Strategic Planning and Data Architecture for TMS Implementation

Business Case and Current Landscape for TMS Implementation

  • Project Context: The discussion serves as a follow-up to a previous presentation regarding the Treasury Management System (TMS) implementation for MAPFRE. The focus is on transitioning from the current priority landscape to a future-state architecture using instructions and feedback provided by the Bank of America team (specifically Mike and Abhishek).

  • Current State Assessment: MAPFRE is currently defining and scoping detailed processes for inclusion in their Request for Information (RFI) document. While the document is not yet 100%100\% complete, key sections regarding the company profile, value landscape, and pain points have been established.

  • Primary Operational Pain Points:

    • Heavy reliance on manual treasury workflows.
    • Daily balancing and reconciliations conducted via spreadsheets.
    • Manual data uploads into SAP.
    • Limited centralized visibility into global cash positions.

Prioritizing Requirements in the RFI

  • Categorization Framework: The RFI includes three distinct categories for requirements:

    1. Nice to have: Desirable features that are not critical for immediate go-live.
    2. Desirable: Features that provide significant value but may be secondary to core fixes.
    3. Postponed/Later: Requirements that can be deferred to subsequent implementation phases.
  • Strategic Objective: The goal is to structure these requirements specifically enough to avoid receiving generic, "templated" responses from vendors, while still providing enough long-term vision so vendors understand the future scope.

Foundations of Data Strategy and Integrity

  • The Primacy of Data: Mike suggests moving away from the rigid "agenda" of categorizing must-haves versus nice-to-haves and focusing instead on data strategy. The core problem for clients implementing a TMS is disorganized and inaccessible data across multiple source systems.

  • Data Strategy as Foundation: A data strategy must precede both selection and implementation. It is a shared responsibility between MAPFRE and the chosen provider.

  • Core Data Elements to Capture:

    • Account Structure: How bank accounts are categorized and linked.
    • Legal Entity Structure: The hierarchy of company entities.
    • Subdivisions: Specific units within legal entities that may maintain independent bank accounts.
    • Data Repository: A centralized location for all data elements to facilitate reporting and analytics.
  • Output Design: The system design should be driven by how MAPFRE wants to view its cash (e.g., entity level, bank account level, or cross-system views). Vendors must be able to translate MAPFRE's raw data into "system speak" to facilitate efficient cash management.

Data Architecture: Direct Integration vs. Data Lake

  • Direct Integration: In this model, core source systems (such as SAP) interact directly with the TMS. The TMS uses an ingestion engine to map and translate incoming data fields into its own language.

  • Data Lake/Warehouse Approach: This involves building a repository to scrub, normalize, and enrich data before it reaches the TMS.

    • Normalization: Standardizing naming conventions and terms across different source systems (e.g., deduplicating customer or supplier info).
    • The "Lake" vs. "Swamp": A data lake is effective only if it follows a strict data strategy to prevent it from becoming a "data swamp."
    • Identification of Key Attributes: Mike uses a hypothetical example of identifying 2525 core data attributes critical to the treasury discipline (e.g., cash management, capital structure).
  • Cost and Performance Trade-offs:

    • Data Lake Pros: Better for large-scale initiatives involving AI, Business Intelligence (BI), and cross-departmental reporting (FP\&A, Tax, Accounting).
    • Data Lake Cons: Significant increase in implementation cost and a longer timeline for the TMS to go live.
    • Recommendation: If the project is strictly a treasury-focused effort, direct integration may suffice. If it is part of a broader organizational data initiative, the lake should be prioritized first.

Strategic RFI and Demo Methodology

  • RFI Limitations: Most vendors will respond "Yes" to every requirement in an RFI just to secure a seat at the table.

  • The Interview Approach: The RFI should be treated as an interview of the vendor's people and company culture rather than absolute proof of the system's capabilities.

  • The Demo as Proof: Micro-level design elements and specific manual processes should be evaluated during live demonstrations rather than on paper.

  • Demo Scripting recommendations:

    1. Provide the vendor with real MAPFRE data in advance.
    2. Allocate a 33 to 44 hour block for a "day in the life" session.
    3. Force the vendor to show how they handle specific MAPFRE workflows (the "lines and arrows" in the process flow) rather than their standard sales pitch.

Vendor Landscape and Selection

  • Recommended Market Players: Mike identifies three main vendors suitable for MAPFRE's complexity:

    1. Krave with FIS (Likely Kyriba).
    2. CG Treasury (GTreasury).
    3. Ripple (Likely Reval).
  • The SAP TRM Alternative: While MAPFRE is an SAP shop and must consider SAP TRM (Treasury and Risk Management) for internal compliance/preference reasons, it is noted that SAP TRM is generally perceived as:

    • Lesser in specialized treasury functionality compared to core TMS providers.
    • More expensive to implement.
    • Requires a longer implementation timeframe.

Questions & Discussion

  • Khaled: Asked if they should present the full scope of their "homework" (micro-level process mapping) in the initial RFI or save it for shortlisting.

    • Answer (Mike): Explicitly recommended saving the micro-level details for the demo phase. It saves time during the RFI evaluation and allows for live "pressure testing" of the system.
  • Khaled: Asked about the best way to handle data from systems that are not API-compatible.

    • Answer (Mike): Mentioned that most systems can produce a CSV or XML file for batch processing. The TMS's ingestion engine is responsible for translating these different formats (CSV, XML, etc.) into the internal treasury language.
  • MAPFRE Team: Confirmed they are currently working on replacing PDF workflows with digital data to ensure the data is relevant for the TMS ingestion.