Data Intelligence Monthly

Issue 04 - February 2026

Welcome to the fourth issue of Data Intelligence Monthly. Each month, we will touch-on a specific topic along the data analytics life-cycle. In this issue, we discuss an important stage and approach in data analytics: Data Sourcing.

Data Sourcing as a Core Technique in Data Analytics

Building Reliable Insights from the Ground Up

In data analytics, advanced models and visualization tools often receive the spotlight. Yet even the most sophisticated analytics will fail if the underlying data is incomplete, inconsistent, or poorly understood. Data sourcing is therefore one of the most critical techniques in the analytics lifecycle especially where decisioning is highly impactful, time-sensitive, and financially material.

Effective data sourcing is not a single task completed at the start of a project. It is an iterative process, requiring continuous refinement as understanding of the business problem deepens and analytical findings surface new requirements.

Data Determination and Requirements

Data sourcing begins not with extraction, but with determination identifying what data is required to answer the business problem. This includes defining metrics, dimensions, historical depth, granularity, frequency, and acceptable thresholds for accuracy and completeness. At this stage, analytics teams translate business objectives into technical requirements. Importantly, these requirements often evolve as feasibility constraints and data limitations become visible.

  • Example: A bank seeks to analyze mortgage portfolio profitability. Initial requirements include loan balances, interest rates, payment schedules, and credit risk ratings. However, early exploration reveals the need for additional data such as prepayment behavior, funding costs, or customer segmentation attributes.

  • Takeaway: As insights emerge, requirements are refined—demonstrating the first iteration of the sourcing process. Without clear data determination, teams risk over-collecting irrelevant data or missing critical variables that materially affect results.

Data Collection

Once requirements are defined, attention turns to data collection—locating and accessing the systems, platforms, and external sources that contain the required data. In larger organizations, data often resides across multiple systems, platforms, engines, warehouses, and data-marts. As such, collection methods may include database queries, APIs, flat-file extracts, pipeline development, and external vendor feeds.

Iteration is common at this stage. Initial extracts frequently expose gaps in coverage, misaligned time periods, or inconsistent identifiers that require returning to the determination phase.

  • Example: An investment firm analyzing portfolio performance may initially pull data from its portfolio accounting system. During analysis, the team discovers that benchmark returns are sourced separately from a market data vendor and use different pricing calendars.

  • Takeaway: This insight requires re-collecting data using aligned valuation dates and potentially supplementing internal data with external sources—an iterative refinement driven by early analytical findings.

Assessing Data Quality

Not all data is equally suitable for analytics. Data quality assessment evaluates whether sourced data is fit for its intended purpose. Common dimensions include completeness, accuracy, consistency, timeliness, uniqueness. Data quality issues are not merely technical inconveniences—but a challenge that must be solved. Assessment often reveals issues that require returning to collection or even redefining requirements entirely.

  • Example: A credit risk team is required to source borrower income data to support probability-of-default modeling. During quality assessment, analysts discover incomplete, invalid and missing data elements, inconsistent formatting, and out-of-date self-reported values.

  • Takeaway: These findings may trigger several iterations, including sourcing alternative variables, limiting analysis to reliable time periods, or incorporating proxy indicators such as transaction inflows. Quality assessment transforms assumptions about data into evidence-based understanding.

Data Validation

While data quality focuses on structural integrity, data validation ensures that sourced data accurately reflects real-world behavior and aligns with trusted benchmarks. Validation techniques include reconciliation to applicable reports and independent systems, comparison to prior periods, and trend and variance analysis. Validation is rarely a one-time exercise. As models evolve and new data elements are introduced, validation must be repeated to preserve analytical credibility.

  • Example: An asset manager validating daily net asset value (NAV) data may reconcile portfolio valuations against custodian records. Unexpected discrepancies may reveal timing differences, stale prices, or missing corporate actions.

  • Takeaway: Resolving these issues often requires revisiting and reiterating on data collection logic and/or adjusting transformation rules. Validated data builds confidence among stakeholders and ensures that analytics outputs can withstand scrutiny.

Bringing It All Together

Analytics success is inseparable from data sourcing discipline. Strong analytical outcomes depend not only on advanced tools, but on the rigor applied long before dashboards or models are built. Effective data sourcing practices require:

  • Clearly defining data needs aligned to business objectives

  • Identifying and integrating data across multiple systems and platforms

  • Assessing quality to ensure fitness for analytical use

  • Validating results to maintain trust and compliance

  • Embracing iteration across all steps as understanding deepens

When executed thoughtfully, sourcing data becomes more than a preparatory step—it becomes a strategic capability that strengthens decision-making, reduces risk, and enables analytics to deliver meaningful and defensible insights.

In the end, high-quality analytics does not begin with algorithms—it begins with asking the right questions of the right data, sourced the right way.

If you are interested in discussing, planning or developing your data analytics strategy, please contact us for a free 30-minute consultation.