Skip to content
Interface language: English
Imperium / Signal

Maintaining a company website: who is responsible for what?

Agree responsibilities, checks and evidence for ongoing website maintenance.

Article language: English

A woman in a red apron beside a round glass lens in a gold-coloured housing.
In this article

“Website care” might mean regular checks, incident handling or being available to talk when someone remembers the site. Agree the actual scope before these meanings meet in one request. The offer must stay current, the form must deliver messages, and the company must control its domain and services. A good agreement says who checks what, how often and what result they report. After launch, the website still goes to work.

The scope depends on technology and the site's role. A few-page website needs different work from a portal with user accounts. In both cases, start with the customer journey: arriving, finding information, contacting the company and receiving an answer. Assign responsibility and a check to each part. An attractive first screen will not receive a message or renew a domain by itself.

Map services and responsibilities

List the domain, hosting, code repository, content system, form email service, analytics and important integrations. For each record the account owner, working contact and person responsible for renewal or billing. Keep access credentials in a tool designed for them, not a public project table. The map should show how to control a service, not distribute secrets.

Check dependencies. An email outage can stop contact while the site remains available. Expired access for one person may prevent a correction from being published. Assign cover for key roles and check that the substitute can perform the necessary actions. The guide to receiving code and access after an AI project (in Polish) explains account and documentation acceptance; maintenance develops those arrangements over time.

Do not assign responsibility to a collective “IT” or “marketing”. Name a role able to receive a request and, internally, a person and substitute. For content changes identify the information owner. A technical provider may correct a price-list layout but should not normally decide alone whether the price and offer terms remain valid. These are different responsibilities and different checks.

Separate scheduled checks, requests and development

Scheduled checks cover work done to a timetable, such as reviewing service expiry or testing a form. A request concerns a particular problem or change. Development introduces new scope, such as another language or an integration. When all three share a single allowance without rules, it is hard to see why a required review was missed.

Give each action a frequency or trigger. Do not assume one rhythm for everything. Offer accuracy may depend on a commercial decision; form testing may depend on its importance for enquiries. Also record how the outcome is recognised: check status, address, date and a brief account of deviations. “Site checked” alone does not say whether anyone sent a message.

An example operational scope may include:

  • Checking key addresses and the ability to use the site on a phone.
  • Testing contact with an identifiable message and confirmation of receipt.
  • Reviewing renewal dates for the domain and essential services.
  • Having content owners review contact details, team information and the offer.
  • Reviewing changes to technical dependencies appropriate to the technology.
  • Checking restoration of an approved version in the agreed environment.

This list is a starting point for agreement, not a universal package guaranteeing reliability. The provider should say which actions the offer includes and what materials it needs from the company. The site owner needs to know when a missing decision on its side will stop a change. Otherwise a recurring report can become a list of tasks no one can finish.

Measure the user's journey

A green server status describes a particular infrastructure component. It does not confirm that a customer saw the correct offer or delivered an enquiry. Google's SRE monitoring guidance distinguishes observing internal system behaviour from checking it externally. A practical implication for a company is to check the actual user journey alongside technical data.

For the form, agree a test address and a way to identify the message. A check should confirm not only the message after clicking but receipt at the agreed destination. The guide to B2B contact forms (in Polish) describes the scope in more detail. If messages go to a shared inbox, decide who recognises the test and removes it from the ordinary queue.

Measure performance in a way suited to the problem. One good result does not establish behaviour on all devices and connections. The guide to website speed and Core Web Vitals (in Polish) develops the diagnosis. The maintenance plan should say which deterioration triggers investigation and which data will be used to assess improvement. Do not commission random optimisation merely to refresh a number in a report.

When assessing a report, check the relationship between a metric and the work's goal. Improving one number may not show what happens to the form afterwards or work transferred to another person. Choose supplementary information for the specific user journey. The article on metrics and unintended behaviour (in Polish) examines what a single number can hide.

Agree a safe way to make changes

A request needs the intended result, the page or process and factually approved material. Record who checks the preview and who decides to publish. The provider needs to know whether a correction covers only Polish or also other languages, downloads and automatic messages. Without this information, the offer can diverge across channels.

Aeneia conceptual photography

For each significant change preserve a reference point and a way to reverse it. Test what the correction may affect. A phone-number change calls for checking the dial link; a form change also calls for checking delivery. For larger work use the website acceptance checklist guide (in Polish). A whole audit need not be repeated for every clearly scoped correction and its dependencies.

Agree periods when publication needs extra preparation, such as an important campaign or event. This is not a permanent freeze. The team should know the consequences and availability of people needed to check the result. If publication is planned for the end of a day and no one can receive a form message or assess the preview, change the timing before the operation starts.

Hypothetical scenario: an advisory firm's website

Imagine an advisory company publishing a site in two languages. In its first month marketing changes a service description, sales sends an older PDF and the only person who receives form messages is on leave. The site looks functional, but customers receive conflicting information. This invented example illustrates maintenance scope; it contains no real client data.

The company assigns a service owner, a shared register of current material and inbox cover. The technical provider receives the approved change and information about the language versions. After publication both pages, the file and the contact journey are checked. Only that result closes the request. Uploading text alone is insufficient because the task concerned consistent customer information.

At the next review the team notices that finding the right file manually takes most effort. It improves the source of materials instead of buying more hours for copying. Google's SRE guidance on eliminating toil describes recognising repetitive operational work before improving it. The solution's scale should match the company, not the tools of a large provider.

Review reports through decisions and open issues

A useful report shows checks performed, problems found, action taken and matters awaiting a decision. An open task names an owner and the required material. It need not contain a dozen charts. For a small site, a short evidenced list can be clearer than an automated document giving every check the same unexplained green mark.

At an agreed interval, review whether the scope still fits the site. A new integration or language may add dependencies the earlier agreement did not cover. A change of provider also calls for checking the responsibility map. Keep important reasons in a project decision register so the next person need not guess why an unusual setting exists. The guide to project decision records (in Polish) discusses recording the reasons and reviewing them.

Acceptance of the maintenance process can itself be practical: report a small change, find the current material, identify the service owner and reproduce the result of the last check. If the team cannot do these without one particular person's help, the plan needs improvement. A document works when it enables a task and leaves an understandable trail.

When several people support the site, a guide to internal service catalogues (in Polish) can help. Explain what the team accepts, what input it needs and what result it delivers. A request to correct text then need not follow the same route as a broken form.

How to compare two maintenance proposals

Ask providers to map services to the same tasks. Compare form checks, content updates, incident handling, version restoration and work involving external services. Packages with similar names may cover different scopes. Record the outcome and confirmation method for each action in a table. Only then can the company assess what care actually includes.

Separate time to acknowledge a request from time to diagnose and restore service. A reply is not a repair. If a proposal gives a deadline, check the triggering event and the situations it covers. Ask about availability outside normal hours and matters dependent on external providers. The aim is not to demand an unrealistic guarantee but to understand the handling path for a specific problem.

Check what happens when work needs a decision from the company. The provider should identify the material and owner needed, and the client team should confirm how it will supply them. If a correction waits for text, a report should not describe it as technically impossible. Clear states show the reason for waiting and support planning without shifting responsibility between departments. Include the time required from the service customer: preparing material, approving a preview and participating in acceptance. The article on the customer’s time in a service (in Polish) helps distinguish necessary participation from repeated avoidable work.

Ask where a small change ends. Adding a short paragraph and preparing a new page with a form require different effort and checks. Agree how scope outside maintenance is priced before an urgent task arrives. A clear proposal shows the process for identifying scope and when the company receives information needed to decide.

Walk through one sample request from description to acceptance. Check who has access, who approves content and where the result is recorded. This short exercise reveals more than a general promise of comprehensive support. If the proposal cannot reconstruct a specific task, ask for clarification before choosing a package. The maintenance scope can then be compared and accounted for later.

For website design and plans after launch, Aeneia Design describes its approach to the project and handover. Discuss any further support separately according to the site’s needs.

Aeneia / Aeneia editorial

Editorial article. Source links are included in the text.

About Aeneia

Brands we have worked with

  • L’Oréal
  • Garnier
  • Netflix
  • Carrefour
  • Warner Bros. Discovery
  • Biedronka
  • HBO Max
  • Wolt
  • Żabka
  • COSRX
  • Embassy of China in Poland
  • Mixa
Aeneia / Initium

From knowledge to a decision

Tell us what you want to put in order. We will agree on where to start.

Contact Aeneia

We partner with and use solutions from:

  • Google
  • OpenAI
  • Microsoft
  • Stripe
  • Anthropic
  • Linear