Best of luck to Metriport. I've worked for years in the healthcare interoperability space and there are still a huge number of unsolved problems impacting cost and patient care.
There actually is a standard for converting C-CDA records to FHIR. It isn't 100% complete but serves as a useful starting point. If you find problems with it you can feed those back into the standards process.
FHIR also includes unstructured narrative text so it isn't necessarily better than C-CDA in that regard. You'll find that data quality problems come down more to provider systems configuration and charting policies rather than data formats.
The biggest pain point isn't technical, it's misalignment of incentives in provider organizations. Interoperability standards have existed for years and are widely implemented in commercial and open-source products and online services. Some of these may be a bit expensive or difficult to use but they work well enough, at least for common use cases.
The problem is that most providers still work on a fee-for-service basis, billing payers (insurers) and patients for individual line items. There's no line item billing code for improving clinical data quality or sharing patient records with other authorized organizations. So they mostly do the bare minimum necessary to comply with government regulations and payer coverage rules.
For example, every doctor is supposed to have a Direct Secure Messaging address listed in NPPES by now so that they can securely email patient records to each other. Every major EHR supports this standard and it can also be used through HISP online portals. But a lot of doctors still have no clue how to do this and haven't registered their address in NPPES (or misunderstood the instructions and put in their own personal Hotmail email address or something). So, they still end up sending faxes.
The situation may eventually improve with the shift to value-based care but this will be a slow process.
There actually is a standard for converting C-CDA records to FHIR. It isn't 100% complete but serves as a useful starting point. If you find problems with it you can feed those back into the standards process.
http://hl7.org/fhir/us/ccda/
Microsoft has an open source library which works pretty well and I think implements at least part of that standard, although I haven't used it lately.
https://github.com/microsoft/FHIR-Converter
FHIR also includes unstructured narrative text so it isn't necessarily better than C-CDA in that regard. You'll find that data quality problems come down more to provider systems configuration and charting policies rather than data formats.