investigate. If significant gaps remain, an independent review can help define the work. No public scan or confident conversation replaces private-system evidence.


