Custom software maintenance is the ongoing work that keeps a bespoke system secure, working and useful after launch: fixing faults, applying security and dependency updates, adapting to changes in browsers, operating systems and connected services, monitoring, backups and small improvements. Without it, even well-built software decays, because everything around it keeps changing.
Last updated: 16 November 2026.
This guide explains what custom software maintenance involves in practice, the four types of maintenance work, how to organise it, where a company should start, what drives its cost and how to measure whether it is working.
Why does custom software need maintenance at all?
Code does not wear out, but its surroundings move constantly. Programming languages and frameworks release new versions and stop supporting old ones. Libraries the software depends on publish security fixes. Browsers and mobile operating systems change behaviour. Payment providers, maps, email services and other connected platforms change their interfaces and retire old versions. Data grows, users find new ways to use the system and the business changes its rules.
Each change is small. Left alone for a year or two, they add up to a system that is insecure, hard to change and expensive to bring back up to date. Regular maintenance keeps each step small.
At Odduel, a creative and technology agency in Barcelona and São Paulo, our maintenance and support service looks after websites, applications and bespoke business systems, including software we did not originally build.
In short: maintenance keeps bespoke software secure, compatible and useful. It covers fixes, updates, adaptation to changes around the system, monitoring, backups and small improvements. Organise it with clear ownership, a written agreement on response times and a regular schedule, and measure it by incidents, update lag and how quickly issues are resolved.
What does custom software maintenance involve in practice?
The work falls into four types. A healthy maintenance plan includes all four, not just fixing what breaks.
| Type | What it covers | Examples | When it happens |
|---|---|---|---|
| Corrective | Fixing faults found in use | A calculation error, a broken form, a crash on one device | As issues are reported or detected |
| Adaptive | Keeping up with changes around the system | New browser or OS versions, a payment provider’s new interface, a changed law | When the environment changes |
| Preventive | Reducing the risk of future problems | Dependency updates, security patches, removing unused code, refreshing tests | On a regular schedule |
| Perfective | Improving what exists | Faster screens, small features users ask for, clearer reports | Planned in a backlog |
Alongside these four sit the operational tasks that keep the system running day to day:
- Monitoring of uptime, errors, performance and unusual activity, with alerts to a named person.
- Backups taken automatically, stored separately and restored in a test at least a few times a year.
- Security reviews of access, accounts and logs, and timely patches.
- Hosting upkeep: server updates, certificates, storage and capacity.
- Documentation of how the system works, how to deploy it and how to recover it.
How should maintenance be organised?
Maintenance fails most often through unclear ownership, not lack of skill. Organise it deliberately.
- Name the owners. One business owner who decides priorities and one technical owner, in-house or at your agency, responsible for the system’s health.
- Agree response times. Define severity levels (system down, major feature broken, minor issue, request) and how fast each is acknowledged and resolved.
- Set a schedule. Monthly dependency and security updates, quarterly reviews of performance, access and backups, a yearly look at the platform’s major versions.
- Keep one backlog. Bugs, requests and technical improvements in one list, reviewed together.
- Use a staging copy. Test every update on a copy of the system before it reaches users.
- Record every change. What was changed, why, when and by whom, so issues can be traced and reversed.

In-house, agency or a mix?
Companies with a development team often keep maintenance in-house and bring in outside help for peaks or specialist work. Companies without one usually hand it to an agency under a maintenance agreement. A mix is common: an in-house product owner sets priorities while an agency handles updates, monitoring and fixes. Our article on choosing a website maintenance company sets out what to ask a provider, and the questions apply equally to bespoke software.
What happens when maintenance is neglected?
Neglect rarely causes a single dramatic failure. It shows up as a series of warnings:
- updates that were skipped become too large to apply safely;
- a library with a known security flaw stays in production;
- a connected service retires the interface the system uses, and a feature stops working overnight;
- only one person understands how to deploy, and they leave;
- backups exist but have never been restored, and do not work when needed.
Bringing a neglected system back usually costs far more than maintaining it would have. Sometimes it forces a rebuild that was not otherwise necessary.
What should a maintenance agreement include?
A good agreement for custom software is specific. It should state:
- the systems, environments and integrations covered;
- the tasks included: updates, monitoring, backups, fixes, a number of hours for improvements;
- severity levels, response and resolution times, and support hours;
- how requests are made and tracked;
- what is excluded and how extra work is quoted;
- who holds the code, the hosting accounts and the credentials, and how they are handed over if the agreement ends;
- security and data protection duties, including those under the GDPR where the system processes personal data, and a data processing agreement where the provider handles that data;
- reporting: what you receive each month.
Insist that the code, documentation and accounts belong to you. A maintenance arrangement should never be the only thing standing between you and access to your own software.
Where should a company start with custom software maintenance?
Start with a health check. Before agreeing any plan, find out where the system stands:
- Inventory. The system’s parts, hosting, connected services, accounts and who has access.
- Versions. The language, framework and key library versions, and how far behind current supported versions they are.
- Security. Known vulnerabilities in dependencies, access rights, password and sign-in practices.
- Backups. Where they are, how often they run, and whether a restore has been tested.
- Documentation. Whether someone new could deploy and recover the system.
- Known issues. Open bugs and the workarounds staff use.
The health check turns into a short catch-up plan for the urgent items, then a routine schedule. For a booking engine, a portal or a back office, the catch-up is often the largest piece of work; our article on booking system development shows how many connected services such a system relies on, each one a reason for regular care.
What drives the cost of custom software maintenance?
The cost depends on the system’s size and complexity, the number of connected services, how current its technology is, how critical it is to the business and the response times you need. A small internal tool with few integrations needs little; a revenue-critical platform with round-the-clock support needs much more.
The largest swing comes from the system’s starting state. Software kept current costs a fraction to maintain compared with software several major versions behind. A fixed monthly plan follows a health check, with improvements quoted separately or drawn from an agreed number of hours.
How do you measure whether maintenance is working?
Track a few measures every month and review them quarterly with the business owner.
- Uptime and incidents: how often the system is unavailable or a key function fails.
- Time to resolve issues by severity, against the agreed times.
- Update lag: how far the language, framework and libraries are behind supported versions.
- Open vulnerabilities in dependencies, and how quickly they are closed.
- Backup restore tests carried out and passed.
- Backlog health: requests waiting, and how long.
If incidents fall, update lag stays small and restores work, maintenance is doing its job. Our website maintenance checklist offers a task list you can adapt for bespoke systems.
Questions people ask
What is custom software maintenance?
Custom software maintenance is the ongoing work that keeps bespoke software secure, compatible and useful after launch. It includes fixing faults, applying security and dependency updates, adapting to changes in browsers, operating systems and connected services, monitoring, backups, documentation and small improvements, organised with clear ownership and an agreed schedule.
What are the four types of software maintenance?
Corrective maintenance fixes faults found in use. Adaptive maintenance keeps the system compatible with changes around it, such as new operating systems or provider interfaces. Preventive maintenance reduces future risk through updates and clean-up. Perfective maintenance improves performance and adds small features. A healthy plan includes all four types.
Where should a company start with custom software maintenance?
Start with a health check: list the system’s parts, hosting, services and access; check how far its technology is behind supported versions; look for known security vulnerabilities; confirm backups can be restored; and review documentation and open issues. The results give a catch-up plan for urgent items and a routine schedule.
Can another company maintain software it did not build?
Yes, if the code, documentation, hosting and accounts are available. The new provider usually starts with a health check and a handover period to understand the system. Missing documentation or access slows this down, which is why every maintenance agreement should confirm that the client owns its code and accounts.
How do you measure whether software maintenance is working?
Track uptime and incidents, time to resolve issues by severity, how far frameworks and libraries lag behind supported versions, open security vulnerabilities, backup restore tests and the age of waiting requests. Falling incidents, small update lag and successful restores show maintenance is working; rising figures call for a review.
Run a simple health check on your most important bespoke system this month: versions, vulnerabilities, backups and who can deploy it. If you want someone to take care of it properly, see our maintenance and support service or tell us about your system.


