How HippoScripts Used Medallion Architecture in Microsoft Fabric to Help a Leading Audit Firm Scale Its Data Platform

Audit firms depend on trusted data. Whether the use case is financial reporting, internal analytics, audit quality monitoring, client portfolio analysis or operational performance, the underlying data platform needs to be reliable, governed and easy to scale.

 

As the volume of data grows, traditional reporting approaches often begin to struggle. Data is copied across different systems, definitions become inconsistent, reporting logic gets repeated, and teams spend too much time reconciling numbers instead of using them.

 

HippoScripts worked with a leading audit firm to support the design and implementation of a modern data platform using Microsoft Fabric and Medallion Architecture. The objective was to create a scalable foundation where data could move from raw ingestion to trusted business outputs in a structured and controlled way.

The project was not only about building pipelines. It was about helping the organisation think more clearly about data architecture, governance, cost control and long-term scalability.

 

 

 

 

The challenge

 

The audit firm needed a data platform that could support current reporting needs while preparing the organisation for future analytics and AI use cases.

Before introducing a layered architecture, several common challenges had to be addressed:

 

• Data was coming from multiple operational and business systems
• Reporting logic risked being repeated across different outputs
• Business users needed clearer access to trusted datasets
• Technical teams needed a more maintainable structure for pipelines and transformations
• The organisation wanted to avoid over-engineering the platform too early
• Microsoft Fabric costs and capacity usage had to be managed carefully from the start

This is a familiar pattern for growing organisations. The problem is rarely that data does not exist. The harder problem is making sure the data is organised, governed and reusable enough to support better decisions.

 

Why Microsoft Fabric was selected

 

Microsoft Fabric provided a strong fit because it brings together data engineering, lakehouse storage, data warehousing, semantic modelling and Power BI within one analytics platform.

 

For an audit firm already operating within the Microsoft ecosystem, Fabric offered a practical way to modernise the data platform without introducing unnecessary fragmentation across too many tools.

 

The key advantages included:

• OneLake as a unified data foundation
• Lakehouse capabilities for structured and semi-structured data
• Delta table support for reliability and performance
• Integration with Power BI and semantic models
• A scalable capacity model that could grow with usage
• A single environment for engineering, analytics and governance

 

The important decision was not simply to “use Fabric”. The real value came from designing Fabric properly around the organisation’s operating model.

 

Why Medallion Architecture was used

 

HippoScripts recommended a Medallion Architecture approach because it gives data teams a clear structure for organising data by quality and purpose.

The architecture was designed around three main layers:

 

Bronze: preserving the raw truth

 

The Bronze layer was used as the landing area for source data. The goal of this layer was to preserve data as close as possible to its original form.

This provided a recoverable foundation. If transformation logic changed later, the team could return to the original data instead of depending on source systems to resend it.

The Bronze layer also helped create a clearer audit trail, which is especially important in professional services and audit environments where traceability matters.

 

Silver: creating clean and reusable data

 

The Silver layer was used to clean, standardise and enrich the raw data.

This included work such as:

• Standardising naming conventions
• Aligning data types
• Handling duplicates
• Applying basic quality rules
• Creating reusable datasets
• Preparing data for downstream reporting and analysis

The purpose of Silver was to avoid rebuilding the same cleaning logic again and again. Once data was cleaned and structured properly, multiple reports and analytical outputs could reuse the same foundation.

 

Gold: preparing trusted business outputs

 

The Gold layer was designed for business consumption.

This is where data was shaped around reporting and analytical use cases. Instead of asking business users to interpret raw source tables, Gold datasets provided curated outputs that could support dashboards, KPIs and decision-making.

Examples of Gold-level outputs included:

• Management reporting datasets
• Operational performance views
• Finance and billing-related analysis
• Audit portfolio reporting
• Business development and client analysis
• Power BI-ready semantic models

The goal was to make trusted data easier to find, easier to understand and easier to use.

 

Designing for scale without over-engineering

 

A major principle of the project was to build a platform that could scale without becoming unnecessarily complex on day one. This meant taking a pragmatic approach to architecture.

Not every dataset needed a complicated pipeline. Not every use case needed every layer immediately. Not every table needed to be modelled before the business had validated the need.

 

HippoScripts helped the firm design a structure that could grow gradually. The platform could support early reporting needs while keeping enough discipline to add more sources, more use cases and more users over time.

This balance matters. A data platform that is too simple becomes difficult to control later. A platform that is too complex too early becomes expensive and hard to adopt.

 

Cost and billing optimisation in Microsoft Fabric

 

One of the most important parts of the engagement was helping the organisation think about Microsoft Fabric costs from the beginning.

Fabric capacity can provide strong value, but it needs to be managed carefully. Without proper monitoring and workload planning, organisations can either overpay for capacity they do not need or under-provision the platform and create performance issues.

 

HippoScripts supported a cost-conscious Fabric approach by focusing on:

• Starting with an appropriate capacity size rather than overcommitting too early
• Reviewing workload patterns before scaling
• Separating development, testing and production considerations
• Monitoring utilisation through Fabric capacity metrics
• Understanding which workloads were interactive and which were background workloads
• Designing pipelines to avoid unnecessary refreshes and repeated processing
• Keeping Gold outputs focused and purposeful
• Avoiding uncontrolled duplication of datasets
• Reviewing when capacity should be scaled, paused or resumed depending on usage patterns

 

This helped the organisation treat Fabric not only as a technical platform, but also as an operating cost that needed governance.

 

Avoiding unnecessary data duplication

 

Another important design decision was to avoid copying data unnecessarily.

With OneLake and Microsoft Fabric, organisations can design a more unified data estate where the same data foundation can support different workloads. The architecture was designed to reduce duplicated storage and repeated transformations where possible. This supported both technical efficiency and cost control.

Instead of creating separate pipelines for every report, the platform encouraged reusable Silver datasets and curated Gold outputs. This meant that new analytical use cases could be added more quickly because the foundations were already in place.

 

Governance and ownership

 

For an audit firm, governance is not optional. Data needs to be trusted, traceable and controlled.

The architecture therefore considered:

• Clear naming standards
• Layer-level ownership
• Controlled access to different data layers
• Separation between raw, cleaned and curated data
• Documentation of transformation logic
• Clearer business definitions for key metrics
• A structure that could support future data quality and lineage requirements

 

This helped reduce ambiguity. Engineers could understand where transformations belonged. Analysts could understand which datasets were ready to use. Business users could have more confidence in the outputs.

 

The outcome

 

The engagement helped the audit firm move towards a more scalable and governed data platform using Microsoft Fabric.

The key outcomes included:

• A clearer data architecture based on Bronze, Silver and Gold layers
• A more structured Microsoft Fabric environment
• Better separation between raw ingestion, data cleaning and business reporting
• Improved reuse of cleaned datasets
• Stronger foundations for Power BI and future analytics
• Better cost awareness around Fabric capacity and billing
• A platform design that could scale with more data sources and users over time

 

The result was not just a technical implementation. It was a stronger data operating model.

The organisation gained a clearer path for building trusted analytics while keeping the platform understandable, maintainable and cost-conscious.

 

Lessons learned

 

Several lessons from this project apply to many organisations considering Microsoft Fabric and Medallion Architecture.

First, architecture should follow the business problem. Bronze, Silver and Gold are useful layers, but they should not be applied mechanically. The design needs to reflect how the organisation actually uses data.

Second, cost optimisation should start early. Waiting until the platform becomes expensive makes optimisation harder. Capacity planning, monitoring and workload discipline should be part of the design from the beginning.

Third, governance should be practical. The aim is not to slow teams down. The aim is to make data easier to trust and reuse.

Finally, Medallion Architecture works best when it becomes a shared language between technical and business teams. It helps everyone understand where data comes from, how it is improved and which outputs are ready for decision-making.

 

Final thoughts

 

Microsoft Fabric gives organisations a powerful platform for modern data engineering, analytics and business intelligence. However, the value of Fabric depends heavily on how it is designed.

By applying Medallion Architecture, HippoScripts helped a leading audit firm build a scalable data foundation that could support reporting today and future analytics tomorrow.

The project showed that strong data platforms do not need to be overcomplicated. They need to be structured, governed, cost-aware and aligned with how the business makes decisions.

 

HippoScripts is a Microsoft-certified data engineering company headquartered in Malta, working with an international customer base to help organisations design reliable data platforms, improve operational efficiency and make better use of their data.

If your organisation is considering Microsoft Fabric, Medallion Architecture or a modern data platform, book a discovery workshop with HippoScripts to explore the right approach for your business.

 

Book a discovery workshop with HippoScripts: Click here