Data Analytics for Accountants: Descriptive vs Predictive vs Prescriptive
By SimplyCPA · 11 October 2026
On this page (6)
- Descriptive: What happened? — summaries, dashboards, ageing reports.
- Diagnostic: Why did it happen? — drill-downs, variance analysis, correlations.
- Predictive: What will happen? — forecasting, regression, credit-risk scoring.
- Prescriptive: What should we do? — optimisation and recommendations, such as inventory reorder points.
- Garbage in, garbage out: Analytics are only as good as the data's completeness, accuracy, consistency and timeliness — which is why data governance matters.
The four types of analytics
| Type | Question | Accounting example | Typical techniques |
|---|---|---|---|
| Descriptive | What happened? | Monthly sales by region; accounts receivable ageing | Aggregation, summary statistics, dashboards |
| Diagnostic | Why did it happen? | Why gross margin fell in Q3 — mix, price or cost? | Drill-down, variance analysis, correlation |
| Predictive | What is likely to happen? | Cash flow forecast; probability that a customer defaults | Regression, time-series forecasting, classification models |
| Prescriptive | What should we do about it? | Optimal reorder quantity; which customers to offer early-payment discounts | Optimisation, simulation, decision rules |
Complexity — and value — generally increase from descriptive to prescriptive.
The analytics workflow
- Ask the question — what decision will the analysis support?
- Get the data — identify sources; extract, transform and load (ETL) it into a usable form.
- Prepare the data — clean duplicates, missing values and inconsistent formats; validate totals against the ledger.
- Analyse — apply the right technique.
- Communicate — visualise and explain the result to the decision-maker.
Choosing a visualisation
- Line chart — trends over time.
- Bar chart — comparing categories.
- Scatter plot — the relationship between two variables (for example, advertising spend vs sales).
- Histogram — the distribution of a single variable (for example, invoice amounts — useful for spotting unusual clusters).
Data quality dimensions
- Completeness — all records are present.
- Accuracy — values are correct.
- Consistency — the same data agrees across systems.
- Timeliness — data is current enough for the decision.
- Validity — data conforms to defined formats and rules.
Analytics in audit
Auditors use analytics to test 100% of a population — for example, every journal entry posted on weekends, by unusual users or just below approval limits. PCAOB amendments to AS 1105 and AS 2301 on technology-assisted analysis (effective for fiscal years beginning on or after December 15, 2025) clarify the auditor's responsibilities when using such tools and when evaluating the reliability of the client's data.
Exam tip: If a question asks which analysis tells management what action to take, the answer is prescriptive. "Predict" questions are predictive — even if the model is simple.
Continue with database management and data governance.