What is duplicate data?
Duplicate data refers to duplicate transactions, accounts, invoices, etc. in our database. In other words, when two transactions are returned via GET /transactions for only one transaction that exists in the financial institution.
What causes duplicate data?
Each time a single link is updated via POST or a recurrent link is updated automatically, the freshly collected data is compared to the existing data in our database and our database is updated with the new data collected, that is new transactions, new accounts, new incomes, etc.
At Belvo we have sophisticated methods to identify if the data we retrieve is in fact new data or data that we already have stored in our database. However, changes to the data by institutions or collecting data from different sources at the same time can make it difficult for us to distinguish new data from existing data in certain cases.
Accounts
Change in an account unique identifier _ If an institution changes the name, account_number, or type of an account it is more difficult to confirm if it is the same account and thus increases the probability of being duplicated in our accounts database.
POST /accounts request to a recurrent link _ Transactions can be duplicated if a recurrent link's transactions are being updated automatically and at the same time a POST /transactions request is manually sent to the same link. For this reason, we only suggest sending POST requests to populate single links.
Transactions
Change in a transaction unique identifier _ If an institution changes the amount, description, or value_date of a transaction after it was already collected by our API, it is more difficult to confirm if it is the same transaction and thus increases the probability of being duplicated in our transaction database.
POST /transactions request to a recurrent link _ Accounts can be duplicated if a recurrent link's accounts are being refreshed automatically and at the same time a POST /accounts request is manually sent to the same link. For this reason, we only suggest sending POST requests to populate single links.