Data warehouse investigations
Platform dashboards make claims. A warehouse lets you test them against orders, refunds, repeat purchase, margin and customer cohorts.
Published: 31 May 2026 | Updated: 31 May 2026
The principle
Treat every platform report as a claim until it survives contact with business data.
Google, Microsoft, Meta and Klaviyo can all claim influence over the same customer. That is not fraud. It is how attribution windows and channel incentives work. The warehouse job is to separate useful signal from duplicated credit.
Common investigations
- Which campaigns drive first orders that produce repeat customers?
- Which products look weak on first-order ROAS but strong on LTV?
- Which channel claims disappear when matched against transaction IDs?
- Where do Meta, Google and email overlap on the same orders?
- Which conversions are inflated by refunds, cancellations or duplicate tracking?
- Which markets, brands or product groups deserve different CAC targets?
Minimum warehouse inputs
InputWhy it mattersTypical source
Ad spendCalculates CAC, ROAS and MERGoogle Ads, Microsoft Ads, Meta
TransactionsGives the order-level source of truthEcommerce platform or ERP
Order linesEnables product-level LTV analysisEcommerce order export
Email revenueShows lifecycle and retention claimsKlaviyo or email platform
RefundsStops bad revenue from looking profitablePayment processor or ecommerce platform
Customer IDsConnects first purchase to future valueCRM or ecommerce customer table
The output
A good investigation ends with an action, not a prettier dashboard. The answer should change bids, budgets, campaign structure, product focus, landing pages or conversion feedback.