Background
A North American logistics company offering transportation services and with more than 75 distribution sites called on Keyrus for a problem related to process automation. Specifically, the customer had a legacy rates setup process involving lots of manual intervention. The process included manual data validation of exported financial reports, downloading inflation figures from government websites, and running Excel macros. Most importantly, the process also included a manual review of thousands of inconsistently structured legal documents (i.e. contracts) ranging from 40 lines to over 60 pages! As a result, defining the right rate for each customer was often missed resulting in inaccurate customer billing and lost revenue. To avoid this revenue loss, Keyrus was asked to analyze and recommend the right solution and technology mix (Snowflake was already selected for data storage) to meet its automation and unstructured data processing needs.
Approach
1. Discovery Phase:
Keyrus proposed to start the engagement using a discovery and solution high-level design phase. This allowed both the customer and Keyrus to confidently assess the options at hand and weigh in the solution cost and ROI prior to committing to a larger project.
During the discovery phase, Keyrus analyzed the solution requirements and evaluated 4 different tools/options (Alteryx, Leverton, Python, and DocuSign API) for the legal documents parsing portion of the process. Ultimately, Keyrus recommended Alteryx due to its balance of functionalities, cost, ease of use, and reusability. In addition, Keyrus also recommended to leverage Rivery for integrating data from the government websites API and internal ERP database into Snowflake. Finally, Keyrus provided the customer with the refined solution implementation plan including timeline, risks and costs.
2. Implementation:
Throughout the implementation phase, Keyrus provided the following deliverables:
Redesigned the end-to-end customer rates setup process
Developed an Alteryx workflow to parse the legal contracts with great accuracy
Processed the initial batch of legacy contracts.
Snowflake account created and configured with appropriate database schemas and permissions
Rivery account created and configured with appropriate logins and scheduled data uploads.
Documentation and knowledge transfer sessions to enable the client team to take ownership of Snowflake, Rivery and Alteryx tools and establish future data pipelines.
Challenge
Variability & Complexity:
Automating a historically manual process that involves reading thousands of contracts ranging from a few lines of text to over 60 pages in multiple languages, containing virtually no consistency between them, and having only a few weeks to do so... can seem like a near impossible task.
Fortunately, after much discussion and multiple iterations, Keyrus experts began to find consistencies that allowed them to pinpoint relevant data points and analyze them with a high degree (~85%) of accuracy.
Benefits
Three new tools were configured for current use cases and future developments:
Alteryx
Rivery
Snowflake
The entire process was redesigned and configured to work for all existing contracts, resulting in the following:
80-120% increase in revenue from annual rates adjustments, which were not previously accounted for.
7,655 legacy contracts analyzed with an 85% match rate (target was >50%) for key fields.
Automated monthly extraction of inflation figures from government website APIs using Rivery.
Improved ability to change pricing logic and generate reports on a schedule with Alteryx and Rivery.
Extensive knowledge transfer was performed with the client's future development plans in mind, including requirements for a Phase 2 project.