02
Retail and Consumer Goods
Enterprise AI · Personalization · Demand intelligence · Consumer credit decision
III. Sectors · 02
02
Enterprise AI · Personalization · Demand intelligence · Consumer credit decision
Support activities
Primary activities
In retail, customer preference is earned with every interaction and can be lost in just one. Tight margins, consumers who compare options in seconds, and an expanding number of channels demand faster, more precise decisions across the entire value chain.
Artificial intelligence drives value across five key areas:
The result is a retail business that makes faster decisions, minimizes errors, and turns every sale into the beginning of the next.

Casas Bahia
2020–2021
ChallengeMoving from transactional retail to a relationship platform meant prioritizing each customer by their value over time.
SolutionStudy of CLV modeling to prioritize customers and experience, geared to conversion, retention and basket size.

Casas Bahia
2020–2021
ChallengeCross-selling in e-commerce was low, and salespeople had no support to choose the right offer for each customer.
SolutionPredictive modeling and customer segmentation, with purchase history and pre-approved credit as the basis of the next best offer in Via+; e-commerce cross-selling grew 20%.

Casas Bahia
2020
ChallengeConverted orders were declined along the RPA, gateway, acquirer, issuer and anti-fraud chain, with no clarity on where the bottlenecks were.
SolutionStudy of the end-to-end approval flow, from order to automated and manual anti-fraud, mapping declines, acquirer retries and order recovery.

Casas Bahia
2020–2021
ChallengeFraud in e-commerce and the marketplace caused losses in an order flow running through RPA, gateway, acquirer and issuer.
SolutionUser authentication, address and payment checks, purchase limits, a transaction fraud score, device fingerprinting, velocity and link analysis, with automated anti-fraud and manual review.

Casas Bahia
2020–2021
ChallengeThe drivers of dissatisfaction in the digital journey were poorly understood.
SolutionStudy of root causes combining NPS with text analytics of complaints on Reclame Aqui, a Brazilian consumer complaints site.

Casas Bahia
2020–2021
ChallengeMarketing campaigns needed more engagement and better returns.
SolutionMachine learning to personalize and target campaign audiences, with propensity models and remarketing; customer engagement grew 15%.

Casas Bahia
2020
ChallengePartner sellers and online vendors, on WhatsApp and other platforms, with offending behavior exposed the marketplace to fraud.
SolutionStudy of a seller approval model built on record data, transactions by payment method, customer profiles, products and service history, confirmed by the Fraud Prevention team.

Casas Bahia
2020–2021
ChallengeThe group's digital financial operation required risk and fraud control.
SolutionRisk management and fraud prevention at banQi, with data integration and automated analysis and mitigation.

Burger King
Grupo Boticário · GAVB · 2021
ChallengeOverlapping delivery coverage and revenue-linked rent made it hard to weigh every combination to cut occupancy costs without hurting operations and experience.
SolutionInteger linear programming that recommends switching delivery on or off at each company-owned store, with sales absorbed by nearby stores, operational impact and a visual simulator.

GPA
Grupo Boticário · GAVB · 2021–2022
ChallengeRecommendations in the Cliente Mais and Clube Extra loyalty apps relied on a vendor due to be replaced by the third quarter, with little autonomy and off-strategy offers.
SolutionA four-wave pipeline of next best offer, basket-based cross-selling, in-category upselling and private labels, with final allocation by business rules; conversion reached 65% and 57% in the first releases.

Grupo Boticário
GAVB · 2021
ChallengeFraud prevention teams depended, at a cost, on an outside vendor's credit score.
SolutionMVP of an in-house credit score, mapping flows and challenges with the Grupo Boticário team and organizing the data.

Quero-Quero
Grupo Boticário · GAVB · 2021
ChallengeEnd customers had to receive the product of greatest interest, in line with Quero-Quero's commercial strategy.
SolutionProposal for testing recommendation paradigms and purchase propensity models, with dashboards, documentation and handover, in a fixed ten-week scope.

Tenda
Grupo Boticário · GAVB · 2021
ChallengeThe Member Get Member program ran inefficiently, with manual lead handling and little use of registration data to understand buying behavior.
SolutionPropensity models for who refers and who converts, profile and persona segmentation, and referral and conversion simulators, in a fixed eight-sprint scope.

Zé Delivery
Grupo Boticário · GAVB · 2022
ChallengeZé Delivery needed specialized data and AI professionals for its initiatives.
SolutionProposal for outsourcing data scientists and engineers, MLOps, an architect, BI, QA and DevOps, under a minimum four-month contract.

St Marche
Grupo Boticário · GAVB · 2022
ChallengeA commercial approach with no problem defined in the material, to bring data and analytics to grocery retail.
SolutionProposal for data and analytics with recommendation cases in grocery retail, CRM and segmentation.

St Marche
Grupo Boticário · GAVB · 2022–2023
ChallengeStores suffered stockouts and weak assortment decisions.
SolutionProduct range classification, safety stock forecasting and complementary product recommendations, with AI-based store clustering and seasonality analysis.

Ambev
Grupo Boticário · GAVB · 2023
ChallengeSupplier delays and short deliveries were detected too late.
SolutionThe propensity of each supplier to deliver late or short, so operators can act ahead of time.

C&A
Grupo Boticário · GAVB · 2023
ChallengeMany visitors to the Morumbi and Tamboré stores left without buying, and their profile was unknown.
SolutionProposal for a fixed-scope MVP: in-store journey, personas centered on non-buyers, a look-alike model and a relationship cadence to monetize the customer.

Grupo Boticário
GAVB · 2023
ChallengeFranchisee uptake of additional services had to grow, counting a cross-sell when a franchisee buys at least one of four services.
SolutionFranchisee clusters with their own journey, purchase propensity and best offer timing, and a second phase with half-yearly recalculation, a control group, approach tests and a CRM cadence.

Leroy Merlin
Grupo Boticário · GAVB · 2023
ChallengeCustomers and staff could not identify the toilet model or the matching seat, causing wrong purchases and dissatisfaction.
SolutionMVP of a deep learning model that identifies the product from a photo sent through the app, with data augmentation for scarce images and heat maps explaining each answer.

Leroy Merlin
Grupo Boticário · GAVB · 2023
ChallengeThe CRM team lacked accurate segmentation for targeted actions across the customer base.
SolutionMVP with eleven personas by channel, ticket, frequency and department, simulating the gain from lifting customers below their persona's average, and a relationship cadence per persona.

Lojas Renner
Grupo Boticário · GAVB · 2023
ChallengeA commercial approach with no problem defined in the material, to bring AI to fashion retail.
SolutionProposal for a portfolio of recommendation, fashion CRM analytics and retail segmentation cases.

Lojas Renner
Grupo Boticário · GAVB · 2023
ChallengeCustomer data was scattered across brands and channels, with no single view or scalable segmentation to personalize communication within Brazil's data-protection law (LGPD).
SolutionA golden record by brand and corporate, a 360° customer view and AI personas in relationship cadences and remarketing; email campaigns brought in 30% more revenue across six brands and over 7 million customers.

Grupo Boticário
GAVB · 2021
ChallengeRising complaints strained relationships with resellers and consumers, exposing the brand to consumer agencies, social media and the press, with no way to spot the critical profiles for specialized desks.
SolutionComplaint score by recency, frequency and channel severity, in nine segments; three times a week, the IVR routes resellers with a recent complaint to a dedicated desk.

Grupo Boticário
GAVB · 2021
ChallengeCRM could not tell whether each purchase was a gift or for personal use, limiting journeys, kits and personas, and the previous model could no longer be retrained.
SolutionA classifier trained on post-purchase survey labels and sales, product, customer and holiday signals, with models for stores and e-commerce and accuracy monitoring.

Grupo Boticário
GAVB · 2021
ChallengeCRM and e-commerce storefronts of three brands had to recommend new products, repurchases and similar items to each consumer.
SolutionImplicit-feedback collaborative filtering on store and e-commerce history, delivering new products, repurchases and item-to-item similars, with metrics logged every cycle.

Grupo Boticário
GAVB · 2021
ChallengeWith no defined churn concept, Retention could not prioritize active customers by their likelihood of buying again.
SolutionLogistic regression on recency, frequency and value, category mix, seasonality, region and loyalty, in propensity bands that set the discount in the anti-churn journey.

Grupo Boticário
GAVB · 2021
ChallengeConsumers who struggled to choose perfumes needed recommendations based on their olfactory preferences.
SolutionA fragrance recommendation API based on product-to-product similarity and item ranking, with automated cloud deployment.

Grupo Boticário
GAVB · 2021
ChallengeMeasurement required homogeneous test and control groups, and the business needed to understand buying profiles for sharper campaigns.
SolutionClusters for store-only, e-commerce-only (single or repeat buyers) and omnichannel shoppers, with variables picked by importance and a name for each group.

Grupo Boticário
GAVB · 2021
ChallengeTechnology teams needed self-service to create and remove components, access and monitoring, with consistent security and design standards.
SolutionA developer portal with guided scaffolding and cost estimates, access and component-removal APIs, maturity and security badges, automated monitoring and a design system.

Grupo Boticário
GAVB · 2021
ChallengeCreating, continuously delivering and governing the group's applications and libraries needed standardization.
SolutionThe Alquimia developer platform, with language templates, delivery pipelines with DevSecOps, a library for the group's services and code discussion notifications.

Grupo Boticário
GAVB · 2021
ChallengeTracking cosmetics and data and AI patents published worldwide was manual and slow for R&D.
SolutionBots that query the European Patent Office by cosmetics codes and keywords, collect patent families and images, and produce reports ranking the main players.

Grupo Boticário
GAVB · 2021
ChallengeData teams needed shared standards for pipelines, SQL versioning, reference architectures and event contracts, without rework or inconsistency.
SolutionIngestion DataOps with versioned SQL, orchestration and continuous delivery by environment, reference architectures and data leadership ADRs, and an event schema registry.

Grupo Boticário
GAVB · 2021
ChallengeDeploying workflows, functions and data warehouse layers happened with no standards or validation.
SolutionContinuous delivery pipelines for orchestrated workflows, cloud functions and SQL procedures in the curated layers, with structure and naming checks and a documented architecture.

Grupo Boticário
GAVB · 2021
ChallengeEach team built ML pipelines its own way, making it hard to productize, ensure data quality, monitor and document models.
SolutionAn internal library of pipeline components, covering tables, data quality and alerts, and a repository template with code structure, continuous delivery and business and technical documentation.

Grupo Boticário
GAVB · 2021
ChallengeDirect Selling planning needed to estimate, every cycle, each reseller's chance of buying each category and brand to target offers.
SolutionA multi-target propensity model by category, brand and reference cycle, which went through four versions until it ran orchestrated in the cloud.

Grupo Boticário
GAVB · 2021
ChallengeResellers needed suggestions of new and similar products in the app, drawing on millions of customers and thousands of products with sparse data.
SolutionImplicit collaborative filtering through matrix factorization, with per-person recommendations, similar products by item, weekly removal of items already bought and app integration.

Grupo Boticário
GAVB · 2022
ChallengeCognitive analyses depended on transcribing audio.
SolutionMVP of an audio transcription pipeline, built from the standard MLOps template.

Grupo Boticário
GAVB · 2022
ChallengeConsumer models kept rebuilding the same features across brands and channels.
SolutionA pipeline that merges the single consumer base, store and e-commerce receipts and sales into features by consumer and product for five brands, with data quality checks.

Grupo Boticário
GAVB · 2022
ChallengeRetention and reactivation journeys depended on anticipating which customers would stop buying, and individual lifetime value required that probability.
SolutionLogistic regression on recency, frequency, ticket and one-time buying, with probability bands for active and inactive customers, drift monitoring, alerts and a monthly run.

Grupo Boticário
GAVB · 2022
ChallengeHundreds of macroeconomic indicators affect the cost breakdown and the budget, and the market tool gave inconsistent forecasts for supplier negotiations.
SolutionA model tournament per indicator, with configurable lags and hyperparameter search, picking the lowest-error model and forecasting twelve months and the monthly cost change.

Grupo Boticário
GAVB · 2022
ChallengeExpansion had to decide where and in which format to open stores and kiosks, maximizing revenue while controlling cannibalization between stores.
SolutionMVP of mixed-integer optimization with a gravity attraction model, a scenario simulator and maps, choosing the location and format of each new point.

Grupo Boticário
GAVB · 2022
ChallengeSetting each item's target inventory had to account for inventory and stockout history.
SolutionStudy of target inventory by item and cycle, combining inventory and stockout data and applying stockout limits.

Grupo Boticário
GAVB · 2022
ChallengeThe brand had to forecast sell-out by product and cycle across stores, e-commerce and direct selling, accounting for campaigns, virtual kits and partner brands.
SolutionA forecasting model with hierarchical normalization, seasonality, campaigns and trend correction, balancing bias and error, plus an app that prepares data for planning.

Grupo Boticário
GAVB · 2022
ChallengeDemand planning needed standardized forecasts of sell-out and sell-in sales and purchases by brand and channel.
SolutionA single repository of explainable forecasting models with campaign, price and active-material features per cycle, data validation and performance metrics for a dashboard.

Grupo Boticário
GAVB · 2022
ChallengeCollecting external web data had no standard.
SolutionAn internal, configurable collection library with automated browsing, proxies, state recovery and cloud storage, published to the private package repository.

Grupo Boticário
GAVB · 2022
ChallengeTeams kept rewriting computer vision utilities without a standard.
SolutionAn internal computer vision library published to the private package repository, with continuous quality and security checks.

Grupo Boticário
GAVB · 2022
ChallengeThe group's ML platform costs were growing with no visibility by pipeline.
SolutionPipelines that collect metadata from each ML run and a periodic cost report for the teams.

Grupo Boticário
GAVB · 2022
ChallengeAssessing portfolio health required each product's revenue trend free of promotional effects and a single base of attributes and indicators.
SolutionA baseline by product group with anomaly detection to discard promoted cycles, and a consolidated base of sell-out, stockouts, new resellers and generative-AI attributes.

Grupo Boticário
GAVB · 2022
ChallengeEach cycle's promotional levers were judged against a simple median, unable to estimate what would have sold without the mass promotion and its hangover.
SolutionTime series by product group, state and channel, with open stores, relative price, hangover and holidays, estimating baseline, hangover, margin and incremental sales.

Grupo Boticário
GAVB · 2022
ChallengeDeciding what to promote, when and at what discount across the year was manual, rule-heavy and with no guarantee of the best result.
SolutionMathematical optimization over forecast scenarios, with business rules as constraints, in short-term and full-year versions, triggered from a spreadsheet and with dashboards.

Grupo Boticário
GAVB · 2022
ChallengePromotion planning required forecasting volume, revenue and margin by product, region and channel under different prices and mechanics, including cannibalization among similar items.
SolutionModels trained in parallel, a grid of price and mechanic scenarios for consumers and resellers with forecast KPIs, and a ranking of cannibalizers by similarity.

Grupo Boticário
GAVB · 2022
ChallengeFulfilling franchisee orders required balancing the use of factory inventory, from preparation to shipping orders.
SolutionLinear and coverage-based allocation and minimum-cost optimization by product and distribution center, with coverage targets, scheduled or on-demand runs and daily consolidation.

Grupo Boticário
GAVB · 2022
ChallengeThere was no understanding of the moments in a reseller's journey, from acquisition to churn, or of the niches most affected by it.
SolutionStudy of the reseller journey, with exploratory profile analysis and proposed churn and activation cases, such as prioritizing first-order contacts and activating via WhatsApp.

Grupo Boticário
GAVB · 2022
ChallengeReseller sign-ups were prioritized only by sign-up time, ignoring the chance of becoming an active reseller.
SolutionA classifier of the chance of a first order within three days, in five bands combined with sign-up time and with no gender variables, on ethical grounds; conversion rose 2 p.p. in a two-week case-control pilot.

Grupo Boticário
GAVB · 2023
ChallengeThe Eudora reseller service center received many record-data questions, such as status, address, password and reactivation, that could be automated.
SolutionMVP of a generative AI assistant with answers by topic, session history and human handoff, plus a validator that classifies, summarizes and evaluates each interaction.

Grupo Boticário
GAVB · 2023
ChallengeMass communication and single-brand recommendation missed each consumer's interests across the four brands: they only got offers from brands they already bought.
SolutionCollaborative filtering on a consumer-by-brand-and-subcategory matrix, weighting store and e-commerce purchases, site and app interactions and CRM engagement, plus business rules.

Grupo Boticário
GAVB · 2023
ChallengeTurning survey answers and Q&A into announcements and news required manual writing.
SolutionMVP with an LLM that drafts titled news from questions and answers, with a conversational interface and a vector base.

Grupo Boticário
GAVB · 2023
ChallengeFranchisees raised many repetitive operational questions, answered by hand from FAQs and the knowledge base.
SolutionMVP of RAG over the FAQs, evolved into a backend with conversation history and token and cost control, on the corporate LLM SDK.

Grupo Boticário
GAVB · 2023
ChallengeExpansion needed to estimate a new store's average revenue by city, focusing on smaller cities.
SolutionA forecasting model using last year's store data, socio-demographic variables, traffic generators and IBGE's urban network data, run on demand.

Grupo Boticário
GAVB · 2023
ChallengeThe complaint base was too small to train the subcategory model.
SolutionMVP that generates fictitious complaints with generative AI to feed the subcategory model.

Grupo Boticário
GAVB · 2023
ChallengeEach data science team rebuilt features in silos, with no consistency between training and inference, no governance and no reuse.
SolutionFeature groups by value stream, native and custom transformations, label generation, offline and online serving and drift detection, with scheduled pipelines.

Grupo Boticário
GAVB · 2023
ChallengeTeams needed to experiment with LLMs under usage, cost and compliance governance, without scattered access keys.
SolutionA single generative AI API with team and global quotas, chat and embedding endpoints and ready-made examples for the internal hackathon.

Grupo Boticário
GAVB · 2023
ChallengeTeams using LLMs integrated directly with several providers, with no standard for authentication, governance or usage control.
SolutionA gateway with adapters for more than one provider and an internal SDK, with a model catalog, file upload and LLM framework integration.

Grupo Boticário
GAVB · 2023
ChallengeKeeping, boosting or discontinuing products depended on reading each item's life stage by hand.
SolutionA score that classifies each product, unit and channel as diffusing, maturing, declining or post-mortem, combining ABC curves, customer and revenue growth and trend.

Grupo Boticário
GAVB · 2023
ChallengeFranchise smart purchasing depended on forecasting sell-out by store and product; without it, purchase suggestions caused stockouts or overstocking.
SolutionA multi-brand forecasting pipeline by business unit, handling new stores, with a reliability layer, tracking indicators and data quality checks.

Grupo Boticário
GAVB · 2023
ChallengeDiscontinuing products was slow, keeping a portfolio less lean, efficient and aligned with the group's strategy.
SolutionAn index that prioritizes products for discontinuation, validated with the business in a dashboard and run in production, with development, staging and production environments.

Grupo Boticário
GAVB · 2023
ChallengePortfolio analysts needed to simulate order fulfillment before running the allocation.
SolutionAn analytical order base and on-demand allocation simulation for analysts, with the result sent back to the order management system.

Grupo Boticário
GAVB · 2023
ChallengeReplenishment for four businesses used fixed coverage by store and product, ignoring margin and demand variability, causing excess, shortages and rush orders.
SolutionStudy of safety stock with service levels set by supply chain segmentation, converted into days of coverage and applied to the sell-out forecast, net of inventory and orders.

Grupo Boticário
GAVB · 2023
ChallengeDirect Selling CRM wanted to shift from mass to segmented promotions, without knowing which resellers could grow most in each brand, category and line.
SolutionAn analysis of purchases over six cycles by loyalty segment and portfolio, sorting resellers into three leverage groups, with backtests and a revenue ranking.

Grupo Boticário
GAVB · 2023
ChallengeThe O.U.I brand needed wider penetration among resellers, identifying those most likely to make a first purchase.
SolutionStudy of propensity comparing buyers and non-buyers, using a decision tree to select the audience.

Grupo Boticário
GAVB · 2023
ChallengeDirect Selling CRM needed to group resellers by recency, frequency and value to target productivity and activity actions.
SolutionRule-based RFM segmentation over a six-cycle window, with VIP, high, medium and low value bands, recalculated every cycle for the active base.

Grupo Boticário
GAVB · 2023
ChallengeSkincare consumers needed recommendations for products and facial care routines.
SolutionMVP of facial recommendation, started from the standard MLOps template.

Grupo Boticário
GAVB · 2023
ChallengeConsumers did not know which products and routines to choose for their own skin's needs.
SolutionMVP that captures the face, measures the skin through facial analysis and uses LLM agents to diagnose in plain language and recommend routines and products, such as the Botik line.

Grupo Boticário
GAVB · 2023
ChallengeThe effect of digital store missions on customers could not be measured without a daily control group and a consistent round history.
SolutionA daily pipeline that checks the source, draws and blocks control customers and logs the history of blocks and missions for each round.

Grupo Boticário
GAVB · 2023
ChallengeThe promotional grid changed every cycle of about three weeks and was the same for everyone, with many low-relevance offers for consumers.
SolutionHybrid recommendation combining next best offer, catalog similarity and two years of history, suggesting each customer's twenty promoted products per cycle.

Grupo Boticário
GAVB · 2023
ChallengeBoleto Turbinado offers were generic, ignoring each loyalty customer's profile in stores.
SolutionNext best offer recommendations crossed with the customer profile and product similarity, generating personalized items per customer for three brands, with data quality checks.

Grupo Boticário
GAVB · 2023
ChallengeData teams needed to learn how to build applications with LLMs.
SolutionStudy of LLM applications in a hands-on workshop: marketing titles, a shopping chatbot, memory, structured outputs and hallucination control.

Grupo Boticário
GAVB · 2023
ChallengeWithout anticipating which resellers would buy in the opening cycle, activation actions had no direction.
SolutionStudy of activation propensity in the reference cycle, structured from the standard data science template.

Grupo Boticário
GAVB · 2023
ChallengeThe storefront for resellers yet to make a first purchase was built by hand from a CRM dashboard, which limited its scale.
SolutionA ranking of the thirty best-selling first-purchase products over the last three cycles, updated weekly and built into the app.