All articles
11 min read

Application Support Models: How to Choose the Right One

Three cream rotary telephones in a row on a steel counter, their cords running to a patch panel with one small amber indicator lamp lit.

There are three basic ways to support an application after launch: your own people do it, an outside team does it, or the two share the work. For most startups and small to mid-size teams, the shared model fits best. You keep the knowledge of your business in-house, and a senior technical team handles the code, the integrations and the incidents you cannot afford to fumble. At Ampersand we build and support software this way, and the rest of this article explains how to set it up so it works.

TL;DR

  • Three models cover almost every case: fully internal, fully outsourced, and mixed. Microsoft’s guidance calls the mixed model the most common one.
  • Support is organized in tiers. Basic user questions stay close to your team; code-level problems go to whoever can read and change the source code.
  • “Support” and “maintenance” are different jobs. Keeping an application running is not the same as keeping its code, dependencies and infrastructure current.
  • Before you sign anything, define fault classes, response times, support hours, who owns the data and what happens if you leave.
  • Choose the model by asking who can fix a code-level fault at 9 a.m. on a Monday, not by comparing monthly fees.

What an application support model actually is

An application support model answers one question: when something breaks, changes or confuses people after go-live, who does what, how fast, and under which agreement?

It covers four kinds of work. ERP Research, an industry content site, breaks ongoing support into break-fix, maintenance, user support and small enhancements. Break-fix means diagnosing errors, failed integrations and stopped jobs. Maintenance means patches, security fixes and updates. User support means answering questions and closing training gaps. Small enhancements are new reports or tweaked workflows. The same source notes that a contract covering only the first two leaves the other two on your internal team by default. That is where many companies get surprised.

A model also needs a pricing logic. A German guide to software support services describes monthly retainers or per-incident pricing for support, retainers or planned hour blocks for maintenance, and a monthly fee per service, environment or user count for managed services. It adds that many mature programs end up with a hybrid: a retainer for baseline operations plus time-and-materials for development.

The three main models

Fully internal

Your own staff handle everything, from the first user question to the code fix. Microsoft’s guidance for Dynamics 365 says this suits larger projects with broad internal resources and specialists for custom code and integrations. You get control, but you must manage swings in demand yourself.

The practical risk for a small team is key-person dependency. If the one developer who understands the integration with your shop or ERP is on holiday, ill or gone, support stops. ERP Research flags the same key-person risk for in-house teams.

Fully outsourced

An outside provider carries support end to end. Microsoft describes this as the best fit for smaller projects without many critical business processes, and notes it needs a strong relationship with the partner. Deloitte’s description of a full outsourcing arrangement is long-term support with no internal capability building.

That last point is the thing to weigh. If the provider holds all the knowledge and you hold none, switching later is expensive. A good outsourced arrangement therefore includes documentation, access to your own repositories and hosting accounts, and a clear handover path from day one.

Mixed (co-sourced)

Internal staff handle simple issues, training and triage. External specialists provide strategic advice, peak coverage and custom-code support. Microsoft calls this the most common model.

Deloitte describes a related pattern it calls co-sourcing, where the provider and the client support the system together. Its example: the provider takes integration support while the client takes functional support. A simple illustration is a sales team running a CRM. Your staff handle questions about how deals are tracked, while the engineers handle the sync between the CRM and other systems. A write-up on Attio CRM integrations shows how many moving parts one CRM connection can involve, which is why the integration half of that split is worth giving to someone who does it daily.

Deloitte also describes a fourth arrangement, staff augmentation, where an outside party supplies one or two people to build your internal capacity. It suits companies that want to grow their own team and need experienced help while they do.

A quick comparison

Fully internal Fully outsourced Mixed
Who fixes code-level faults Your developers The provider The provider
Who answers user questions Your team The provider’s service desk Your team
Main strength Control No hiring, predictable fee Knowledge stays with you, expertise on call
Main thing to manage Key-person risk, uneven workload Knowledge transfer and exit terms A clear line between the two sides
Typical fit Larger teams with specialists Smaller, less critical systems Most startups and growing businesses

The table follows the three models in Microsoft’s guidance. For a team of three to thirty people with one business-critical application, the mixed column usually wins, because hiring a full-time specialist for code and integrations is hard to justify.

Support tiers: who handles what

Whichever model you choose, faults travel up a ladder. The clearest public description comes from a software maintenance agreement published by INEXTO:

  • First level (L1): registers and classifies each request and starts resolving it right away, escalating if it stays unsolved. In that agreement the client provides L1.
  • Second level (L2): has in-depth knowledge of the software’s architecture and infrastructure. It handles troubleshooting, configuration, work-arounds, data patching and issue analysis, and escalates source-code problems.
  • Third level (L3): has source-code access and does code patching, bug correction, full root-cause analysis and disaster recovery. In the same agreement the service provider is responsible for L3.

Microsoft’s guidance uses five levels, running from super-user assistance and a helpdesk to an internal application team and, at the top, an external partner or internal IT team handling complex issues such as custom code and integrations. The names differ, but the logic is the same: the closer a fault is to the code, the fewer people can fix it.

For a small company, this gives a simple rule. Keep L1 where the users are, because your colleagues know your processes. Put L2 and L3 with people who can see the code, the servers and the logs. If one person on your payroll covers all three, you have a single point of failure, not a support model.

Support, maintenance and managed services are not the same thing

Buyers often compare quotes that cover different jobs. The German guide cited above separates three:

  • Software support: keeping the application running through incident handling, user requests, monitoring and small fixes.
  • Software maintenance: keeping code, dependencies and infrastructure current through patches, updates, refactoring and performance work.
  • Managed services: operational ownership of a defined service, such as hosting, monitoring, security operations, backups, a service desk and reporting.

ERP Research adds a point that applies well beyond ERP. The common confusion is to treat vendor maintenance and an application management contract as alternatives. Vendor maintenance keeps the software itself supported, while application management supports your own implementation: the customizations, integrations and reports. Many organizations need both, and support costs are routinely underestimated.

For a custom-built website, web app or platform, there is no vendor shipping patches for your specific code. The team that built it is, in practice, the vendor. That is the case for choosing a support partner who knows the build.

How managed application support works

In a managed arrangement, a provider takes operational ownership of a defined scope and reports against it. The scope is the key word. A typical scope names the environments covered (production, staging), the monitoring in place, the backup routine, the release and patch process, and the channel through which your team raises requests.

The fee is usually monthly. The deinitteam guide describes managed-service fees as set per service, environment or user count. That makes the cost predictable, and it makes the scope definition the most important page of the contract. Anything outside it, such as a new feature, is normally billed separately, which is why the hybrid of a baseline retainer plus project work is so common.

Managed support fits you when:

  • you run one or a few business-critical applications and do not want to hire a full-time operations person;
  • your team knows the business but not the infrastructure;
  • you would rather pay a stable monthly amount than absorb surprise bills after every incident.

Deloitte cites figures that explain why this matters: 29% of businesses cite a lack of technical skills as a barrier to technology adoption, and 50% point to maintenance and ongoing support after launch as a key barrier to scaling. Support is not an afterthought. It is what decides whether a product can grow.

What to define in any support agreement

A model is only as good as its written terms. A Swiss example makes this concrete. The SaaS maintenance and support contract of Swiss software company Zetcom shows the elements worth specifying:

  • Fault classes. Zetcom uses classes A, B and C. Debugging starts within 4 working hours for A, 1 working day for B and 4 working days for C, with maximum debugging deadlines of 2, 4 and 10 working days.
  • Support hours. The same contract states support is available Monday to Friday, 09:00 to 12:00 and 13:30 to 17:00. Note how that differs from a 24/7 expectation.
  • Data ownership. The customer remains the sole beneficiary of its data and can require a handover.
  • Infrastructure and maintenance windows. The contract sets out redundant infrastructure and scheduled maintenance windows.

You do not need to copy those numbers. You need to answer the same questions for your own application:

  1. What counts as critical? Checkout down and a typo on the about page are not the same fault. Write down the classes.
  2. How fast is the first response, and how fast is the fix? These are two different clocks.
  3. During which hours is support available, and what happens outside them?
  4. Who owns the code, the data and the accounts, and can you take them to another provider?
  5. How are changes billed? Define what is a fix inside the retainer and what is a new piece of work.
  6. What documentation exists so another engineer could take over?

The exit question is not pessimism. It is what keeps a support relationship honest. A provider confident in its work has no reason to resist it.

How to choose: five questions

Research on large German and Swiss companies points to the same lesson. A 2013 case study of eleven of them found that arm’s-length outsourcing went with high internal maturity in managing the application portfolio, low-to-medium complexity and high standardization. In plain terms: the more standard and well documented your systems, the easier it is to hand them over. The more customized they are, the more you need a partner who understands them closely. Large corporations have dedicated staff for this. You probably do not, so use these questions instead:

  1. Do you have anyone who can read and change the code? If not, L3 must be external.
  2. How critical is the application? If an hour of downtime costs real revenue, you need defined response times, not best effort.
  3. How custom is it? Heavily customized systems with integrations to a shop, CRM or ERP need support from people who know those connections.
  4. How much knowledge do you want to keep? If you want your team to grow into the system, choose a mixed model, or staff augmentation to build capacity.
  5. How stable is your demand? Spiky workloads, such as a launch or a seasonal peak, favor an external partner for peak coverage.

If you answer “no” to the first question and “yes” to the second and third, you are looking at a mixed or outsourced model with a senior team behind it.

Where Ampersand fits

We are a small Zurich studio that builds websites, web applications and platforms, and keeps them running. Every project is led by senior engineers from the first consultation to handover, so the people who scoped your system are the people who support it. We aim for your independence rather than lock-in: the goal is that you understand and control what we built.

Our pricing page lists the support and advisory plans, among them Run & monitor at CHF 750 per month and Guidance at CHF 799 per month, along with Run & improve and the build, automation, integration and migration services that connect to them. For a mixed model, we take the code-level and integration work while your team keeps first-line support and the business knowledge. Clients such as UBS and the City of Lugano work with us.

FAQ

What is the difference between application support and production support?

Both keep a live application working: handling incidents, answering user requests and monitoring. The names vary by company, and some use “production support” for the operations side only. What matters is that your agreement lists the tasks, because maintenance such as patches and dependency updates is a separate job from support.

Which support model is best for a startup?

Usually the mixed model. Your team handles user questions and triage, and an external senior team handles code, integrations and incidents. It avoids hiring a full-time specialist too early, and it keeps knowledge in your company. Microsoft’s guidance identifies mixed as the most common model.

What are L1, L2 and L3 support?

L1 registers and classifies requests and resolves simple ones. L2 does deeper troubleshooting, configuration and work-arounds. L3 has source-code access and fixes bugs at the root. In the INEXTO agreement, the client provides L1 and the provider is responsible for L3.

What should a support contract include?

Fault classes with response and fix times, support hours, maintenance windows, data ownership with a handover right, and a clear line between what is covered and what is billed as new work. The Zetcom SaaS contract is a Swiss example that covers all of these.

Is managed support more expensive than hiring someone?

It is billed monthly against a defined scope, so the cost is predictable and you carry no recruitment, holiday-cover or key-person risk. Compare it to the full cost of a specialist hire plus cover for absences, not to a salary alone.

Sources

Talk to us

Have a project this touches on?

A free 10-minute call is the fastest way to find out whether we are the right studio for it.

Book a free 10-min call