regarding data quality, i've run into a few issues using Plaid for pulling my own financial data. I've had instances where the API returns duplicate charges that I confirmed were not duplicate on my bank/credit card accounts. I've also had issues with their unique id's too. When a pending charge is returned by the API they give it a unique id, but when the charge is approved they generate a new unique id for the approved charge rather than updating the existing id. They provide a field that lists the pending charge id for you to match up but i've had instances where those id's don't match up. I also run into issues where the access token for some of the institutions i have connected expire or suddenly don't work anymore so I need to re-authenticate.
I've emailed them about these issues and they've been pretty helpful in getting them resolved, but it made me reconsider building a consumer facing product on their API. I still use the API for personal use however
Totally understand on the duplicate issue - its a hard problem to solve with the inconstancies from the end financial institution, and it definitely caused issues for us in the past. That being said - we've spent the past couple months making some huge improvements - I definitely encourage you to give it another shot.
I've emailed them about these issues and they've been pretty helpful in getting them resolved, but it made me reconsider building a consumer facing product on their API. I still use the API for personal use however