Illustrative example. This case study describes a typical engagement of this kind, not a specific client. Figures are representative of the work, not a guarantee of results.
Data and analytics · Retail
One version of margin for a direct-to-consumer homewares brand
A fast-growing online retailer had five systems, three versions of revenue and a month-end process that took a week. A small, well-tested data platform gave everyone the same numbers, and showed which products were quietly losing money.
Results
Measured over the first two month-end cycles.
The problem
Orders lived in Shopify, accounts in Xero, email marketing in Klaviyo, advertising across Meta and Google, and fulfilment data in CSV files from a third-party logistics provider. The finance manager spent the first week of every month exporting, pasting and reconciling.
Marketing reported return on ad spend from the ad platforms, finance reported margin from Xero, and the numbers never matched. Leadership meetings started with an argument about whose figure was right.
Nobody could answer the question the founders cared about most: after shipping, payment fees, discounts and returns, which products actually made money?
What we did
A one-week audit, then a six-week build in the client's own cloud accounts.
Week 1: Audit
We reviewed each system and the existing reports, and ran a workshop with finance and marketing to agree written definitions for twelve core metrics, starting with revenue and contribution margin.
Weeks 2–4: Foundation
Automated pipelines from all five sources into a cloud data warehouse, with dbt models that allocate shipping, payment fees and returns to individual orders. Automated tests catch missing or duplicated data before it reaches a report.
Weeks 5–7: Dashboards
A leadership dashboard, a month-end finance pack and a marketing view that reconciles to finance. Each metric links to its written definition.
Handover
Documentation, recorded walkthroughs and two training sessions for the finance manager and a marketing analyst, who now add new reports themselves.
Stack
What made the difference
The most valuable deliverable was a one-page list of metric definitions, agreed before any code was written. The platform simply made those definitions automatic.
Allocating costs to individual orders, rather than averaging them, is what revealed the unprofitable products. Most were heavy, low-priced items with high return rates.
Choosing tools the team could maintain mattered more than choosing the most powerful ones. The running cost of the platform is a small monthly figure, and no new hire was needed.