RappiPlus: From Data to Business Decisions
RappiPlus: From Data to Business Decisions
Turning messy multi-source data into a profitability, funnel, retention, and experimentation analysis for a delivery subscription service.
Turning messy multi-source data into a profitability, funnel, retention, and experimentation analysis for a delivery subscription service.
RappiPlus needed a data-driven answer to a deceptively simple question: is the business actually working? That meant reconciling three separate data sources (order transactions, a product catalog, and marketing spend) each with its own quality issues (missing values, duplicate orders, inconsistent amounts), and then going further: understanding not just whether the business was profitable, but where users were dropping off, whether they came back after signing up, and whether a recent checkout redesign actually moved the needle.
The results from this analysis are also available as an interactive Power BI dashboard.
The analysis moved through five connected stages:
Data quality. Cleaned and validated three raw datasets in Python (fixing date types, handling missing values, removing invalid and duplicate records, and standardizing categorical fields) before trusting any number downstream.
Profitability. Merged orders with product costs and marketing spend to calculate revenue, total cost, profit, margin, average order value, and marketing efficiency.
Conversion funnel. Queried a PostgreSQL database directly with SQL (via SQLAlchemy) to trace users through each stage of the purchase journey and calculate step-by-step conversion rates.
Cohort retention. Built a weekly retention analysis by registration cohort, using window functions and conditional aggregation in SQL to see how many users stayed active in the weeks after signing up.
Experimentation. Ran a two-proportion Z-test on an A/B test of a new checkout UI, to validate (statistically, not just anecdotally) whether the change affected conversion.
Across 24,906 orders from 7,640 unique users (Jan–Jun 2025), RappiPlus generated $9,610,018.94 in revenue. Product costs brought gross profit to $5,781,149.93 (60.16% margin), but once the full $2,871,843.53 in marketing spend is factored in, net profit lands at $2,909,306.40 — a 30.27% margin. That 30-point gap between gross and net margin is the single clearest signal in the data: marketing spend is absorbing roughly half of the business's raw profitability, making marketing efficiency (not just sales volume) a first-order lever for the business.
By category, Home is both the top revenue driver and the most profitable ($3.23M revenue, 62.46% margin), edging out Fashion (59.32%) and Electronics (58.66%) — a gap worth investigating given all three categories sell at roughly the same volume. At the product level, Vacuum-Pro-Black, Blender-XL-Red, Jacket-Winter-M, and Sneakers-Urban-42 each sold over 6,100 units, comfortably ahead of the electronics lineup (~4,100–4,200 units each). On the marketing side, spend splits almost evenly between social ($918K) and organic ($913.5K), with paid search close behind ($863K); though a small share of spend (~$177K) couldn't be attributed to a known channel, a data-quality gap worth closing.