Customer analytics: turn customer data into business decisions

Customer analytics turns recorded purchases into decisions about account priorities, buying frequency, and product mix. Build analytics customer teams can verify by agreeing on the customer identifier, transaction grain, and reporting period before comparing accounts.

What customer analytics measures

A useful customer analysis connects an observable action to a business question. Revenue identifies the accounts contributing sales. Purchase frequency distinguishes repeated ordering from an isolated purchase. Product mix shows which categories an account already buys and which categories need a conversation.

These measures describe recorded business with your company. They do not reveal everything a customer buys elsewhere, why an account made a purchase, or whether an additional sale is available. Keep those questions separate from the arithmetic.

The starting point is a sales export with customer ID, order ID, date, category, and net sales. Additional customer analytics data sources can extend the analysis once their identifiers and dates match. Sales operations owns the customer mapping; finance confirms the sales definition and reporting total.

The formulas and their denominators

Revenue per purchasing customer = Net sales / Distinct purchasing customers

Purchase frequency = Distinct orders / Distinct purchasing customers

Average order value = Net sales / Distinct orders

Repeat-purchaser share = Customers with at least two orders / Distinct purchasing customers

Category revenue share = Category net sales / Total net sales

Use purchasing customers as the denominator for these measures. An account master containing inactive accounts answers a different question. State whether orders are placed, shipped, or invoiced, and keep that definition throughout the report.

For this example, each row is a completed order containing a single category. A real line-item export needs distinct order counting because an order can occupy several rows. Net sales exclude sales tax and include the discounts and returns belonging to the reporting scope.

A worked sales export

An illustrative US industrial supplier reviews January 1 through March 31, 2026. All amounts are USD. Put this table in columns A:E, with the first order on row 2. Dates must be stored as dates, and the export contains only the reporting period.

Create a customer summary in G:I. Enter the distinct customer IDs in G2:G5. In Google Sheets, enter revenue in H2 and orders in I2, then fill down:

=SUMIFS($E$2:$E$9,$B$2:$B$9,G2)
=COUNTIFS($B$2:$B$9,G2)

Google documents SUMIFS as summing values that meet specified criteria. Its COUNTIFS documentation explains the corresponding conditional count; here counting rows equals counting orders because the order IDs are unique.

The purchasing-customer count is four, taken from the distinct IDs. The next table works each measure from those rows. Percentages are shown to one decimal place; no intermediate calculation is rounded.

The checks that prove the report

Customer revenue must equal export revenue: $3,000 + $1,500 + $700 + $800 = $6,000. Category revenue must also equal that amount: $3,900 + $2,100 = $6,000. Compare the export total with finance's sales report for the same dates and exclusions.

The measures also satisfy an identity:

Customers × Purchase frequency × Average order value = Net sales

Here, 4 × 2 × $750 = $6,000. This identity checks arithmetic consistency; duplicate orders could still inflate every input together. Check order IDs separately before approving the report.

Turn the comparison into a decision

Customer A supplies half the revenue, so the account owner should review continuity of that relationship. Customer D orders repeatedly but buys only Safety. A discussion about Packaging could be useful, provided its needs support that conversation. Customer C purchased once; that alone does not establish a retention problem.

For a structured comparison, segment customers from ledger evidence and define the comparison group. Keep account size and observation time consistent. Retail customer analytics applies the questions to shopper records and store context.

Where it goes wrong

• Names used as identifiers. Spelling changes split one customer into several accounts. Reconcile stable IDs before grouping.
• Line counts called orders. Multiple lines on an invoice inflate frequency unless you count distinct orders.
• Unequal observation periods. A recently acquired customer has had less time to purchase again. Compare equally observed groups.
• Sales described as profit. Revenue alone contains no cost evidence. Add matching cost records before making a margin claim.
• Anonymous transactions treated as customers. An unidentified receipt supports basket analysis but cannot prove an individual buying history.

Covirage is not affiliated with Google.

Analyze your own customer file in Covirage

Bring a sample of your own sales export or Google Sheet with customer IDs, dates, order IDs, categories, and net sales. Covirage maps the columns, its tools compute and check the comparisons, and the external AI model explains the results with every figure traced to its rows. Create a free Covirage account, then set up your business and bring your file.

See sales insights for the account questions these comparisons support.