NPS Service Experience Study · Research · Dai Min

Brief overview

This project examines after‑sales service as a dense but under‑analysed source of cognitive friction in e‑commerce. Using an enterprise‑scale service experience study with 3,017 valid responses from a mid‑scale Chinese lifestyle e‑commerce platform, it tracks how friction concentrates across user lifecycle segments and purchase cohorts around Double 11.

Standard CX indicators (NPS, CSAT, CES and 30+ service KPIs) are combined with open‑text feedback and internal operational notes. The analysis shows that after‑sales scores deteriorate most sharply for general‑retention and first‑order users, with process opacity and escalation gaps driving dissatisfaction more than final resolution outcomes.

Cognitive friction in after‑sales

We treat stuck moments in refunds, repairs and price‑protection as cognitive friction: situations where users cannot easily infer what system is doing, what they are allowed to do next, or how long resolution will take.

Three friction types are used to read metrics and qualitative accounts:

Legibility friction – status, rules or timelines are hard to read (e.g. repair progress, logistics tracking).
Escalation friction – moving a case to right channel or level is effortful or blocked (repeat contacts, low first‑contact resolution).
Expectation‑gap friction – behaviour diverges from expectations set by marketing and past experience (Double 11 promises vs. actual handling).

Rather than attributing these patterns to user deficits or "lack of digital literacy", study treats them as features of sociotechnical system that can be measured within routine monitoring.

Study design

Context & Period

Platform: Mid‑scale e‑commerce platform.

Period: Nov 26 – Dec 3, 2018 (Double 11 After‑sales).

Methodology

Mixed-method study: N=3,017 responses via integrated channel delivery.

Data weighted by actual user distribution to ensure representation across cohorts.

User segmentation (lifecycle‑based)

Segment	Brief definition
Active users	Registered / paid within last month; high current engagement.
General‑retention	Previously paid, browsed in last month, but no recent payment.
High‑risk / Churned	Last visit 1–3 months ago; last payment >2 months ago.
Unpaid users	Registered but never paid; high friction at acquisition stage.

Note: Segmentation based on behavioral RFM (Recency, Frequency, Monetary) indicators.

This segmentation allows us to see how friction patterns change as users move from acquisition to retention and risk.

Key metrics and patterns

Composite Indicators

NPS variation by segment: Active (positive) vs. General-retention (negative).
After-sales identified as the primary weakness across nearly all purchase cohorts.

Item‑level Contrast

High satisfaction with Outcomes (Exchanges) vs. Low satisfaction with Process (Repair status).
Process opacity drives stronger negative reactions than final resolution gaps.

Lifecycle Effects

First-order users show sharpest NPS decline during peak friction log periods.
Multi-order users develop "learned friction tolerance" through repeated exposure.

Invisible Labour

Consultation volume spikes while first-contact resolution (FCR) drops significantly.
Frontline agents shoulder the "stitch-work" required by brittle system processes.

Cognitive Friction Framework

The study demonstrates that cognitive friction can be effectively detected using standard CX telemetry. We focus on three critical design nodes:

Process Legibility

Can users infer state and next actions from the interface alone?

Escalation Pathways

The cost of moving between automated and human-led support channels.

Expectation Management

Alignment between marketing promises and operational reality.

Programme Integration

This study serves as the empirical anchor for the shared data schema:

Values from this study populate the friction_record and metric_entity types.
Segmentation logic is integrated into the tag_entity taxonomy for cross-project querying.

Paper & Future Work

"Cognitive Friction in After‑sales Service" is currently in preparation for **CHI 2026**. Future extensions will focus on neurodivergent labor in public service service-desks.