Charts of Account Mapping

Case Studies

The Situation

As part of a Realpage Pilot, a global investment firm requested Lionpoint to support the mapping of a few dozen charts of accounts.

A global investment firm required assistance mapping a 15,000-row, non-structured Chart of Accounts spreadsheet to 300 accounts. To streamline the process and avoid a lot of manual effort, Lionpoint’s on-site team sought support from Lionpoint’s data and AI team to automate the effort. 


The Approach

Lionpoint used natural language processing techniques to automate a highly manual mapping exercise. 

The initiative’s primary goal was to create an automated solution capable of suggesting corresponding account names and codes based on a given account description. To achieve this objective, Lionpoint’s data team utilized natural language processing (NLP) techniques. By leveraging NLP techniques and building on components of Lionpoint’s LITTIL framework, the team was able to generate precise matches rapidly.  

Overall, the model mapped 91% of the accounts to account codes based on assessing text similarity between account names. However, the data contained many similar account names and account names with minimal similarity. Also, inconsistencies within account codes also affected accuracy. 

To evaluate the model’s accuracy, the data team assessed its performance by comparing the top 5 suggestions it provided with a subset of manually mapped accounts, which accounted for 8% of the total data. The assessment revealed that the model achieved a 61% accuracy in correctly determining the account codes.  


The Impact

An estimated time saving of approximately 25% was gained compared to manual mapping. 

The initiative improved efficiency and saved time. While the model’s accuracy could be further refined, the initiative demonstrated the potential to enhance financial reporting accuracy and streamline financial processes. The data team plans to continue refining the solution and expanding its capabilities to improve accuracy and efficiency in CoA management further.