Mohammed Rashad
← all projects
Power BIBIMarketingMySQLDAXSMART Goals

Olist Sales Funnel Insights (Brazilian E-commerce)

problem · Help a Brazilian e-commerce marketplace cut client acquisition cost while protecting merchant revenue as post-pandemic tailwinds reversed.

repo · Rashad-Mohammed02/OlistSalesFunnelBIupdated · 3 days agosize · 0 B

what shipped

  • Built a Power BI dashboard from a MySQL extract of 8,000 leads and 842 converted merchants, mapping a 10.5% overall conversion rate across acquisition channels.
  • Surfaced Paid Search (12.3%) and Direct Traffic (11.2%) as the strongest channels vs Social (5.6%) and Email (3.0%); recommended reallocating 30% of underperforming spend to lift conversion to 13%.
  • Quantified revenue concentration risk: construction tools generated R$50.7M (~80% of merchant revenue), with manufacturers outperforming resellers 5 to 1.
  • Diagnosed 4 ETL data-quality issues (text-typed fields, Portuguese-locale dates, blank segments, broken one-to-many join on reviews) constraining causal funnel analysis.

repository readme

fetched from github · rebuilt daily

Olist Sales Funnel Insights — Brazilian E-commerce BI

A Power BI dashboard analyzing Olist's lead-conversion and merchant-performance data to help cut client acquisition cost as post-pandemic e-commerce growth reversed.

Built for DAT-8564 Business Intelligence (MS Business Analytics, Hult International Business School, Spring 2026).


The problem

As COVID-19 e-commerce tailwinds faded and Latin American inflation climbed above 10%, Olist's growth decelerated. With $23M in Goldman Sachs funding deployed and 1,400 employees on payroll, the Sales Director needed a clear answer:

How can we reduce client acquisition cost while protecting the revenue our merchants generate?

The data

A MySQL extract from Olist's lead and marketplace tables:

EntityVolume
Leads in pipeline8,000
Converted merchants842
Overall conversion rate10.5%
Avg. review score4.017 / 5

Headline findings

Channel performance

ChannelConversion RateTime to Convert
Paid Search12.3%
Direct Traffic11.2%
Display5.1%189 days (fastest)
Social5.6%238 days (slowest)
Email3.0%
Unclassified ("blank")23.3%— (data integrity issue, see below)

Insight: Paid Search and Direct Traffic outperform Social and Email at comparable or shorter cycle times. Reallocating spend doesn't require new investment.

Revenue concentration

  • Construction tools generated R$50.7M — roughly 80% of all merchant revenue.
  • Phone/mobile a distant second at R$8M; all other segments below R$1M.
  • Manufacturers outperformed resellers 5:1 (R$50.8M vs R$10.3M).
  • DISC profiling: Stability/Green profiles drove R$51.4M vs R$8.6M for Dominant profiles — but this correlates almost entirely with the construction-tools manufacturer cohort, so it's an alignment artifact, not a validated targeting model.

Review scores vs revenue

The Avg. Review Score vs Total Revenue scatter showed no meaningful positive relationship between service quality and sales volume. High-revenue merchants were not consistently better-rated; the broad average of 4.02 means the merchant base is competent overall, but quality alone doesn't explain the post-pandemic revenue dip.

SMART recommendations

1. Channel reallocation (Q2 2026)

Reallocate 30% of Social and Email spend to Paid Search and Direct Traffic campaigns. Target: lift overall conversion from 10.5% → 13% by Q3 close. No incremental investment required — just shift budget toward already-proven channels with shorter conversion cycles.

2. Segment-targeted acquisition (3-month horizon)

Direct 60% of new merchant acquisition effort toward manufacturers in construction tools and phone/mobile, using the Stability DISC profile as the primary targeting criterion for Sales Rep outreach. Manufacturers generate 5× reseller revenue → focusing acquisition lifts revenue per merchant onboarded.

Benchmark performance at 90 days post-signup; target 20% above current average revenue per new merchant. Supports Olist's 22% commission model.

Data quality issues found

Four ETL-level problems constrained the depth of analysis:

  1. All fields ingested as text → required manual reclassification in Power BI.
  2. Date fields formatted in Portuguese locale → caused parsing failures; ISO 8601 at source would eliminate.
  3. landing_page_id field had numeric codes only, no descriptive labels → couldn't evaluate landing-page conversion.
  4. Reviews table couldn't be joined to orders due to a 1-to-many conflict on order_id → blocked time-series analysis of review-score impact on revenue.
  5. Lead/merchant pipeline and marketplace transaction data are largely separate populations — newly acquired merchants had not yet generated order history at extraction time, so funnel and revenue cannot be causally linked.

Deliverables in this repo

Stack

  • Power BI (Power Query for ETL, DAX for measures, multi-page dashboard)
  • MySQL as the source data
  • DISC profile taxonomy for sales-rep targeting analysis

Data source

Olist is a real Brazilian e-commerce SaaS marketplace; this analysis uses a public/educational extract of their leads + marketplace tables for academic coursework.