Insights and Recommendations — E-commerce Analytics¶
Executive Summary¶
Olist generates strong order volume but suffers from a critical structural weakness: an extremely low repeat purchase rate (~3%). The data points clearly to the cause — slow delivery destroys customer satisfaction, and dissatisfied customers don't return. Fixing delivery speed is the single highest-leverage action available, with the potential to lift review scores, repeat rate, and ultimately customer lifetime value.
Findings¶
Finding 1 — Delivery Speed is the #1 Driver of Satisfaction¶
Average review score falls steadily and dramatically with delivery time:
| Delivery Speed | Avg Review Score |
|---|---|
| Fast (≤7 days) | 4.32 ★ |
| Normal (8-14 days) | 4.15 ★ |
| Slow (15-30 days) | 3.74 ★ |
| Very Slow (30+ days) | 2.28 ★ |
A very slow delivery cuts the average review almost in half. Since reviews drive future purchases and marketplace trust, slow delivery has compounding negative effects.
Finding 2 — The Repeat Purchase Rate is Critically Low¶
Only about 3% of customers ever place a second order. For a marketplace, this is alarming — it means Olist is almost entirely dependent on acquiring new customers, with no retention engine. The cost of constant acquisition is unsustainable.
This low repeat rate is almost certainly linked to Finding 1: a customer who waits a month for their order and leaves a 2-star review is not coming back.
Finding 3 — Customer Lifetime Value is Thin¶
Because most customers buy only once, LTV ≈ average order value (~R$160). There's no compounding of value over time. The entire upside of an LTV strategy — where retained customers become more valuable each year — is being left on the table.
Finding 4 — Geography Drives Delivery Times¶
Delivery times vary widely by state, with remote regions experiencing the slowest delivery and consequently the worst reviews. The delivery problem isn't uniform — it's concentrated in specific geographies and likely specific sellers.
Finding 5 — Certain Categories Underperform on Reviews¶
Some product categories consistently receive poor reviews, independent of delivery speed — suggesting product quality or expectation-mismatch issues that go beyond logistics.
Recommendations¶
Recommendation 1 — Make Delivery Speed the Top Operational Priority¶
Action: - Set delivery SLAs by region and hold sellers accountable - Identify the slowest sellers/regions and address bottlenecks (warehousing closer to demand, better carrier partnerships) - Set realistic delivery estimates — and beat them (under-promise, over-deliver)
Expected impact: Moving "Very Slow" and "Slow" orders into the "Normal" band would lift thousands of reviews from ~2-3 stars to ~4 stars, directly improving marketplace trust and repeat likelihood.
Recommendation 2 — Build a Retention Engine¶
Action: - Post-purchase email flows (order follow-up, satisfaction check, reorder prompts) - Loyalty incentives for second purchases - Win-back campaigns for customers who haven't returned
Expected impact: Even lifting the repeat rate from 3% to 8% would meaningfully increase LTV and reduce dependence on expensive acquisition.
Recommendation 3 — Use RFM Segments for Targeted Marketing¶
Action: - Champions (recent, frequent, high-value): protect with VIP treatment - At Risk (were valuable, gone quiet): win-back campaigns - New (recent first purchase): nurture toward a second order — the critical conversion
Expected impact: Targeted spend on the right segments is far more efficient than blanket marketing.
Recommendation 4 — Investigate Poorly-Rated Categories¶
Action: For categories with low reviews independent of delivery speed, audit product listings, seller quality, and whether product descriptions set accurate expectations.
Expected impact: Removes a drag on overall marketplace ratings and reduces returns.
What I'd Analyse Next¶
- Review text analysis (NLP on the Portuguese comments) to understand why customers are dissatisfied beyond delivery speed
- Seller-level scorecard combining volume, delivery speed, and ratings to identify which sellers to promote or remove
- Cohort retention curves to measure whether retention is improving over time
- Freight cost vs satisfaction — does high shipping cost suppress repeat purchases?