Lojas Renner, Brazil’s largest fashion retailer, grew retail revenue 9.2% in 2025 while reducing total inventory 3% and cutting aged stock 16%. The Q4 2025 earnings make this the most detailed operational dataset from a Latin American retailer running AI-driven inventory allocation at scale. Both figures trace to a system Renner began piloting in 2019, now operating across 650 stores in a retail market worth $231 billion. No other retailer in the region has published comparable before-and-after metrics from an AI allocation deployment at this scope.

The pilot was narrow by design. In the second half of 2019, Renner started routing 8.5% of its merchandise to stores using AI models that forecast demand by individual item, color, size, and geographic location. Those algorithmically allocated products sold 12% better and needed 18% less inventory than conventionally distributed goods, as reported by TI Inside. In-store service levels — the rate at which a customer finds the right size and color on the rack — jumped from 78% to 93% in that same half-year. By 2020, the target was 17% of total assortment under algorithmic allocation, with projected gains of 5% additional sales on 10% less stock.

Then Renner built the infrastructure to scale it. Between 2021 and mid-2024, the company invested R$3.3 billion in technology and innovation — its largest-ever capital cycle. Every product across 650 stores in Brazil, Argentina, and Uruguay now carries an RFID tag, generating 4.8 million in-store reads daily and enabling monthly inventory counts completed in hours rather than the annual counts they replaced. Stock-outs fell 87% and inventory accuracy improved 64% after the RFID rollout, per Sensormatic. The company expanded from 16 to 54 algorithmic store clusters, each receiving assortments calibrated to local demand signals. Production lead time dropped from 95 days to 30.

By full-year 2025, the compound effect reached the income statement. Gross margin hit 56.1%, Renner’s highest in six years, driven by lower markdowns and stronger full-price sell-through. Sales per square meter rose to R$17,200, approximately 45% above the company’s nearest peers. Store operations productivity improved roughly 40%, and replenishment efficiency gained around 60%. Net income climbed 21.8% to a record R$1.5 billion. Products recommended by Renner’s algorithms converted at 135% higher rates on the company’s digital channels, according to CEO Fabio Faccio at South Summit Brazil.

The gap between Renner and the regional average is widening, not closing.

Brazil anchors Latin America’s largest retail market at $231 billion, yet most of the region’s retailers operate nowhere near Renner’s data maturity. An analysis of 30 major Latin American e-commerce platforms found that 80% show critical deficiencies in AI adoption: 16 of 30 returned no results for misspelled queries, only five offered personalized search, and none supported multimodal search. The average page load across those platforms was 4.94 seconds, nearly double the ideal threshold. ILIA 2025, the region’s AI readiness index, classifies only Brazil, Chile, and Uruguay as “pioneer” markets out of 19 countries surveyed. Across those 19 markets, roughly one in five organizations has not inventoried its data assets, and only half of available data is ready for AI processing.

The counter-argument writes itself: if proven AI allocation requires R$3.3 billion, 100% RFID tagging, and five years of iteration, this is a story about scale advantages, not a replicable template. This is plausible, but only if you assume the investment must precede the result. Renner’s 2019 pilot covered 8.5% of products and still delivered a 12% sales lift with 18% less stock, before most of that capital was committed. The 2020 projections forecast 5% additional sales at 17% coverage with 10% less stock, figures that implied diminishing per-SKU returns but still positive economics at every increment. The infrastructure followed the proof of concept, not the other way around.

Faccio told attendees at NRF 2026 in January that the AI hype phase is over, replaced by what he called pragmatic application. Renner now targets 40% in-season production and 9-13% annual revenue growth through 2030, with its Cabreúva distribution center serving as the allocation engine for the entire network. Digital GMV grew 12.3% in 2025, reaching 15.5% of total sales, as the same algorithmic layer now governs both physical and online distribution. For Latin America’s other retailers, the benchmark that matters is not the R$3.3 billion total. It is the pilot that delivered at 8.5% coverage before the big checks cleared.