01 | Course Lens

From accounting records to decision-ready insight.

Accounting analytics connects a business question to reliable data, an appropriate analytical method, and a supported conclusion.

Ask

What decision or accounting question are we trying to support?

Prepare

Which records are relevant, and what must be cleaned or validated?

Analyze

Which comparison, calculation, model, or visual best fits the question?

Communicate

What does the evidence support, and what should the audience understand?

Course question: What decision will this analysis support?

02 | Analytics Process

Analytics follows a deliberate process.

Click the six steps in the best order. The order matters because weak problem definition creates weak analysis.

Accounting example: Before building an AR aging dashboard, define the decision, choose the aging metric, identify authoritative invoice and payment sources, then prepare and validate the records.

03 | Four Types of Analytics

Same business issue, four different questions.

DescriptiveWhat happened?

Summarize results.

Flip

AR example

Overdue balances increased 14% this month.

DiagnosticWhy?

Investigate drivers.

Flip

AR example

Three large customers explain most of the increase.

PredictiveWhat may happen?

Estimate likely outcomes.

Flip

AR example

Which invoices are most likely to become overdue?

PrescriptiveWhat should we do?

Compare actions.

Flip

AR example

Which collection strategy should be tested first?

Why are delays concentrated in two locations?

Which orders are most likely to be delayed next week?

04 | Applied Process

Duplicate-payment review: find a signal, then investigate it.

Click each workflow stage to see what the accountant actually does.

Ask: Are we paying the same invoice more than once? Define the exception criteria before searching.

Key idea: An anomaly is not proof of error. Analytics identifies where attention should go.

05 | Skills + Data Types

Strong analysts combine judgment, tools, and data awareness.

Technical

Excel, Power Query, Alteryx, Power BI, queries, repeatable workflows.

Functional

Metrics, statistics, analytical methods, validation, interpretation.

Communication

Explain findings clearly, separate observation from interpretation, support action.

General ledger table with fixed columns and data types

Customer complaint paragraphs and attached images

Bridge: Source and format are different. A bank feed is a source. JSON may be the format used to deliver it.

06 | Formats + Sources

Mystery files: identify the format from the structure.

Date,Account,Debit,Credit
2026-01-31,Cash,1200,0
2026-01-31,Revenue,0,1200

Delimited text, easy to ingest, but limited metadata and formatting.

{
  "invoice_id": "INV-1042",
  "customer": "Aster Foods",
  "amount": 8420
}

Key-value structure, common in APIs and integrations.

Tagged structure, often used in system exchange.

Workbook: [Transactions] [Customers] [COA]
Cell F2: =XLOOKUP(D2,Customers!A:A,Customers!C:C)
Tables + formulas + formatting

Rich workbook format, useful but may hide formulas, sheets, and assumptions.

Quarterly Report : Page 12
Revenue ........ $8,420,000
Gross Margin ... 41.8%
Fixed visual layout

Designed for reading. Extracted values should be reconciled before analysis.

Which format would you least want for a 500,000-row transaction analysis?

07 | ETL + Alteryx

ETL turns raw records into analysis-ready data.

Extract

Read invoices, payments, customer terms, ERP exports, APIs, or databases.

Transform

Join, clean, standardize, calculate, validate, and reshape the records.

Load

Create one clean dataset for Excel, Power BI, or the next reporting cycle.

Explore the AR aging workflow: click each step and watch the table change.

Step 1: InputBring in the source tables before changing anything.

Invoices

Invoice_ID	Customer_ID	Invoice_Date	Amount
INV-1042	C1006	8/26/26	$12,800.72
INV-1043	C1006	08-30-2026	$6,301.51
INV-1044	C1009	2026/09/28	$1,269.24

Payments

Payment_ID	Invoice_ID	Payment_Date	Amount_Applied
PAY-721	INV-1043	09/15/26	$2,000.00
PAY-722	INV-1044	10/01/26	$1,269.24

Notice the inconsistent date formats and the extra space in C1006. At this stage, preserve the raw evidence.

Step 2: JoinMatch payments to invoices and calculate what remains open.

Invoice_ID	Customer_ID	Invoice_Amount	Amount_Applied	Open_Balance
INV-1042	C1006	$12,800.72	$0.00	$12,800.72
INV-1043	C1006	$6,301.51	$2,000.00	$4,301.51
INV-1044	C1009	$1,269.24	$1,269.24	$0.00

A bad join key can duplicate or drop invoices. Check row counts and total invoice amounts before moving on.

Step 3: EnrichAdd customer credit terms from the customer master.

Customer master

Customer_ID	Customer_Name	Terms_Days
C1006	Finch & Vale Consulting	30
C1009	Juniper Bistro Group	15

After enrichment

Invoice_ID	Customer_ID	Invoice_Date	Terms_Days	Due_Date
INV-1042	C1006	2026-08-26	30	2026-09-25
INV-1043	C1006	2026-08-30	30	2026-09-29
INV-1044	C1009	2026-09-28	15	2026-10-13

The customer ID has been standardized, dates are consistent, and the terms table supplies the information needed to calculate due dates.

Step 4: FormulaCalculate days past due and assign an aging bucket.

Invoice_ID	Open_Balance	Due_Date	As_of_Date	Days_Past_Due	Aging_Bucket
INV-1042	$12,800.72	2026-09-25	2026-12-01	67	61+
INV-1043	$4,301.51	2026-09-29	2026-12-01	63	61+
INV-1044	$0.00	2026-10-13	2026-12-01	0	Paid

The same rule is applied consistently to every open invoice. Paid invoices remain out of overdue aging buckets.

Step 5: ValidateRun control checks before trusting the output.

Control checks

Check	Expected	Result
Invoice row count	3	3 ✓
Invoice total	$20,371.47	$20,371.47 ✓
Applied payments	$3,269.24	$3,269.24 ✓
Missing customer IDs	0	0 ✓

Exception review

Invoice_ID	Issue	Status
INV-1043	Source customer ID had trailing space	Standardized
INV-1044	Open balance is zero	Classified Paid

Validation is not cosmetic. It is evidence that the joins, formulas, and transformations did not distort the accounting records.

Step 6: OutputDeliver one clean, repeatable AR aging table.

Invoice_ID	Customer	Open_Balance	Due_Date	Days_Past_Due	Aging_Bucket
INV-1042	Finch & Vale Consulting	$12,800.72	2026-09-25	67	61+
INV-1043	Finch & Vale Consulting	$4,301.51	2026-09-29	63	61+
INV-1044	Juniper Bistro Group	$0.00	2026-10-13	0	Paid

This output can feed Excel or Power BI next month without rebuilding the preparation logic.

08 | Wrangling + Statistics

Prepare carefully, then choose statistics that answer the question.

Wrangling checks

Click a risk to see why it matters.

Statistics lens

Descriptive

Mean, median, dispersion, skewness, distribution.

Inferential

Confidence intervals, hypothesis tests, regression.

You want to estimate whether processing time is associated with error rate. Which approach best answers that question?

Do not confuse correlation with causation: a relationship can be useful without proving that one variable caused the other.

09 | Choose the Analysis

Start broad when exploring. Test a defined expectation when confirming.

Exploratory

Search a population broadly for patterns, anomalies, or questions you did not anticipate.

Confirmatory

Evaluate whether evidence is consistent with a specific expectation or hypothesis.

Search all expense transactions for unexpected merchant and amount patterns.

Test whether a known policy change reduced late approvals as expected.

Professional sequence: Explore what changed, diagnose likely drivers, estimate what may happen next, then compare possible actions.

10 | Visualization + Stakeholders

Choose the simplest visual that makes the intended pattern obvious.

Click a chart to see when to use it.

Analyst

Granular fields, filters, exceptions.

Manager

Drivers, context, actions.

Executive

Decision, pattern, impact.

Auditor

Evidence, coverage, reliability.

Story discipline: Separate what the data show, what you think it means, and what you recommend doing.

11 | AI in the Workflow

AI can prioritize attention. The professional evaluates the evidence.

Acquire relevant structured and unstructured data from approved, authoritative sources.

AI flags a high-risk journal entry. What is the strongest next step?

12 | Prompting with RCCO

Give AI a job, context, boundaries, and a useful output.

R

Role

Who should the AI act as?

C

Context

What data, business question, and definitions matter?

C

Constraints

What must it not assume? What must it verify?

O

Output

Table, chart, workbook logic, memo, JSON, dashboard plan?

Role

Context

Constraints

Output

Select all four RCCO components.

13 | Governance + Judgment

Responsible AI use is part of the analytical workflow.

Click each risk, then identify the control that should reduce it.

Choose a governance risk.

AI found 42 unusual records and concluded that all 42 are misstated. What is wrong?

14 | Excel Diagnostic

Excel Skills Check

This is a low-stakes Excel skills check. Complete the five tasks using any Excel methods you know. The goal is to see your current Excel comfort level and identify the skills we should practice in this course.