The client’s challenge
Most documents follow the ideal processing flow for archiving, identifying documents and clients through direct comparison of relevant indicators. The challenge arises with the remaining volume, where Banca Transilvania encountered two distinct types of situations:
Missing, incorrect, or uncertain identifiers
In some cases, one or more identifiers associated with a document were missing, incorrect, or ambiguous. Following the acquisition of other banking institutions, their documents were absorbed into the system, documents created using different patterns that nonetheless had to be incorporated into the current archiving flow. As a result, establishing hierarchical associations between documents, as well as linking documents to clients based on exact identifier matching, became impossible to do with certainty.
Desynchronization in the archiving flow
Desynchronizations could occur across the preparation, dispatch, transport, and processing stages of the archiving workflow, including situations where certain documents arrived at the archive before appearing in the bank’s internal registers. All of these cases needed to be identified quickly and accurately to enable the automatic normalization of the largest possible share of unarchived documents.
Essensys Approach
Building identity sets for clients
We mapped out each client’s individual data variations, identifying and addressing every possible scenario, so a single client could be recognized across multiple identities, exactly as they appeared in the bank’s records or in the archive.
Building identity sets for documents
We applied the same approach to documents, addressing cases of incorrect formatting and erroneous date entries.
Matching documents to clients
Data from the bank’s internal registers was compared against records from physically archived documents using fuzzy matching algorithms, with configurable scores, weights and thresholds, producing a ranked list of potential matches.
Key features
Automated processing algorithms eliminate the need for manual handling of source data.
Identity sets are built per client, accounting for real-world data variations: reversed first and last names, missing or present hyphens, missing first names, name changes, missing or incorrect ID numbers, inconsistent diacritics, extra or missing spaces, and inconsistent punctuation (e.g. “SRL” vs “S.R.L.”).
A client can be identified through any of their associated identities, as they appear in the bank’s records or in the archive.
A set of documents is always processed within the context of a client’s primary identity, even when that client has multiple associated identities.
Two-step associations enable more precise client identification.
Prevents documents from being incorrectly matched to the wrong clients when processing uncertain or ambiguous data.
Identity sets for documents are built based on document number and date, handling formatting errors and incorrect calendar dates.
Compares data from the bank’s internal registers against records from physically archived documents, generating a ranked list of potential matches ordered by descending match score.
Identifies both documents that arrived at the archive with incorrect data, and documents that have already been archived but have not yet been matched in the internal registers due to uncertain data.
Project impact
- 90% of documents outside the ideal flow reconciled automatically
- Significant reduction in the need for manual review and identification,
- Fast, scalable results – batch and incremental processing for large data volumes, with no routine manual intervention required.
Testimonial
Essensys demonstrated a strong understanding of the data complexity and the specific challenges involved in our archiving process, as well as a high level of technical expertise throughout the entire project. The solution they developed automated the normalization of data for documents that could not be matched through exact matching, significantly reducing the need for manual verification and enabling the efficient processing of large data volumes, all while fully meeting our security and governance requirements. Essensys proved to be a reliable partner, capable of delivering a solution tailored to a project of considerable complexity.
Deputy Director, Physical and Electronic Document Management,
Banca Transilvania
Conclusion
A bank built through the integration of multiple financial institutions inherits more than clients, it inherits years of documents with inconsistent identifiers: names spelled differently, missing ID numbers and data patterns that vary from one acquisition to the next.
For Banca Transilvania, this reality turned a legally required process, the physical archiving of documents, into a costly manual effort.
Through identity resolution and fuzzy matching algorithms, Essensys automated the data normalization process that exact matching alone couldn’t solve, enabling the automatic reconciliation of over 90% of documents outside the ideal archiving flow.