Even Without Building E-Signatures Myself, the Hard Parts Remained
On choosing Modusign integration at 21N: what an external service removes, and the state and failure boundaries a product still has to own.
Joohwan Yu · Founder, Full Stack Developer
Toris · Published
Contents
For electronic contracts at 21N (21앤), we integrated Modusign (모두싸인) instead of building e-signatures ourselves. Under the conditions of handling the app, the API, the admin screens, and the infrastructure together, building document signing myself would have greatly expanded the maintenance scope. I drew the boundary on the side of using a service that already provides the needed capability.
What You Can Hand Off and What Remains
Using a signing service does not automatically sort out our product’s contract state. Which document corresponds to which contract, how far the user has gotten, and how to reflect external results on our screens are still the product’s responsibility.
In the public case study at the time, I recorded the Modusign integration and its connection to contracts, points, and settlement. Looking back now, the key was not to confuse a successful external API call with the feature being done. The moment a document creation request is accepted is different from the moment the user finishes signing. The wording shown to the user also has to reflect that difference.
If I Reviewed It Again, I’d Break Down the Failures
I would separate the case where it failed before the request, the case where it was processed on the external system but we never got the response, and the case where we confirmed the result but the internal update failed. If you attach the same retry button to all three situations, duplicate requests can result. Before requesting again, you first have to look at whether you can check the existing external document or its processing state.
The recovery path is a part I’d want to design more explicitly now. The link between the external request identifier and the internal contract identifier has to remain so a person can trace the situation. Rather than just recording “integration error,” I would record which request for which contract got how far.
The Value of Judging to Cut Back on Building It Yourself
There was no need to see the choice not to build it ourselves as the technically less challenging path. The real task was connecting the service’s core flow all the way through. Comparing the time to build a signing tool with the time to show contract state accurately, the latter was more directly valuable to this product’s users.
The lesson that stayed with me is that an external service doesn’t eliminate complexity. It moves part of the complexity outside the contracted API boundary. Understanding the state and failures at that boundary and translating them into our product’s language remained the developer’s job.
Related post: 21N’s External Integrations and Operational Judgment
Advertisement