A3 M9 Audit Data Analytics

Definition and Purpose of ADAs

Techniques that enable auditors to analyze financial and nonfinancial data to discover patterns, relationships, and anomalies, changing how audits are performed due to increased data availability and auditing technology.

Applicability of ADAs

Can be applied across all areas of an audit—risk assessment, tests of controls, substantive procedures, and overall conclusions—and a single ADA can concurrently provide evidence for multiple areas if objectives are met.

Benefits of ADAs

Enhance audit quality by providing better client understanding, advanced risk assessment, expanded coverage (up to entire populations), metadata insights, increased efficiency, enhanced fraud detection, and improved visual reporting.

Steps in Applying ADAs

1. Plan the ADA (objective, data needed, techniques, tools)

2. Access and obtain data (sourcing and ETL cleaning)

3. Review relevance and reliability of data

4. Perform the ADA

5. Evaluate outcomes and address whether objectives were achieved.

ADA Tools and Technology Types

• Data Extraction & Preparation (ETL)

• Data Modeling

• Data Visualization

Data Extraction & Preparation (ETL):

Spreadsheet tools, SQL explorer, data transformation/cleaning software, and Robotics Process Automation (RPA).

Data Modeling

Analytics software/plug-ins, data mining software, programming scripts, and cloud-based services (e.g., Benford's law, sampling).

Data Visualization

Charts, graphs, visualization software, and Natural Language Processing (NLP) tools.

Four Broad Categories of ADA Techniques

1. Descriptive Analytics

2. Diagnostic Analytics

3. Predictive Analytics

4. Prescriptive Analytics

Descriptive Analytics

Explains what happened or is happening (e.g., summary statistics, mean/median, standard deviation, sorting/filtering, aging).

Diagnostic Analytics

Explains why something happened by uncovering correlations and causes (e.g., clustering, drill-down/drill-through, variance analysis, period-over-period, sequence checks).

Predictive Analytics

Uses historical data to predict future events (e.g., regression analysis, forecasting, classification, sentiment analysis).

Prescriptive Analytics

Prescribes optimal courses of action to achieve desired outcomes (e.g., what-if analysis, machine learning, NLP, decision support).

ADA Applications by Audit Area

• Risk Assessment

• Test of Controls

• Substantive Procedures

• Concluding the Audit

Risk Assessment

Identifies previously unknown risks, assesses financial statement/assertion/fraud risks, and helps plan further procedures.

Test of Controls

Evaluates external data, analyzes internal data to verify control effectiveness, detects control failure anomalies, and assists in reperformance (e.g., fuzzy logic matching).

Substantive Procedures

Used in tests of details (sequence checks, full population testing, structure/content analyses) and substantive analytical procedures (regression, trend, ratio, period-over-period).

Concluding the Audit

Applies analytics with advanced entity knowledge to confirm no material misstatements were missed and evaluate updated financial figures.

Data Sources & Storage Functions

Sourced from information systems (AIS, ERP, CRM), internal logs/ledgers, and external sources. Stored in repositories ranging from Data Lakes (structured/unstructured), Data Warehouses, Data Marts, Data Cubes, Databases, Tables, to Spreadsheets.

Data File Formats

Includes tab-separated (.txt), comma-separated (.csv), Excel (.xlsx), Access database (.accdb), Extensible Markup Language (.xml), hypercube (.hyper), and compressed (.zip) files.

Structured Data

Organized in relational databases using tables, attributes (columns), records (rows), and fields. Uses primary keys (unique identifiers), foreign keys, and composite keys. Reduced in redundancy via normalization.

Unstructured Data

Unorganized and inconsistent original formats (e.g., social media posts, sensor data, video/images, transcripts) typically stored in data lakes.

Data Attributes

an be quantitative/numeric (discrete whole numbers, continuous decimals, interval scale, ratio scale), qualitative/text (nominal categories, ordinal ranks), date/time data, or geographic data.

Data Reliability Verification & Enhancements

Verified using GITC testing, flowcharts, confirmations, recalculations, SOC 1 reports, sequence tests, batch/hash total validations, and reconciliations (e.g., subledger to general ledger). Reliability is highest when the auditor sources data directly, independently, from effective control environments, via original documents, and in written/documented form.

Visualizations & Interpreting Results

Visuals (scatter plots, bullet charts, column/line charts, pie charts) must use appropriate scaling (starting y-axis at zero) and unbiased design. Outliers in classification plots or unexpected divergence in trend analysis require drill-down procedures and further audit investigation.

Evaluating & Grouping Outcomes

Items flagged by ADAs are categorized as either clearly inconsequential (no RMM individually or in aggregate; rationale documented) or not clearly inconsequential (grouped by common characteristics, requiring further substantive procedures, fraud risk assessments, or control failure evaluations).

Information Systems

Information systems utilized by an entity to store data include Accounting Information Systems (AIS), Management Information Systems (MIS), Executive Information Systems (EIS), Decision Support Systems (DSS), Customer Relationship Management systems (CRMs), Supply Chain Management systems (SCM), Inventory Management Systems (IMS), Knowledge Management Systems (KMS), and Enterprise Resource Planning systems (ERPs).

Data Storage Functions

Repositories used to store data, listed from largest to smallest, include:

Data Lake

Stores all structured and unstructured data.

Data Warehouse

Structured and organized database tables available for analysis.

Data Mart

A subset of database tables used for specific business segments.

Data Cubes

Database tables transformed for drilling

Databases

Structured tables available for specific analysis.

Tables

Single sheets of attributes and records.

Spreadsheets

Data files that may contain tables and/or other values.

Internal and External Sources

Data for ADAs can be sourced internally or externally:

Internal & Reporting Sources

Audited financial statements, transaction logs, subledgers, general ledgers, source documents (invoices, purchase orders, receiving reports), stand

External Sources

Governmental external sources, private external sources, and service organizations (e.g., cloud computing providers).

Structured Data Elements

Components of structured data in relational databases:

Tables

Objects in a database stored in files containing rows and columns.

Attributes

Columns in a table representing characteristics or properties describing objects.

Records

Rows in a table containing information about a single entity or object.

Fields

The space at the intersection of a column and row where a data value is entered.

Database Keys

Attributes that uniquely identify records or facilitate relationships between tables:

Primary Key

A required attribute in every table containing a unique identifier (e.g., customer number or email address).

Foreign Key

An attribute in one table containing values corresponding to a primary key in another table.

Composite Key

A combination of more than one attribute used to create a unique identifier when a single attribute is insufficient.

Normalization

A database design technique that reduces data redundancy by dividing large tables into smaller tables linked together with foreign keys.

Relational Databases

An effective method to reduce data redundancies across a structured dataset by restructuring data into multiple linked tables using primary and foreign keys.

Visualization Design Best Practices

Guidelines to ensure ethical and effective decision

Choose the Right Type of Visualization

Ensure the selected chart type (e.g., scatter plot, pie chart, line chart) properly communicates the intended information.

Apply Correct Scaling

Start y

Utilize Appropriate Colors

Select color schemes while considering the audience and cultural context to avoid misinterpretation.

Emphasize Focus Areas

Design visualizations to focus directly on the target ADA objective while avoiding bias.

ADA Visual Interpretation Techniques

Regression Analysis, Variance Analysis, Period over Period Analysis, Classification Analysis, Trend Analysis

Regression Analysis

Evaluates relationships between dependent and independent variables using scatter plots and regression lines ($R^2$ values) to set expectations for comparison against actual results.

Variance Analysis

Compares forecasted/budgeted values to actual values using tools like bullet charts; significant deviations past acceptable margins trigger further audit procedures.

Period

over

Classification

A predictive analytic using scatter plots, pie charts, or tree maps to assign new data points to historical categories; outliers require increased audit judgment and investigation.

Trend Analysis

Uses line charts to evaluate patterns over time and set expectations; inconsistent trends relative to comparative data warrant drilling up/down into timelines.

Evaluating & Grouping Misstatements

Clearly Inconsequential, Not Clearly Inconsequential, and Additional Procedures

Clearly Inconsequential

Items or groups of items determined not to pose a risk of material misstatement individually or in aggregate; rationale based on amount or nature must be documented.

Not Clearly Inconsequential

Possible misstatements requiring grouping by common characteristics and further analysis to determine if actual misstatements exist.

Additional Procedures

Further qualitative and quantitative evaluations applied to items that remain "not clearly inconsequential," including assessing potential fraud risk, control/process failures, and expanding substantive tests.