Client name withheld under a non-disclosure agreement.
The challenge
The group sells apparel and accessories through its own stores, a brand app and marketplace storefronts in mainland China, Hong Kong and Taiwan. Its marketing team bought media across search, social platforms, marketplace advertising and influencer campaigns, and each platform reported its own conversions. Added together, the platforms claimed more sales than the group had actually made.
Sales data sat in the point-of-sale system, the app backend and each marketplace’s seller portal. Customer records were matched by hand, when they were matched at all. Budget meetings ran on screenshots from platform dashboards, and the finance team could not say which campaigns had paid for themselves.
What we delivered
- Multi-Source Data Integration & Analytics. One customer and transaction dataset built from the point-of-sale system, the app, the marketplace seller portals, the CRM and the advertising platforms. Records are matched on consistent keys, refreshed daily, and every figure can be traced back to its source.
- Marketing Attribution & Revenue Intelligence. An attribution model that shares credit for each sale across the touchpoints that led to it, rather than counting it once per platform. The model separates sales that marketing drove from sales that would have happened anyway, and reports return by channel, campaign and market.
- Business Intelligence & Performance Analytics. Dashboards for the marketing, e-commerce and finance teams that show spend, attributed revenue and return on the same definitions, with a monthly budget review pack that reaches management automatically.
We started with the marketing team’s own spreadsheet of campaign results and agreed which questions the model had to answer before building anything. We then compared the model’s output against two quarters of past campaigns and holdout tests before any budget was moved.
Timeline
- Weeks 1 to 4: data audit, source connections and shared definitions for a sale, a customer and a touchpoint
- Weeks 5 to 11: integrated dataset, attribution model and validation against past campaigns
- Weeks 12 to 16: dashboards, budget review pack, team training and handover
Outcomes
- 15 to 25% better return on paid media spend after the first two budget cycles
- Marketing, e-commerce and finance now plan from one set of numbers
- Budget decisions moved from quarterly platform reviews to a monthly review of attributed return
