Blog · Board and management reporting
Diagnostic analytics breaks a business change into recorded contributors. Follow a worked revenue and cost investigation, the checks behind its explanation and the distinction between an observed driver and a proven cause.
Diagnostic analytics investigates a business change by breaking it into recorded contributors and checking that they reconcile to the total. It can show which products, customers, periods or cost movements explain a result mathematically. That evidence tells managers where to investigate; it does not automatically establish why the underlying behavior occurred.
A diagnostic question has a defined measure and comparison: why did gross profit fall between two completed months, for example? Establish the change before looking for explanations. If the starting total is wrong, a detailed explanation will only make the error harder to spot.
Descriptive reporting establishes what happened; diagnostic work examines its recorded contributors. The broader descriptive versus diagnostic analytics comparison covers that distinction. Here the focus is the investigation itself, from sales and cost rows to a checked explanation.
Use ordinary business exports: invoice lines, product quantities, net sales, matched costs and dates. Keep a stable product identifier, the currency, the reporting cutoff and a definition of which credits belong in each period. The SEC's financial statement guide explains the basic relationship: costs of sales are deducted from net revenue to arrive at gross profit.
Segmentation asks where the change occurred: by customer, location, product or department. Timing checks ask whether activity moved across a cutoff. A driver bridge separates a dollar movement into recorded quantities, realized prices and unit costs.
Exploratory comparisons also help test assumptions. NIST's introduction to exploratory data analysis emphasizes gaining insight, finding structure and checking underlying assumptions. In business work, that means inspecting unexpected records instead of treating an unusual result as self-explanatory.
Choose the simplest method that answers the question with fields you actually have. Without quantity, revenue changes cannot be split reliably into price and volume effects. Without transaction dates, a suspected cutoff issue remains a question to investigate.
An illustrative US distributor compares August with September 2026. The following table is the complete input for this example. Realized price means net revenue per unit after recorded discounts; unit cost uses the same cost policy in both months.
| Product | August units | August price (USD/unit) | August cost (USD/unit) | September units | September price (USD/unit) | September cost (USD/unit) |
|---|---|---|---|---|---|---|
| Standard | 100 | 1,000 | 600 | 90 | 950 | 630 |
| Premium | 50 | 2,000 | 1,200 | 55 | 2,000 | 1,200 |
Multiply units by price for revenue and units by cost for cost of goods sold. The resulting figures are:
| Product | August revenue (USD) | September revenue (USD) | August COGS (USD) | September COGS (USD) | August gross profit (USD) | September gross profit (USD) |
|---|---|---|---|---|---|---|
| Standard | 100,000 | 85,500 | 60,000 | 56,700 | 40,000 | 28,800 |
| Premium | 100,000 | 110,000 | 60,000 | 66,000 | 40,000 | 44,000 |
| Total | 200,000 | 195,500 | 120,000 | 122,700 | 80,000 | 72,800 |
September revenue fell by 195,500 - 200,000 = $4,500. Cost of goods sold rose by 122,700 - 120,000 = $2,700. Gross profit therefore changed by -4,500 - 2,700 = -$7,200, matching 72,800 - 80,000.
Review the cost of goods sold definition before interpreting the increase. A change in cost classification can look like a commercial deterioration even when the underlying spending has not changed.
Use one explicit calculation order: value the unit change at the old price, then value the price change at current units. This assigns the interaction between price and units to the price effect.
Revenue quantity effect = (Current units - Prior units) × Prior price
Revenue price effect = Current units × (Current price - Prior price)
Apply the same order to cost, substituting unit cost for price. The resulting bridge is exact for the rows above.
| Product | Revenue quantity effect (USD) | Revenue price effect (USD) | COGS quantity effect (USD) | COGS unit-cost effect (USD) |
|---|---|---|---|---|
| Standard | -10,000 | -4,500 | -6,000 | 2,700 |
| Premium | 10,000 | 0 | 6,000 | 0 |
| Total | 0 | -4,500 | 0 | 2,700 |
For Standard, (90 - 100) × 1,000 = -$10,000 of revenue quantity effect. Its price effect is 90 × (950 - 1,000) = -$4,500. Together they equal 85,500 - 100,000 = -$14,500.
Premium's quantity growth offsets Standard's revenue quantity loss in dollars. Across the company, the net revenue bridge is 0 - 4,500 = -$4,500; the cost bridge is 0 + 2,700 = $2,700. The gross profit bridge is therefore -4,500 - 2,700 = -$7,200. All dollar results are exact; no rounding is needed.
This is a product-level quantity and price bridge, not a separate mix decomposition. Keep the calculation order consistent; price-volume-mix analysis covers the broader method.
The recorded price decrease and unit-cost increase explain the net gross profit decline under this bridge. They do not show why Standard sold fewer units, whether discounts changed demand or whether a supplier increase was unavoidable.
Even a clear relationship needs care. NIST's scatterplot guidance explains that an association does not prove cause and effect. A comparison of discounts and sales could reflect customer selection, product availability or timing as well as a pricing response.
Finance owns the reconciliation and cost definitions. Sales investigates approved prices and credits. Procurement or operations checks cost and fulfillment records. The next review should distinguish verified transactions, supported explanations and unresolved questions, with an owner for each action.
The same approach applies to manufacturing business analysis, provided quantities, products and costs use comparable definitions. Change the business context, then rebuild the bridge from its actual records.
Upload your sales and matching cost files: Covirage maps the columns, and its tools compute and check the segment and driver comparisons supported by those records. The external AI model explains the results with every figure traced to its rows. Create a free Covirage account and bring a sample of your own data.
See self-service analytics for the supporting investigation workflow.
A finance team sees gross profit fall and separates the movement into product quantity, realized price and unit cost effects. The contribution of each effect must reconcile to the reported profit change. Managers then investigate the transactions and business circumstances behind the largest contributors.
Common techniques include period comparisons, segmentation, transaction inspection, price and quantity bridges, timing checks and association analysis. Choose the technique that matches the question and available fields. A method is useful when its result is reproducible and its limitations are explicit.
Usually it identifies measured contributors and relationships rather than proving causation. A discount recorded on an invoice can explain a lower realized price mathematically, but establishing why the discount was offered or how it affected demand requires additional business evidence.
The metric owner checks definitions and reconciliations, while the operational owner investigates the business records behind the change. Finance may own the revenue and cost totals, with sales or operations responsible for verifying pricing decisions, fulfillment timing and the resulting actions.