Custom Connector Development: Build, Buy or Commission? A Time-and-Cost Model
Last updated: November 2026
When a system needs to be integrated and no connector exists, organizations have three options: build the connector in-house, commission a system integrator, or use an integration platform that delivers custom connectors as part of its service. The right choice depends on total cost of ownership, time to value, maintenance burden, security and the number of systems involved. This article provides a reusable time-and-cost model, explains the hidden costs of each option and lists the questions to ask before deciding. ZigiOps, which typically builds new connectors in about three days, is used as an example of the platform-delivered option.
Connector gaps are a normal part of enterprise integration. No catalog covers every internal application, legacy system or niche product. When a gap appears, the decision about how to fill it is often made quickly, based on who is available rather than on cost and risk. That can lead to integrations that are expensive to maintain, hard to secure and dependent on a few individuals.
The pressure to decide quickly is understandable. A business team is waiting, manual work continues every day, and a developer offers to "write something over the weekend". The resulting script may work, but it rarely includes retries, monitoring, credential rotation or documentation, and it quietly becomes part of the organization's critical infrastructure. Months later, nobody remembers how it works, and every change becomes a small project.
A simple, structured model helps make this decision deliberately.
The three options
Option A: Build in-house
Internal developers write the code that connects the system to the target platform, usually as a service, script or plug-in.
Advantages: full control, deep knowledge of internal systems, no external dependency.
Disadvantages: developer time is diverted from core products; error handling, security, logging and bi-directional logic must be built from scratch; long-term maintenance stays internal.
Option B: Commission a system integrator
A consulting partner or system integrator designs and builds the connector, often as part of a broader project.
Advantages: access to specialist skills, defined project scope, no internal development effort.
Disadvantages: higher hourly costs, change requests for every modification, knowledge transfer challenges, and ongoing dependency on the partner.
Option C: Platform-delivered custom connector
An integration platform vendor builds the connector for its platform, after which it is configured like any standard connector.
Advantages: fast delivery, inherits the platform's security, logging, mapping and synchronization features, maintained as part of the platform.
Disadvantages: dependency on the platform vendor; only viable if the vendor can reach the system's interfaces.
The time-and-cost model
Total cost of ownership should cover at least three years, because most of the cost of a connector comes after go-live. The model below lists the cost components to estimate for each option.
| Cost component | Build in-house | Commission integrator | Platform-delivered |
|---|---|---|---|
| Initial design and build | Developer time for connector, mapping, sync logic, error handling | Project fees | Connector delivery, often included or fixed-fee |
| Security review | Full review of new code and infrastructure | Review of partner code and access | Platform already reviewed; connector-specific review only |
| Hosting and operations | Servers, monitoring, on-call | Depends on contract | Part of platform deployment |
| Maintenance per year | Fixes, upgrades, API changes, new fields | Change requests and support contract | Configuration changes by administrators |
| Knowledge risk | High if few developers understand it | Medium to high, depends on handover | Lower, standard platform skills |
| Opportunity cost | Developer time not spent on products | Internal time managing the partner | Minimal |
Typical timelines
Timelines vary widely, but some patterns are consistent. An in-house connector that includes proper error handling, security, logging and bi-directional sync often takes several weeks to a few months, depending on complexity and developer availability. A commissioned integrator project usually takes a similar or longer time once scoping, contracting and handover are included. A platform-delivered connector can be much faster, because the platform already provides everything except the system-specific access layer.
ZigiOps, for example, typically builds a new connector in about three days. Once delivered, the integration is configured without code using the same field mapping, conditions and synchronization features as for ServiceNow, Jira or Salesforce.
| Option | Indicative timeline | Main drivers |
|---|---|---|
| Build in-house | Several weeks to months | Developer availability, scope of sync logic, security review |
| Commission integrator | Several weeks to months | Contracting, scoping, partner capacity, handover |
| Platform-delivered | Days to a few weeks | Connector delivery, configuration and testing |
The hidden costs of custom code
The initial build is the visible part of a custom connector. The hidden costs come later:
- API and schema changes. Each change in the source or target system can break the connector.
- Credential management. Tokens expire and must be rotated; secrets must be stored securely.
- Incident handling. When the connector fails, someone must diagnose and fix it, often urgently.
- Feature requests. Users ask for new fields, filters or directions, each requiring development.
- Knowledge loss. When the original developer leaves, maintenance becomes slower and riskier.
- Audit and compliance. Custom code must be included in security reviews, penetration tests and audits.
Many organizations underestimate these costs by a wide margin because they are spread across teams and years.
When each option makes sense
Build in-house when
- The integration is a core differentiator of your product.
- You have dedicated integration engineers with capacity for long-term maintenance.
- The requirement is so unusual that no platform can support it.
Commission an integrator when
- The connector is part of a larger transformation project the integrator is already delivering.
- You need specialist expertise in a particular platform for a one-time effort.
Use a platform-delivered connector when
- You need several integrations over time, not just one.
- Speed matters.
- Administrators, not developers, should maintain the integration.
- You want consistent security, logging and governance across all integrations.
- The system can be reached through an API, SOAP, a database or a library.
Questions to ask before deciding
- How many systems will need custom connectors in the next three years?
- Who will maintain the connector after go-live, and do they have capacity?
- What happens when the source system is upgraded?
- Does the option support bi-directional synchronization, filtering and error handling out of the box?
- Does it store transferred data, and where?
- How will security teams review it?
- Can the system be reached without building a new API?
Security considerations
Every custom connector expands the attack surface. In-house and integrator-built connectors need full security reviews of code, infrastructure and credentials. Platform-delivered connectors inherit the platform's security controls, so reviews can focus on the connector's access to the specific system.
When evaluating a platform, check its certifications, deployment options and data handling. ZigiOps is ISO 27001 certified, can run on-premise or in the cloud, and does not store the data it transfers, which simplifies security reviews for new connectors.
A worked comparison
Consider an organization that needs to connect an internally built maintenance system, with a SQL database and no API, to ServiceNow with bi-directional status sync.
- In-house: a developer builds a service that polls the database, calls the ServiceNow API, handles correlation and writes back status. After testing and security review, the integration goes live after several weeks. Each later change requires the developer.
- Integrator: a partner scopes, builds and hands over a similar service. Timeline is similar, with added contracting time. Changes require new change requests.
- Platform: the platform vendor delivers a database connector; administrators configure mappings and sync rules; the integration goes live within days or a few weeks. Changes are made through configuration.
Multiply this by the number of custom systems expected over three years, and the difference in total cost and time becomes substantial.
Building your own cost model
To compare options for your organization, estimate the following inputs for each candidate system and option. Keep the assumptions visible so stakeholders can challenge and refine them.
- Build effort: person-days for design, development, testing and documentation.
- Day rates: internal loaded cost per developer day, or the integrator's rate.
- Platform costs: subscription share and any connector fee, if applicable.
- Security review: person-days for review and remediation.
- Hosting: infrastructure and monitoring costs per year.
- Maintenance: expected changes per year multiplied by effort per change.
- Incidents: expected failures per year multiplied by time to resolve.
- Opportunity cost: value of the product work developers would otherwise deliver.
- Delay cost: value lost for each month the integration is not live, such as manual work that continues.
Delay cost is often the deciding factor. If manual work costs a team many hours per week, an integration that goes live in days rather than months pays back quickly, even before maintenance savings are counted.
Portfolio thinking: the second and third connector
Decisions are often made one connector at a time. But most organizations need many custom connectors over time, and the economics change as the number grows. The first in-house connector may seem manageable; the fifth means five codebases with different logging, security and deployment approaches. With a platform, each additional connector adds little operational overhead, because monitoring, credentials, mapping and governance are shared.
For this reason, it is worth estimating the expected number of custom connectors over three years before choosing an approach for the first one. Many organizations discover the long tail is longer than they assumed once they inventory internal, legacy and niche systems.
Contract and support questions for platform-delivered connectors
- Is custom connector development included in the subscription, or charged separately?
- Who maintains the connector when the source system is upgraded?
- What support response times apply to custom connectors?
- Does the custom connector receive the same features and updates as standard connectors?
- Can the connector be reused for other instances or business units?
Frequently asked questions
What is a custom connector?
A custom connector is a component that lets an integration tool communicate with a system that is not supported by standard connectors, such as an internal application, legacy system or niche product.
How much does it cost to build a custom connector?
Costs vary with complexity, but the largest share usually comes after go-live, through maintenance, upgrades, incident handling and security reviews. Evaluate total cost over at least three years.
How long does it take to build a custom connector?
In-house or integrator-built connectors typically take weeks to months. Some platforms deliver much faster; ZigiOps typically builds new connectors in about three days.
Is building integrations in-house cheaper?
It may look cheaper initially, but long-term maintenance, knowledge risk and opportunity costs often make it more expensive than a platform approach.
Can a custom connector be built for a system without an API?
Yes. Connectors can use databases, SOAP services or libraries. ZigiOps supports all of these methods.
Who should maintain custom connectors?
Ideally the team that manages other integrations, using the same platform and processes, so custom connectors are not treated as exceptions.
What is delay cost in integration decisions?
Delay cost is the value lost while an integration is not yet live, such as continued manual work, slower incident handling or missed revenue. It often outweighs differences in build cost.
Can a platform-delivered connector be customized later?
Usually, yes. Once the connector exists, mappings, filters and synchronization rules are adjusted through configuration rather than code, so changes do not require a new development project.
What if the platform vendor cannot reach my system?
Ask early which interfaces the vendor supports. If the system offers an API, SOAP, a database or a library, most capable platforms can connect. If it offers only a proprietary protocol, in-house development may still be needed.
Key takeaways
- Custom connector gaps are normal; the way you fill them determines long-term cost and risk.
- Include delay cost and the expected number of future connectors in every decision.
- Evaluate build, commission and platform options over at least three years, including hidden costs.
- In-house builds suit differentiating or highly unusual integrations; integrators suit one-time specialist needs.
- Platform-delivered connectors are usually fastest and easiest to maintain when several systems need integration.
- ZigiOps typically delivers new connectors in about three days, with no-code configuration and no storage of transferred data.
