How to organise maintenance for your business website

Maintenance keeps a website useful after publication. It includes correct content, journeys that continue to work, updated components and the ability to recover from a failure. The plan should match what the website contains and the impact of an interruption. A task report only has value if it shows what was checked and the result.

By Lyntev. Updated on .
About 5 minutes to read.

From data to the next decisionChoose the question, define the calculation and act on the result.

Translation published on . Original Portuguese guide; source revision .

By the end, you can:

  • Distinguish technical maintenance from content updates
  • Prepare changes with a recovery path
  • Record failures, tests and responsibilities

Create an inventory another person can use

List the domain, DNS, hosting, code, content, database, files, email and external services. For each item, identify the supplier, owner, renewal, access and dependencies. Include components that do not appear in the interface, such as scheduled tasks and notifications. Keep credentials in an appropriate place; the inventory points to that place without reproducing secrets.

Separate who approves, who carries out the work and who follows up the business activity. The technical team can fix form delivery, but someone at the company needs to confirm that enquiries are handled. Define a substitute for critical tasks and a contact channel if the service becomes unavailable. The plan should remain workable when the usual person is absent or the supplier changes.

Organise the routine around impact and changes

Distinguish technical maintenance, content updates and new features. Fixing a sending failure, changing opening hours and adding a customer area have different scopes. The proposal and work record should make clear what is included and what needs additional assessment. A small wording change may be simple; a small change to a commercial rule may affect several journeys.

Define frequency according to risk and use. A shop with continuous orders needs different monitoring from a rarely updated information page. Some checks belong to every publication; others to an agreed cycle or a specific alert. Avoid a rigid monthly list that postpones a critical failure until the next report. Priority should follow the observed impact.

Organise the routine around impact and changes
WorkWhen to reviewEvidence
Contact journeyAfter changes and on the agreed cycleTest enquiry received
Terms and contact detailsWhen the business changes informationApproved source and updated pages
Dependencies and platformAccording to updates and identified riskVersion and change tests
RecoveryIn a planned trial and before significant changesRestore demonstrated in an appropriate environment

Prepare the update before applying it

Read what is changing and identify affected components. Save the previous state, confirm a recovery method and test in an appropriate environment. An available version does not, by itself, tell you whether the change is compatible with the website. Check integrations, extensions and custom functions. In WordPress documentation, for example, preparation includes backups and verification; on other platforms, follow the corresponding procedure.

Keep changes sufficiently bounded to investigate a regression. Record versions before and after, altered configuration and tests performed. After publication, confirm that the browser receives current resources and the cache is not keeping an old combination. Test real functions, not just whether the homepage opens. A form can fail while the website looks entirely normal.

Test restoration, not just whether a backup exists

Identify the set needed to rebuild the service: files, data, configuration and secure references to access credentials. In WordPress, the documentation distinguishes the database from files; downloading only one part may be insufficient. On a static website, code and publication may be recoverable, but forms and orders may depend on data stored in another service.

Define the data loss and downtime the operation can tolerate, and choose the procedure accordingly. Test recovery at a controlled destination without triggering real emails or payments. Compare expected pages, attachments and records. Record duration, gaps and missing access. A backup can exist and still be unusable because of corruption, incompatibility or missing configuration.

Fictitious example: the website opens but enquiries do not arrive

At a fictitious company, a review finds that the form shows confirmation but notifications stopped arriving after an email change. The team starts by checking enquiry records and the configured destination. It establishes whether the problem concerns saving, notification or handling by the team. These possibilities need different actions; resending everything without analysis can create duplicates and confusion.

In the illustrative trial, enquiries were saved and the failure affected notification. The destination is corrected, an identified enquiry is sent and receipt is confirmed. The company reviews pending enquiries through the authorised record and decides the appropriate follow-up. The report explains the cause, period observed, correction and prevention: test the complete journey whenever an associated email service changes.

Investigate a contact failure
  1. Interface

    What was shown to the person?

  2. Record

    Was the request saved?

  3. Notification

    Did the notification reach the right destination?

  4. Handling

    Is someone responsible for follow-up?

Monitor signals that lead to an action

Choose checks that represent useful website availability. Receiving an HTTP response may show that a server responds, but not that the content and function are correct. A controlled journey test should have identification, limits and appropriate cleanup. Define who receives each alert, which information they need to consult and when they should escalate the problem.

Common errors include alerts with no owner, backups stored only alongside the service they may need to restore, updates with no record and reports listing actions without results. Also avoid impulsively restoring old data: new enquiries may exist that were not in the backup. Before recovery, identify what changed since the saved version and prepare to preserve or reconcile that data.

Close every intervention with a result and next step

The worksheet should explain why the work was done, what changed and how it was confirmed. Include outstanding issues with an owner and review date. If nothing needed changing, record the tests and results rather than inventing work. Continuity improves when another person can distinguish a conscious decision from a forgotten task.

Worksheet for your project

Service, domain and owners:
Inventory and dependencies affected:
Reason for the intervention or review:
Previous and new version/configuration:
Backup and recovery procedure:
New data to preserve if restoring:
Tests before the change:
Tests at the published address and results:
Incidents, cause and period observed:
Outstanding issues, priority and owner:
Alerts and renewals to follow up:
Date of the next review:

Sources and criteria

Content prepared with AI assistance and research using technical sources, with checks of the relevant examples and calculations. The references support the guide’s decisions; examples identified as fictitious let you try out the method.

Wellinton Oliveira

Written by

CEO of Lyntev

Technology with context, original analysis and sources you can consult.