Blog · Board and management reporting
Compare the observed change in a business measure with the segment and driver breakdowns behind it. Work the same revenue rows both ways and see what the evidence does, and does not, establish.
Descriptive and diagnostic analytics use the same records to answer different questions: what changed, and which recorded components account for that change. A revenue total describes the result; a product and price breakdown gives the team a specific investigation. Neither a segment comparison nor a tidy arithmetic bridge proves what caused customers to behave differently.
Descriptive analytics summarizes an observed measure over a defined period. Diagnostic analytics compares the components of that measure across periods, groups, or operating conditions. The distinction is the question and the evidence needed, rather than the appearance of the report.
| Question | Descriptive analytics | Diagnostic analytics |
|---|---|---|
| Starting point | What was revenue, and how much did it change? | Which products and recorded drivers account for the change? |
| Calculation | Totals, differences, and rates | Segment differences and reconciled driver contributions |
| Evidence needed | A complete, consistently defined measure | The same measure plus relevant detail and comparison fields |
| Useful output | A confirmed change worth attention | A quantified finding and a focused follow-up question |
| Limit | A company total can hide offsetting movements | A contribution is not proof of causation |
The broader types of business analytics include questions about future outcomes and possible actions. Keep those separate from explaining the records already supplied.
A US equipment distributor compares August and September net product revenue. Its export contains invoice date, product tier, units, revenue after discounts, currency, and credit adjustments. The table below is the reconciled product summary; realized price is revenue divided by units, rather than the list price in a catalog.
| Product tier | August units | August realized price (USD) | August revenue (USD) | September units | September realized price (USD) | September revenue (USD) |
|---|---|---|---|---|---|---|
| Standard | 1,000 | 100 | 100,000 | 900 | 105 | 94,500 |
| Premium | 500 | 200 | 100,000 | 600 | 190 | 114,000 |
| Total | 1,500 | 200,000 | 1,500 | 208,500 |
Each revenue cell equals that row's units multiplied by realized price. The revenue totals must tie to the distributor's net sales report for the same periods. Product tiers must be mutually exclusive, and a credit adjustment must appear once under the agreed treatment.
A spreadsheet can sum only the records matching a period and product tier. Microsoft's SUMIFS documentation describes summing values that meet multiple criteria, which supports rebuilding each grouped total from the invoice rows. Covirage is not affiliated with Microsoft Excel.
The company revenue change is the current total minus the previous total. Its percentage change uses the previous total as the denominator.
Revenue change = September revenue - August revenue
Revenue change percentage = Revenue change / August revenue
| Measure | Calculation | Result |
|---|---|---|
| Revenue change (USD) | 208,500 - 200,000 | 8,500 |
| Revenue growth | 8,500 / 200,000 | 4.25% |
| Unit change | 1,500 - 1,500 | 0 |
| August average realized price (USD) | 200,000 / 1,500 | 133.33 |
| September average realized price (USD) | 208,500 / 1,500 | 139.00 |
Prices are rounded to cents; calculations use unrounded values. The descriptive conclusion is clear: revenue increased by $8,500, or 4.25%, while total units were unchanged. Average revenue per unit increased, but that alone does not show whether individual prices rose or the company sold a different mix.
First separate the product movements. Standard revenue fell by $5,500, while Premium revenue increased by $14,000. Those movements sum to the company's $8,500 increase, so the segment breakdown closes.
Then split each segment into a unit effect at the previous price and a price effect on current units:
Unit effect = (September units - August units) × August price
Price effect = September units × (September price - August price)
| Product tier | Revenue change (USD) | Unit effect calculation | Unit effect (USD) | Price effect calculation | Price effect (USD) |
|---|---|---|---|---|---|
| Standard | -5,500 | (900 - 1,000) × 100 | -10,000 | 900 × (105 - 100) | 4,500 |
| Premium | 14,000 | (600 - 500) × 200 | 20,000 | 600 × (190 - 200) | -6,000 |
| Total | 8,500 | 10,000 | -1,500 |
The check is $10,000 - $1,500 = $8,500. Both segment rows also reconcile individually. This convention assigns the interaction between unit and price changes to the price effect; use the same convention in each review.
Here, total units stayed constant. The positive unit effect reflects a shift toward the higher-priced Premium tier, valued at August prices. The recorded price effect is negative overall, even though company average revenue per unit rose. The diagnostic analytics methods page covers the investigation beyond this pairwise comparison.
The sales lead owns the follow-up on Premium pricing. The operations lead checks whether Standard availability constrained shipments. Finance owns the net revenue definition and the reconciliation. Together they can distinguish a recorded movement from an explanation supported by operating evidence.
Do not conclude that a discount created Premium demand simply because its realized price fell while units rose. Account mix, shipment timing, availability, and commercial terms could all matter. NIST's guidance on association explains why an observed relationship does not by itself establish causation.
Use diagnostic analytics tools when choosing how to reproduce the investigation. If the next question is what might happen next month, the descriptive versus predictive comparison covers that separate job.
Bring an invoice export with dates, product groups, units, and net revenue; Covirage maps the columns, and its tools compute and check the period and segment comparisons. The external AI model explains the result with every figure traced to its rows. Create a free Covirage account, which takes about a minute, then set up your business and bring a sample of your own data.
See self-service analytics for the business workflow behind those comparisons.
Yes. The same invoice export can provide a revenue total and a breakdown by customer, product, or period. Diagnostic work needs enough detail to separate the proposed drivers, so a company total alone cannot explain a change in product mix.
It can establish which recorded segments and arithmetic components account for the change. Explaining why customers ordered less requires additional evidence, such as availability records, account conversations, or a controlled test. A reconciled breakdown supports that investigation without proving its cause.
Start with a descriptive measure that ties to the source report. Then investigate a meaningful change using consistent segments and driver definitions. If the opening total is wrong, a detailed diagnostic breakdown can produce a precise explanation of an incorrect figure.