Platform Overhaul Before Peak Season: What Conrad, Douglas and Redcare Are Changing in Their Data Structures
Six weeks for the retail-media overhaul at Jumia, 200% scaling at Redcare – the Mirakl summer playlist shows which levers still work before Q4.
August is the last window for platform operators and sellers to close structural gaps before Black Friday traffic exposes every weakness in the catalog. The Mirakl Summer Playlist bundles four formats recorded in the first half of 2026 at K5, OMR and the Mirakl Summit. The common denominator: it is not the AI strategy that decides, but the operational excellence beneath it.
What the Overhaul Costs – and What It Delivers
The most concrete cost-benefit proof comes from Jumia. The African marketplace operator swapped its legacy retail-media system for AI-powered Sponsored Product Ads. Rollout across eight countries, duration six weeks. The result, according to the vendor: +28% ad revenue, +12% seller GMV and the strongest Black Friday in company history. Jumia does not disclose the migration costs, but the six-week timeframe provides a benchmark for staffing and project budgets.
At the same time, Havas media chief Alena Beutler-Thönes cites a conversion metric that shifts budget decisions: traffic from AI-powered searches converts at 14%, classic search at 3%. Anyone still managing retail media by cost-per-click is leaving margin on the table.
| Initiative | Timeframe | Reported Effect | Source |
|---|---|---|---|
| Legacy to AI retail media (8 countries) | 6 weeks | +28% ad revenue, +12% seller GMV | Jumia (vendor) |
| AI search vs. classic search | – | 14% vs. 3% conversion | Havas |
| Campaign automation rate | – | 95% fully automated | Mirakl Ads (vendor) |
| Marketplace scaling, regulated market | – | +200% | Redcare Pharmacy |
What Breaks in Migration
Conrad CEO Ralf Bühler, Douglas e-commerce chief Marcel Uphues and Redcare CCO Dirk Brüse describe three different breaking points at the Mirakl Summit 2026:
- Conrad: B2B process complexity – article master data with many variants, customer-specific prices, approval workflows. Clean data structures are a prerequisite here so that AI agents can order autonomously later.
- Douglas: brand protection in a curated assortment. The marketplace must not dilute the brand world. Data quality becomes the gatekeeper for seller onboarding.
- Redcare: regulation (pharmacy law, prescription requirements). The 200% scaling was only possible because compliance checks were integrated into the data flow, not handled downstream.
Limango (22 million members) and Benuta (seller brand) confirm at the K5 Masterclass: those who cannot keep up today fail not because of missing AI, but because of unsolved basics – margin logic, data quality, multichannel complexity. Operators and sellers see the same peak season from two sides; both make their decisions in the weeks before Q4.
What the Switch Requires Internally
Hood founder Steffen Röling describes the shift from rigid WKZ business (advertising cost allowance) to dynamic self-service. Organizationally, that means: online marketing and retail media are merged, and sales teams no longer have to set up seller campaigns manually but instead manage the self-service platform. Limango CEO Lennard Kressin has gone through the full cycle: introduce the market standard, build in-house, then deliberately return to Sponsored Product Ads because the business model dictates the logic.
Moritz Zaudtke (Mirakl Ads) puts the automation rate at 95%: 95% of all campaigns already run automated. For the internal switch, that means: fewer campaign managers, more data analysts who maintain rule sets and manage exceptions.
Realistic Timeline to Black Friday
From today, it is a good three months until Black Friday week. The Jumia case study shows that a retail-media system switch is feasible in six weeks – provided the data structures are right. That is exactly where many are struggling: Conrad, Douglas and Redcare are currently investing in catalog cleanup, attribute completion and interfaces to ERP and PIM. Anyone who does not do this homework in August will scale not revenue in November, but the error rate.
Decision-makers should check: are product data available in a form that AI agents can read them (14% conversion)? Is the retail-media infrastructure switched to self-service so that Q4 brand budgets can flow directly in September? Are 95% of campaigns running automated, or is the team still tied up with manual WKZ processing?
The four formats of the summer playlist should be understood as working material – not as inspiration, but as a checklist for the next twelve weeks.