Skip to content
All field notes
Comparisons13 August 20266 min read

Build or buy a document parser: the real costs

When choosing between building a document parser and buying a managed solution, building in-house is more expensive and riskier than it appears on paper for most operational teams. The initial build costs are still manageable, but it is the maintenance required when document formats change that truly breaks the model. Below are the concrete trade-offs, including the situations where custom development is the right choice.

By Yeslin Beljaars

What does building a document parser actually cost?

A developer building a parser on top of an open-source toolkit or cloud API (Google Document AI, AWS Textract, Azure AI Document Intelligence) can have something working within a few weeks. The problem is not that first version, but everything that follows. Document formats are not stable: suppliers update their PDF layouts, a new customer sends a poor-quality scan, a bill of lading suddenly includes an extra line block. Every variation requires adjustments to the extraction rules. Without a team to handle those adjustments, accuracy degrades slowly but steadily. That maintenance is structural, not a one-time effort, and it is rarely factored into most business cases.

Subscription costs versus build costs: how does the comparison stack up?

A managed solution charges per document or per month. That feels like an ongoing cost, while building sounds like a one-time investment. But the calculation shifts quickly when you are honest about the full cost of building: initial development work, integration with your ERP or TMS, testing edge cases, and then ongoing maintenance for every document variation. With managed solutions like dottle, integrations with systems such as Exact, AFAS, SAP, or Dynamics are pre-built, scalability is handled, and time-to-value is typically two weeks rather than months. That difference in time also carries a price: every month your team is still manually retyping data is capacity that is unavailable elsewhere.

Document variations are the biggest maintenance problem

This is where most in-house build projects fall apart. A parser that works well on twenty known formats from twenty fixed suppliers starts to break down the moment a format appears that is structured slightly differently. Rule-based extraction is inherently fragile when layouts vary: you can keep extending the rules, but the pile of exceptions grows along with them. Managed solutions do not solve this by magic, but they place the maintenance responsibility with the provider, not with your IT department. New formats are added in a controlled way without requiring you to schedule a sprint for it.

When does building in-house make more sense?

There are situations where custom development is the more logical choice. First: when you face strict data-residency requirements or air-gapped environments where documents cannot leave your own infrastructure. Second: when your document formats are so unique and stable that a generic approach consistently falls short. Third: when your volume is high enough and document structure predictable enough that the management costs of a custom solution are lower over time than a per-document rate. In practice, that last scenario is only realistic at very high volumes with a dedicated technical team actively maintaining the parser. For most operational teams in transport, food, or wholesale distribution, that does not apply: their IT department has other priorities and document formats vary more than expected.

What are the honest limitations of a managed solution?

A managed document parser is not a silver bullet. If your source documents are structurally poor quality (blurry, handwritten, inconsistent), no system will handle them fully. If your process is not well-defined, meaning it is unclear which data you need and for what purpose, then automation is premature regardless of the tool. And if your volume is low (a few dozen documents per month), the investment will not pay off. dottle is straightforward about that boundary: when volume is too low or a process is not yet mature, we advise waiting. The value lies in repetitive, predictable, high-volume document work.

Conclusion: make-or-buy for document parsing

For most operational teams, building your own document parser costs more than it appears and delivers less than expected. The build costs are visible, but the maintenance burden, the delay in time-to-value, and the internal capacity tied up in the effort are far less so. A managed solution like dottle is live within two weeks, connects to the systems you already use, and keeps the responsibility for handling format variations with the provider. Custom development is the better choice when there are specific privacy requirements, stable unique formats, or extremely high volumes supported by a technical team that can sustain it structurally. In all other cases, it is a more costly path than necessary.

Seeing this in your own document flow?

Book a demo on your own documents

Frequently asked questions

What does it cost to build your own document parser?

Initial build costs vary widely but are generally still estimable. Maintenance is not: every new document variation or updated layout requires changes to the extraction rules. That ongoing work is rarely included in most business cases, and it is structural, not a one-time effort.

What is the difference between building a document parser and using a managed solution?

When you build in-house, you are responsible for development, integration, testing, and maintenance. With a managed solution, you use a ready-made service with pre-built integrations and a provider that handles changing formats. The biggest practical difference is time-to-value: weeks versus months, and who carries the maintenance responsibility.

When is custom development the better choice?

When there are strict data-residency requirements (air-gapped environments), when document formats are so unique and stable that a generic approach consistently falls short, or when volumes are extremely high with a dedicated technical team in place. For most operational teams in transport, food, or wholesale distribution, these conditions do not apply.

How quickly can a managed document parser go live compared to building your own?

A managed solution like dottle is typically operational within two weeks, including the connection to your ERP or TMS. An in-house build project takes several months in practice, especially when edge cases, integrations, and an approval workflow are part of the scope.