Spreadsheet Design, Data Quality, and Error Control

Introduction

Excel is often introduced as a tool for entering numbers and performing calculations. In business analytics, however, a spreadsheet is better understood as a small information system. It stores data, applies rules, generates analytical results, and communicates evidence for decisions. A well-designed workbook can support budgeting, forecasting, pricing, inventory analysis, financial reporting, and operational control. A poorly designed workbook can produce incorrect decisions while appearing entirely credible.

Spreadsheet quality therefore involves more than correct formulas. It includes the quality of the underlying data, the clarity of the workbook’s structure, the traceability of assumptions, the control of errors, and the ability of another person to reproduce and audit the results. These concerns are particularly important because spreadsheets are frequently created by individual analysts but later used by managers, auditors, colleagues, regulators, or clients.

This chapter develops a practical design framework based on six principles:

1. Separate inputs, calculations, and outputs.
2. Use consistent conventions and transparent structure.
3. Control data quality at the point of entry.
4. Detect and manage formula and logic errors.
5. Document assumptions, sources, and procedures.
6. Design for reproducibility, review, and auditability.

Learning Objectives

By the end of this chapter, you should be able to:

• explain why spreadsheet architecture affects reliability;
• design a workbook using separate input, calculation, and output areas;
• apply naming, formatting, and documentation conventions;
• distinguish validity, accuracy, completeness, consistency, timeliness, uniqueness, and relevance in data quality;
• use Excel validation, conditional formatting, structured tables, and error checks;
• identify common formula errors, circular references, hard-coded assumptions, and hidden dependencies;
• evaluate a spreadsheet for reproducibility and auditability; and
• recommend controls appropriate to business risk.

1. The Spreadsheet as an Analytical Model

An analytical spreadsheet transforms information through a sequence of stages:

source data → controlled inputs → calculations → outputs → decisions

The workbook is reliable only when each stage is adequately controlled. For example, a sales forecast may be mathematically correct but still misleading if the sales history contains duplicate invoices, the growth assumption is unexplained, or the output uses an outdated input range.

A useful conceptual model is:

• Inputs: values supplied by users or imported from external systems.
• Calculations: formulas, transformations, summaries, and analytical logic.
• Outputs: reports, dashboards, charts, indicators, and decision measures.
• Controls: checks that test whether inputs and calculations behave as expected.
• Documentation: explanations of purpose, sources, assumptions, ownership, and maintenance.

This model is related to the principle of separation of concerns in information systems. A change to a sales assumption should not require rewriting a chart. Similarly, correcting an imported transaction should update calculations without manually editing several reports.

1.1 Separation of Inputs, Calculations, and Outputs

A basic workbook may contain the following worksheets:

The precise arrangement depends on the task. Small workbooks may combine some areas, while high-risk workbooks should separate them more strongly. The principle is not to create as many sheets as possible; it is to make the flow of information understandable and controlled.

Inputs should be visually distinguishable from formulas. For example, blue font may indicate user-entered cells, black font may indicate formulas, and green cells may indicate imported data. Color alone is not sufficient because it may be inaccessible to some users; labels, protection, and documentation should reinforce the convention.

Outputs should not normally contain manually typed intermediate numbers. If a manager sees a total revenue figure on a dashboard, that figure should be linked to a calculation or controlled summary rather than retyped. Manual duplication creates the possibility that the displayed result will diverge from the underlying model.

1.2 The Principle of One Source of Truth

A value used in several calculations should ideally be stored once and referenced wherever required. Suppose a workbook uses a tax rate in five formulas. Entering the rate separately in five cells creates five opportunities for inconsistency. A single assumption cell, such as TaxRate, is preferable.

This principle does not mean that every number must be stored in one cell. Transaction-level data naturally contains repeated fields. It means that shared assumptions and master definitions should have a controlled, authoritative location.

2. Design Conventions for Maintainable Workbooks

Maintainability is the ease with which a competent user can understand, update, test, and extend a workbook without introducing errors. Maintainable design reduces dependence on the original author.

2.1 Layout and Naming Conventions

A workbook should use a predictable layout. Common conventions include:

• place titles and instructions above or beside the working area rather than inside a data table;
• use one row for column headers and avoid merged cells within datasets;
• keep each column to one variable or attribute;
• use consistent date, currency, percentage, and decimal formats;
• avoid blank rows and columns inside a table;
• avoid storing multiple unrelated tables in the same rectangular range;
• use descriptive worksheet names such as Sales_Input rather than Sheet3; and
• keep formulas consistent down a column whenever possible.

Column headings should be precise. “Date” may be insufficient if a workbook contains invoice date, shipment date, and payment date. Names such as Invoice_Date, Shipment_Date, and Payment_Date reduce ambiguity.

File names should communicate identity and version. A convention such as Regional_Sales_Forecast_2025-06-30_v03.xlsx is more useful than final_final2.xlsx. In controlled environments, version history or document-management software should be preferred to manual filename proliferation.

2.2 Structured Tables

Excel Tables provide a useful structure for datasets. They automatically expand when new rows are added, support filters, preserve formulas in calculated columns, and allow structured references such as Sales[Revenue]. Compared with fixed ranges such as A2:A5000, structured references reduce the risk that newly added records will be excluded.

A sound data table generally has:

• a unique record identifier where appropriate;
• one header row;
• no merged cells;
• consistent data types within each column;
• one observation per row; and
• clearly defined missing-value treatment.

The “one observation per row” rule is central to analysis. If a sales table contains one row per invoice line, the analyst must not interpret each row as one customer or one invoice without appropriate aggregation.

2.3 Named Ranges

A named range assigns a meaningful name to a cell or range. For example, a cell containing an annual growth assumption may be named GrowthRate. A formula such as =Beginning_Sales(1+GrowthRate) is easier to interpret than =B6(1+$F$3).

Names improve readability, reduce dependence on cell coordinates, and can make formulas more robust when worksheet layouts change. They are especially useful for:

• assumptions used in many formulas;
• lookup tables;
• validation lists;
• key output cells; and
• ranges used in dashboards or reports.

Names should be unique, descriptive, and stable. Avoid names that resemble cell references, contain spaces, or have unclear abbreviations. A naming convention such as Assump_TaxRate, List_Region, and Out_TotalRevenue can communicate purpose.

Named ranges do not automatically make a workbook correct. A poorly defined name, an outdated reference, or a name referring to the wrong sheet can create hidden errors. Names should therefore be reviewed through Excel’s Name Manager and documented where their purpose is not obvious.

3. Data Quality in Business Analytics

Data quality is multidimensional. A dataset is not simply “good” or “bad”; it may be accurate but incomplete, complete but outdated, or consistent in format but irrelevant to the decision.

3.1 Major Dimensions of Data Quality

These dimensions may conflict. Increasing completeness by filling every blank with zero can reduce accuracy if a blank means “unknown” rather than “none.” Data-quality treatment must therefore reflect business meaning.

3.2 Missing, Invalid, and Zero Values

A missing value, a zero, and a not-applicable value are not interchangeable. For example:

• blank delivery date: delivery may not yet have occurred;
• zero delivery cost: delivery occurred at no charge;
• N/A: delivery is not relevant to the transaction.

Replacing all blanks with zero may distort averages, totals, ratios, and interpretations. A documented data dictionary should define each field, acceptable values, units, and missing-value conventions.

3.3 Data Profiling

Before analysis, profile the dataset. Useful questions include:

• How many rows and columns are present?
• What is the minimum and maximum date?
• Which fields contain blanks?
• Which identifiers are duplicated?
• Are numerical values within plausible bounds?
• Do category labels use consistent spelling and capitalization?
• Do totals reconcile to a trusted source?

In Excel, profiling can involve COUNT, COUNTA, COUNTBLANK, MIN, MAX, UNIQUE, COUNTIF, PivotTables, filters, and conditional formatting. For example, the number of blank customer IDs in a range can be assessed with =COUNTBLANK(Customer_ID_Range). The result is meaningful only if the range and the business requirement are correctly defined.

4. Validation and Preventive Controls

Error control is strongest when errors are prevented before they enter the model. Excel Data Validation can restrict entries by type, range, date, list, or custom formula.

Examples include:

• restrict a quantity to whole numbers greater than or equal to zero;
• restrict a percentage to a value between 0% and 100%;
• restrict a department field to an approved list;
• require a date within the reporting period; and
• display an input message explaining the expected value.

Validation should be designed around business rules. A rule that permits values from 0 to 100 may be technically valid for a percentage but inappropriate if the measure is a discount rate whose permitted range is 0% to 40%.

Validation is not a complete security mechanism. Users may paste values over validation rules, and rules may be lost when data is copied between workbooks. High-risk workbooks should combine validation with protection, exception reports, reconciliation, and review.

4.1 Conditional Formatting

Conditional formatting is useful for highlighting exceptions rather than decorating a workbook. Appropriate uses include:

• negative inventory;
• overdue dates;
• duplicate identifiers;
• missing required fields;
• values outside control limits; and
• failed reconciliation checks.

Rules should be tested using boundary values. A rule intended to identify amounts greater than 10,000 should be checked with 9,999, 10,000, and 10,001. Visual alerts should also have a documented response: highlighting an error without explaining who investigates it is an incomplete control.

5. Formula Errors and Logic Risks

Excel displays several common error values. Their meanings and typical remedies differ.

Suppressing errors with IFERROR can improve presentation, but indiscriminate use is dangerous. For example, =IFERROR(Revenue/Units,0) treats every error as zero. That may conceal a missing lookup, invalid text value, or broken reference. Error handling should distinguish an expected business condition from a model failure.

A safer design may explicitly test the denominator: =IF(Units=0,"Not meaningful",Revenue/Units). The correct response depends on the measure’s definition and reporting requirements.

5.1 Formula and Logic Risks

Many spreadsheet failures are not syntax errors. Excel may calculate a formula successfully even when the formula represents the wrong business logic. Common risks include:

• incorrect relative or absolute references;
• omitted rows in a total;
• inconsistent formulas copied down a column;
• hard-coded numbers inside formulas;
• incorrect lookup match modes;
• dates stored as text;
• units mixed within a calculation;
• percentages entered as whole numbers rather than decimal fractions;
• hidden rows or columns affecting interpretation;
• stale external links; and
• formulas replaced by values during copy-and-paste.

For example, a formula =B2*C2 may be syntactically valid but wrong if column B is hours and column C is a wage rate expressed per day. Unit consistency is a practical control: multiplying hours by currency per hour produces currency, whereas multiplying hours by currency per day requires a conversion assumption.

5.2 Circular References

A circular reference occurs when a formula depends directly or indirectly on itself. If cell A1 refers to B1 and B1 refers to A1, the calculation has no straightforward starting point. Circularity may arise accidentally through incorrect references or intentionally in models involving interest, tax, or iterative financial calculations.

Accidental circular references should normally be removed by restructuring the model. Intentional iterative calculations require documented assumptions, a convergence policy, and testing. Enabling iterative calculation merely to make a warning disappear is not an adequate control.

6. Control Checks and Reconciliation

A control check compares an expected condition with the actual workbook state. Strong workbooks make failures visible.

Examples include:

• record count check: imported row count agrees with the source extract;
• completeness check: required fields contain no blanks;
• uniqueness check: transaction identifiers are unique where required;
• balance check: debits equal credits or beginning balance plus movements equals ending balance;
• reconciliation check: workbook total agrees with an approved source total; and
• reasonableness check: a result lies within a plausible business range.

A reconciliation difference can be expressed as:

The expected condition is often Difference = 0, subject to a documented tolerance for rounding or measurement. The formula is meaningful only when both totals use the same population, period, currency, and definition.

A percentage variance may be calculated as:

Here, Actual is the observed value and Reference is the comparison value. The formula assumes that Reference ≠ 0 and that both values use the same units and scope. If the reference is zero, a different rule is required; reporting an undefined percentage as zero may be misleading.

Checks should have clear statuses such as OK, WARNING, and FAIL. A dashboard should not present polished performance indicators while a critical check is failing. In high-risk contexts, failed checks should block publication or require formal approval.

7. Documentation, Reproducibility, and Auditability

Documentation explains what a workbook does and how it should be used. At minimum, a professional workbook should identify:

• purpose and intended users;
• data sources and extraction date;
• owner and reviewer;
• reporting period and currency;
• definitions of important measures;
• assumptions and their rationale;
• update procedures;
• known limitations; and
• change history.

A data dictionary can be maintained on a separate sheet. Typical fields include variable name, business definition, data type, unit, valid values, source, and treatment of missing data.

7.1 Reproducibility

A result is reproducible when another competent analyst can obtain the same result by following the documented process with the same source data and assumptions. Reproducibility requires more than saving the final number. It requires preserving the input data, transformation logic, parameter values, and relevant software or process information.

Practical measures include:

• preserve raw data separately from transformed data;
• avoid undocumented manual adjustments;
• record extraction and update dates;
• use stable formulas and named assumptions;
• document refresh steps; and
• retain prior versions or use a controlled version-history system.

7.2 Auditability

Auditability is the ability to trace an output back to its source and understand the transformations applied. Excel’s formula auditing tools, precedents, dependents, Show Formulas, and Evaluate Formula features can support review. However, an audit trail is strongest when the workbook’s architecture is clear before these tools are used.

A reviewer should be able to answer:

1. Where did this number originate?
2. Which assumptions affect it?
3. Which formulas transform the source value?
4. What checks confirm that the result is reasonable?
5. Who changed the workbook and when?

8. Worked Example: Designing a Sales-Margin Workbook

Assume a retailer wants to assess product profitability. The source data contains product code, sales units, selling price, unit cost, and region.

Step 1: Define the data structure

Each row represents one product-region-period observation. The data table includes a unique record ID. Selling price and unit cost are expressed in currency per unit; units are whole units; revenue and cost are currency amounts.

Step 2: Separate inputs and calculations

The Inputs sheet contains the reporting period and an approved regional target margin. The Raw_Data sheet preserves the source extract. The Calculations sheet contains formulas:

The first three equations require consistent units. The margin percentage assumes Revenue ≠ 0. If revenue is zero, the workbook should display an explicitly defined result such as Not meaningful, rather than silently reporting zero.

Step 3: Add validation and checks

Units are validated as non-negative whole numbers. Selling price and unit cost are validated as non-negative currency values. Product codes are checked against the approved product list. A duplicate-record check identifies repeated record IDs. A completeness check identifies missing product codes, units, prices, or costs.

An independent total of revenue is calculated from the source extract and compared with the workbook total. If the difference is not zero, the workbook displays FAIL and the dashboard includes a warning.

Step 4: Design the output

The dashboard reports total revenue, gross margin, margin percentage, and regional performance. These outputs are linked to calculation tables rather than manually typed. A note identifies the extraction date and the definition of gross margin.

Step 5: Review edge cases

The analyst tests:

• zero units;
• zero selling price;
• missing cost;
• a negative quantity entered accidentally;
• an unknown product code;
• a duplicate record; and
• a new region added to the source data.

Testing these cases is more informative than testing only ordinary records because spreadsheet failures often occur at boundaries and exceptions.

9. Applied Case: Monthly Inventory Planning

The following is an illustrative case.

Background

A distribution company uses an Excel workbook to plan monthly inventory replenishment. The workbook combines stock records, open purchase orders, expected demand, and supplier lead times.

Problem

Managers notice that stockout warnings differ between regional versions of the workbook. Investigation reveals that some files use a fixed range ending at row 5,000, while newer data extends beyond that row. In addition, lead times are entered in some files as calendar days and in others as business days. Several reorder-point assumptions are hard-coded in formulas.

Analysis

The problem is not one incorrect cell alone. It involves weak architecture and inconsistent definitions:

• fixed ranges exclude new records;
• separate regional files create multiple versions of assumptions;
• hard-coded values obscure the model’s logic;
• units for lead time are inconsistent; and
• no reconciliation checks compare workbook totals with the inventory system.

Approach and solution

The company redesigns the workbook into five controlled areas. Raw inventory data is imported into an Excel Table. A central assumptions sheet stores service-level targets and lead-time definitions. Calculations use structured references and named ranges. Inputs are validated against product and supplier lists. A control sheet checks record counts, duplicate product-location keys, missing stock balances, and reconciliation to the inventory system.

The reorder-point logic is documented as a business rule rather than hidden in a formula. For a simple model, reorder point may be defined as expected demand during lead time plus a documented safety-stock amount:

The terms must use compatible units: if demand is units per day, lead time must be measured in days, and safety stock must be measured in units. If business days are used, the workbook must state how weekends and holidays are treated.

Outcome

Regional workbooks are replaced by a common template with controlled assumptions and a documented refresh process. New rows are included automatically through the structured table. Failed checks prevent the dashboard from being distributed until investigated.

Learning points

The case demonstrates that spreadsheet reliability depends on definitions, structure, and governance as well as formulas. A technically valid formula can still produce an invalid business result when its ranges, units, assumptions, or population are wrong.

10. Spreadsheet Governance, Ethics, and Contemporary Practice

Spreadsheet governance should be proportional to risk. A personal analysis of low-risk information may require only basic documentation. A workbook used for regulatory reporting, payroll, lending, financial statements, or safety decisions requires stronger controls, independent review, access management, version control, and retention procedures.

Ethically, analysts should not conceal errors by formatting them away, replacing unexplained values with zeros, or presenting estimates as observed facts. Assumptions should be disclosed, uncertainty should be acknowledged, and significant limitations should be communicated to decision-makers.

Confidential data also requires care. Workbooks may contain personal, financial, or commercially sensitive information. Password protection is not a substitute for appropriate access controls, secure storage, data minimization, and organizational policies.

Modern Excel includes Power Query, Power Pivot, dynamic arrays, data types, and integration with cloud services. These features can improve repeatability and scalability, but they introduce new dependencies. A query may rely on a file path, a credential, or a transformation step that another user cannot access. A professional workbook documents these dependencies and tests refresh behavior.

Generative AI can assist with formula explanation, documentation drafts, and test-case generation. It should not be treated as an authority for business definitions or as a substitute for validation. AI-generated formulas must be tested against known examples, boundary cases, and independent calculations. Confidential data should not be submitted to tools without authorization.

11. A Practical Quality-Review Framework

Before releasing a workbook, conduct a structured review.

Structure

• Are inputs, calculations, outputs, and checks distinguishable?
• Are formulas separated from manual entries?
• Are tables complete and free from merged cells inside data areas?

Data

• Are sources, dates, units, and definitions documented?
• Have completeness, validity, duplication, and consistency been tested?
• Are missing, zero, and not-applicable values distinguished?

Formula logic

• Are formulas consistent across rows and columns?
• Are absolute and relative references intentional?
• Are hard-coded assumptions removed or documented?
• Are error values and zero denominators handled appropriately?

Controls

• Do reconciliations agree with independent sources?
• Are failed checks visible and actionable?
• Have boundary and exception cases been tested?

Communication and governance

• Can another analyst reproduce the result?
• Are ownership, review date, and change history recorded?
• Is sensitive information appropriately protected?
• Are limitations and assumptions visible to decision-makers?

Chapter Summary

A spreadsheet used for business analytics is an analytical information system, not merely a collection of cells. Reliable workbooks separate inputs, calculations, outputs, checks, and documentation. Maintainability is improved through consistent layout, meaningful names, structured tables, controlled assumptions, and a single source of truth.

Data quality must be assessed across multiple dimensions, including accuracy, completeness, validity, consistency, uniqueness, timeliness, relevance, and integrity. Validation and conditional formatting provide preventive and detective controls, but they must be supported by reconciliation, documentation, and review.

Formula errors such as #DIV/0!, #N/A, and #REF! have different meanings and require different remedies. Circular references, hard-coded assumptions, incorrect ranges, mixed units, and hidden manual adjustments are important risks even when Excel produces no visible error. Reproducibility requires preserving data, assumptions, and procedures; auditability requires tracing outputs back to sources and transformations. Finally, governance and ethical practice require proportionate controls, transparent assumptions, appropriate protection of sensitive data, and responsible use of automation and AI.

Key Terms

Data validation: Rules that restrict or guide allowable cell entries.

Auditability: The ability to trace results to sources, assumptions, and transformations.

Calculation area: The part of a workbook containing formulas and analytical logic.

Circular reference: A direct or indirect formula dependency on itself.

Completeness: The extent to which required records and values are present.

Control check: A test that compares an actual condition with an expected condition.

Data dictionary: Documentation describing fields, definitions, types, units, and valid values.

Data quality: The suitability of data for its intended use across several dimensions.

Input: A value entered or imported into a model for use in calculations.

Integrity: The reliability of relationships among records, identifiers, and fields.

Named range: A meaningful name assigned to a cell or range.

One source of truth: The principle that a shared assumption or definition should have one controlled authoritative location.

Output: A report, indicator, chart, or result produced by a workbook.

Reconciliation: Comparison of a workbook result with an independent source or expected total.

Relevance: The extent to which data supports the decision or analysis.

Reproducibility: The ability to obtain the same result using the same data, assumptions, and documented process.

Structured table: An Excel table with defined headers, expandable records, and structured references.

Uniqueness: The absence of unintended duplicate records.

Validity: Conformity to permitted formats, ranges, categories, or business rules.

Review Questions

Short Answer Questions

1. Why should inputs, calculations, and outputs be separated in a business workbook?
2. Distinguish between a missing value, a zero value, and a not-applicable value.
3. What is the purpose of a named range?
4. What does the Excel error #REF! usually indicate?
5. Define reproducibility in the context of spreadsheet analytics.

Long Answer Questions

1. Explain how structured tables, naming conventions, and documentation improve spreadsheet maintainability.
2. Discuss at least five dimensions of data quality and give a business example of each.
3. Explain why using IFERROR indiscriminately can create analytical risk.
4. Describe a control framework for a workbook used in monthly management reporting.
5. Compare preventive controls, such as validation, with detective controls, such as reconciliation and exception reporting.

Analytical / Application Questions

1. A workbook calculates profit margin as Profit ÷ Sales. Sales may sometimes equal zero. Design an appropriate error-handling rule and explain why reporting zero margin may be misleading.
2. An imported transaction file contains 2,400 rows, but the summary formula uses a fixed range ending at row 2,000. Identify the risk, recommend a redesign, and propose a control check.
3. A replenishment model combines demand in units per day with lead time recorded in business days and safety stock in units. Explain what must be clarified before the reorder point can be trusted, and identify at least two tests you would perform.

Multiple Choice Questions with Answers

Question 1

Which workbook structure most clearly separates responsibilities and supports auditability?

A. One worksheet containing inputs, formulas, and charts mixed together

B. Separate areas for inputs, calculations, outputs, checks, and documentation

C. A dashboard containing manually typed totals

D. Several worksheets with no stated relationship between them

Correct Answer: B. Separate areas for inputs, calculations, outputs, checks, and documentation

Question 2

Which data-quality dimension concerns unintended duplicate transaction records?

A. Timeliness

B. Relevance

C. Uniqueness

D. Completeness

Correct Answer: C. Uniqueness

Question 3

What is a principal advantage of using an Excel Table for transaction data?

A. It guarantees that all business rules are correct

B. It prevents users from entering invalid data under all circumstances

C. It automatically removes all duplicate records

D. It can expand with new records and support structured references

Correct Answer: D. It can expand with new records and support structured references

Question 4

What does the error value #DIV/0! usually indicate?

A. A formula is dividing by zero or an empty denominator

B. A worksheet has been renamed

C. A lookup found no matching record

D. A dynamic-array result is blocked

Correct Answer: A. A formula is dividing by zero or an empty denominator

Question 5

Why can indiscriminate use of IFERROR be dangerous?

A. It always converts numbers into text

B. It can conceal different underlying failures by replacing them with one result

C. It prevents formulas from being copied

D. It deletes the source data

Correct Answer: B. It can conceal different underlying failures by replacing them with one result

Question 6

Which practice best supports the principle of one source of truth?

A. Typing the same tax rate into every formula

B. Storing a shared tax rate once and referencing it by a documented name

C. Copying assumptions into every worksheet

D. Hard-coding assumptions in dashboard labels

Correct Answer: B. Storing a shared tax rate once and referencing it by a documented name

Question 7

A reconciliation compares a workbook total of 98,500 with an independent source total of 100,000. What is the difference using Workbook total − Independent source total?

A. 1,500

B. −1,500

C. 98,500

D. 100,000

Correct Answer: B. −1,500

Question 8

Which statement about a circular reference is correct?

A. It occurs when a formula depends directly or indirectly on itself

B. It occurs whenever a workbook contains a named range

C. It means every formula in a worksheet is invalid

D. It is always solved by hiding the warning

Correct Answer: A. It occurs when a formula depends directly or indirectly on itself

Question 9

If revenue is 12,000 and cost is 9,000, what is the gross margin percentage?

A. 25%

B. 33.3%

C. 75%

D. 133.3%

Correct Answer: B. 33.3%

Question 10

Which action most directly improves reproducibility?

A. Removing the source data after producing the dashboard

B. Preserving source data, assumptions, transformation steps, and update dates

C. Using more colors in the output sheet

D. Replacing formulas with final values before distribution

Correct Answer: B. Preserving source data, assumptions, transformation steps, and update dates