Why Gojek’s Biggest Complaint Isn’t About Drivers

2026-07-04 · Source: Naturallanguageprocessing on Medium · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, medium

Summary

An analysis of 225,000 Gojek app reviews, conducted using Latent Dirichlet Allocation (LDA), revealed that while driver-related issues are significant, GoPay wallet and transaction problems are disproportionately critical for 1-star reviews. GoPay issues accounted for 20.2% of 1-star reviews, despite making up only 13.9% of the overall corpus, indicating a deeper trust breach. Driver complaints were further segmented into "Driver Availability & Order Issues" (23.4% of 1-star reviews) and "Driver & Delivery Experience." Conversely, 5-star reviews predominantly cited "General Service Satisfaction" (36.6% of 5-star reviews), emphasizing basic reliability like "easy, fast, helpful, safe." The analysis also highlighted the importance of using Coherence Score (Cv) over perplexity for selecting the optimal number of topics (K=7) in LDA.

Key takeaway

For AI Product Managers building super-apps, you should prioritize the reliability of financial services like digital wallets, as failures here disproportionately erode user trust. Segmenting operational complaints, such as driver issues, into granular subcategories will enable more targeted solutions. Focus product efforts on ensuring core services are consistently "easy, fast, helpful, and safe" to drive higher user satisfaction and 5-star reviews, rather than solely pursuing novel features.

Key insights

Financial failures in super-apps break user trust more deeply than service inconveniences.

Principles

• Financial failures erode trust profoundly.
• Operational problems have distinct subcategories.
• Reliability drives positive user sentiment.

Method

Utilize Latent Dirichlet Allocation (LDA) for unsupervised topic modeling on app reviews, ensuring proper preprocessing and using Coherence Score (Cv) for optimal topic number selection (e.g., K=7).

In practice

• Segment "driver complaints" into distinct issues.
• Prioritize financial transaction reliability.
• Focus on core service reliability for 5-star reviews.

Topics

• Gojek
• Super-apps
• Topic Modeling
• Latent Dirichlet Allocation
• User Feedback Analysis
• Fintech Trust
• Coherence Score

Code references

• hamkhadafi/gojek-reviews-topic-modeling

Best for: Data Scientist, Machine Learning Engineer, AI Product Manager

Related on AIssential

• AI Weekly Issue #492: AI slop : A $725B bet on what no one wanted — Massive AI infrastructure investment faces growing user distrust and declining engagement with AI-generated content.
• Bluesky users are mastering the fine art of blaming everything on "vibe coding" — User perception of AI-assisted coding as inherently flawed creates a default blame for tech issues.
• AI Agents Need More Than Wallet Screening to Manage Risk — AI agents require destination protocol risk scores, like CORE3's PoL, to prevent losses beyond basic wallet screening.
• EXCLUSIVE: A 19-Year-Old Just Disrupted India's ₹22 Lakh Crore Auto Industry with AI — Trackeroo unifies fragmented Indian auto dealership operations with AI, achieving rapid growth through solving a critical, overlooked pain point.
• The Download: inside the QuitGPT movement, and EVs in Africa — User dissatisfaction with AI performance fuels "QuitGPT" while EVs and advanced nuclear power gain economic and environmental momentum.
• Why Payout Speed Has Become a Product Quality Signal With AI — Payout speed and transparency are now primary indicators of product quality and user trust in money-handling platforms.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Naturallanguageprocessing on Medium.