Product Manager · Analytics · Data

From
insight
to impact.

Laura Cai · Auckland, New Zealand

I build digital products that move numbers — combining a product manager's delivery instinct with a data practitioner's rigour. 10+ years across ecommerce, fintech, telco, and O2O.

LinkedIn → View projects Email
eCommerce Fintech / Lending O2O Telco Identity & PKI
+50%lead conversion
92%user retention
10M+subscribers
−45%CAC
100K+listings
+18%ARPU
Systems built
Web / App PIM OMS CRM ERP Inventory Data Products Credit & Lending Order Dispatch Engine System Integration Identity & SSO BI Platform
About

Data-first product thinking.
Delivery that sticks.

I'm a Senior Product Manager based in Auckland, New Zealand, with over a decade of experience building digital products across ecommerce, fintech, O2O, and telco.

My edge is the combination: I think like an analyst and execute like a product manager. I don't hand off insights — I own the outcome. That means defining what to track, building the data infrastructure to see clearly, and then driving the product decisions that move the numbers.

"The riskiest thing in product is building the right thing on wrong data. I fix the foundation first."

I've shipped and scaled platforms handling billions in GMV, built credit decision engines with 70%+ automation, designed BI systems for 10M+ user telco datasets, and launched marketplaces from concept to 100K listings in 6 months.

I hold a Master of Information Sciences from Massey University and have been based in Auckland for two years, embedding in the local tech community through Colab Cohort's Advanced PM programme and industry volunteering.

10+
Years in digital
product & data
+50%
Visitor-to-lead
conversion lift
10M+
Users across
telco projects
92%
Annual user
retention rate
Experience

A decade of shipping
things that move numbers.

May 2016 — June 2022
Beijing, China · 6 years
UCAR Inc. Group
Senior eCommerce Product Manager → Product Lead
Led the full digital sales ecosystem for a listed automaker's online retail platform — 100K+ users, CNY billions in annual GMV, 500+ dealerships, 3,000+ internal staff. Progressed from Agile backlog execution to platform portfolio strategy and cross-functional leadership over 6 years.
4 years as Senior PM owning specific product lines end-to-end; 2 years as Product Lead managing a team of 6–10 PMs across 10 product lines and the full pre-sales, sales, and after-sales portfolio.
Built and iterated the conversion funnel (web + app) using A/B testing and funnel analysis — +50% visitor-to-lead, +30% visitor-to-sale, −45% lead CAC over one year.
Designed a sales analytics toolkit: unified customer identity across 6+ systems, built data pipelines, role-specific dashboards for management and sales consultants, automated all daily reporting.
Shipped one-stop auto-loan product (application → approval → repayment): built credit decision engine, managed credit bureau and identity integrations, 70%+ automatic approval, bad-debt under 1%.
Launched API integrations with 4 banks and 5 insurers — fully embedded flow (no redirects), expanding finance and insurance choices at point of purchase.
Built a used-car marketplace from concept to 100K+ listings in 6 months — MVP scoping under extreme time constraint, supporting B2C and C2C supply models.
As Product Lead: set platform-level roadmap and priorities, ran monthly reviews, weekly cross-team syncs, and owned Agile ceremonies across a complex multi-product environment.
eCommerceFunnel optimisationA/B testingData pipelineCredit modellingAPI integrationMarketplaceAgile · SquadTeam leadership
+
Expand
May 2011 — Nov 2013
Beijing, China · 2.5 years
BONC Corporation
BI Business Analyst · Telco Sector
Embedded analytics consultant for major Chinese telco operators — China Unicom and China Mobile — on BI platform upgrades and data-driven marketing programmes covering tens of millions of subscribers.
Led data governance programme for China Unicom Shaanxi: unified 30+ source systems (billing, CRM, network ops, device management, external data), standardised all metric definitions across departments, built automated data quality monitoring across completeness, accuracy, consistency, and timeliness dimensions.
Built BI platform from scratch: ETL pipelines, role-specific dashboards for management and ops, self-serve query tools for analysts, fully automated daily reporting — zero manual effort.
Designed user segmentation system using clustering analysis (SPSS) across 4 dimensions: RFM value, behavioural patterns, lifecycle stage, and preference characteristics. Maintained 5–8 actionable segments — analytically meaningful and operationally executable.
Built predictive models: churn prediction (spend decline + complaint frequency + usage anomaly signals), plan upgrade propensity (data cap saturation + device capability), and convergence selling scores.
Designed and executed targeted campaigns across SMS, app push, outbound calls, and co-marketing channels with banks and e-commerce partners — 92%+ annual retention, +18% average monthly ARPU.
Telco analyticsData governanceBI platformUser segmentationChurn predictionARPU optimisationSPSSETL pipeline
+
Expand
June 2015 — May 2016
Beijing, China · 1 year
Beijing Dakeyoujia
Product Manager · O2O Services Marketplace
Built a household services marketplace from zero — 100K+ users across 20+ cities, 10K+ service providers, routing 10K+ orders per day.
Designed and launched an automated order-dispatch engine with skill and location-based matching — routes 10K+ orders/day with under 3% manual intervention and over 95% fulfillment rate.
Owned Service Provider App and backend admin console — onboarded 10K+ providers (registration, verification, training), supported scheduling, order management, and payouts.
Owned Vendor Management portal for hundreds of agencies — e-contract onboarding, provider management, end-to-end online settlements.
O2OMarketplaceOrder dispatch engine0-to-1
+
Expand
May 2014 — May 2015
Beijing, China · 1 year
GOME Online
eCommerce Product Manager
Owned the product detail page and PIM system for millions of SKUs across consumer electronics and appliances at one of China's largest electronics retailers.
Owned catalogue/PIM system maintaining rich data for millions of SKUs, governing hundreds of categories and tags across appliances, consumer electronics, and apparel.
Built PDP recommendation engines ("Top Sellers", "You May Also Like") — achieved 5%+ CTR and 30%+ add-to-cart conversion on clicked recommendations.
Shipped UGC review system (photos, video, ratings) — buyer review rate above 5%, boosting decision support and platform trust.
Maintained 5–15% add-to-cart conversion rate on product detail pages through iterative UX optimisation.
eCommercePIMRecommendation engineUGCConversion optimisation
+
Expand
Nov 2013 — May 2014
Beijing, China · 6 months
Wangyin Online (JD.com)
Business Analyst · Payments
Payment platform under JD.com — one of China's largest ecommerce groups.
Designed and launched merchant onboarding flow (e-contract sign-up, profile management, settlements) for tens of thousands of merchants.
Delivered 20+ client payment-gateway integrations, enabling end-to-end payment and business workflows.
PaymentsMerchant platformSystem integration
+
Expand
Aug 2009 — May 2011
Beijing, China · 1.75 years
iTrusChina
Business Analyst · PKI & Digital Identity
China's leading SSL/TLS and PKI provider, serving ~400M users with digital certification and cloud e-signature services.
Drove requirements analysis and process re-engineering for cloud-based digital certification — cut institutional onboarding from 20–40 days to under 5 days (2.5× faster).
Enabled 50+ institutions to deploy and integrate PKI/CA with existing systems, issuing 10K+ digital certificates.
Co-built MVP of a certificate-based SSO identity system — requirements analysis, user research, and go-live support.
PKI / IdentitySSOSystem integrationRequirements analysis
+
Expand
Jan 2025 — Mar 2025
Auckland, NZ · 3 months
EcoMatters Environment Trust
Project Coordinator Intern
Coordinated EcoFest — Auckland's month-long environmental festival — managing the end-to-end publishing pipeline for 350+ events across 10+ organisations. Produced performance data reports feeding into the following year's planning.
Projects

Six case studies.
Products, data, and outcomes.

100K visitors 1,500 leads (+50%) orders 1,300 sales (+30%) −60% drop optimised target
eCommerce UCAR · 2016–2022 Product Management
Conversion funnel optimisation
End-to-end redesign of the online vehicle purchase journey — from test-drive booking through deposit, ordering, and CRM. Iterative A/B testing over 12 months across web and app.
+50%visitor → lead
+30%visitor → sale
−45%lead CAC
1% → 1.5%
Visitor-to-lead rate
0.1% → 0.13%
Visitor-to-sale rate
~¥500 → ¥275
Lead acquisition cost
The problem
The platform was new and conversion was low. But "low conversion" is a symptom, not a cause. The first task was finding where users were dropping off and why — which required trustworthy funnel data before any optimisation could begin.
The approach
Audited tracking layer first: cross-checked event data against backend transaction logs, fixed gaps, established a reliable funnel baseline.
Ran iterative optimisation loop: funnel analysis → user research → hypothesis → A/B test → ship → repeat. Multiple cycles over 12 months.
Test-drive booking: discovered drop-off wasn't digital — it was the offline experience. Standardised the dealership reception process, sales script, and lead capture to fix the full funnel.
Online ordering: added 3D car viewer, loan calculator, and transparent purchase journey visualisation to reduce perceived risk and build trust.
CRM: built lead status tracking and sales reporting so consultants could see each lead's visit history, intent signals, and follow-up priority in real time.
Key insight
Visitor-to-lead improved more (+50%) than visitor-to-sale (+30%) because as lead volume grew, so did the mix of intent levels — some natural dilution is expected. The full-funnel CAC still improved meaningfully because efficiency gains compounded across every step.
CRM OMS Inventory Finance ERP App Web unified pipeline management view sales CRM embed analyst self-serve +18%
Data Product UCAR · 2016–2022 BI · Analytics
Sales analytics toolkit
Unified 6+ fragmented systems into a single analytics platform with role-specific dashboards, embedded CRM analytics, automated reporting, and real-time alerts — driving measurable improvement in human performance.
+18%store sales / person
6+systems unified
0manual reports
+18%
Avg sales performance per person
6+
Source systems integrated
0
Manual daily reports (was 100%)
The problem
Sales data was scattered across CRM, OMS, inventory, and finance systems. Managers built Excel reports manually every week. Worse — "conversion" meant different things in different systems, so nobody trusted the numbers. Good data existed; it just couldn't be used.
Foundation first
Unified customer identity across systems — the same customer had different IDs in every system. Built identity resolution to create a single customer journey view.
Established company-wide metrics dictionary: one definition for "lead", one for "closed deal", agreed across all departments. No more metric arguments.
Built ETL pipeline extracting from all source systems, cleaning and standardising, loading into a centralised reporting layer.
Role-specific products
Management: macro dashboards — total leads, test drives, sales, store rankings, inventory, monthly achievement rates. Automated threshold alerts (no manual monitoring).
Sales consultants: analytics embedded directly in the CRM. Open a customer record and see full visit history, browsing behaviour, intent signals, and recommended follow-up actions.
Analysts: self-serve query capability and automated scheduled reports delivered on time, every time.
All: mobile-accessible key metrics. Real-time anomaly alerts pushed to responsible owners automatically.
apply + score auto · 70% expert · 30% approved <1% bad
Fintech · Lending UCAR · Credit Product
One-stop auto-loan
End-to-end loan product from application to repayment, with a hybrid credit decision engine — automatic approval for 70%+, expert review for the rest.
70%+auto-approval
<1%bad debt rate
70%+
Automatic approval
<1%
Bad debt rate
mins
vs. ~1 week before
My role
I owned the end-to-end loan product — from user-facing application flow through to the backend decision engine. The data science team built the scoring models (credit scorecard, blacklist, fraud detection). My job was to turn those models into a working product that balanced automation, user experience, and risk.
Key decisions
Designed 3-tier decision architecture: automatic approval, automatic rejection, and human-assisted review with system-generated analysis reports.
Core product principle: minimum data collection for maximum risk control. Every additional field increases abandonment — I pushed back on risk requirements that didn't justify the UX cost.
Managed external data integrations: credit bureau, national ID verification, third-party credit scoring. Each integration required interface design, encryption spec, and error-state handling.
Ran multiple rounds of A/B testing pre-launch — comparing automated decisions against expert approvals to validate accuracy before going live. Monitored bad-debt rate post-launch alongside the risk team.
30+ sources unified warehouse dashboards self-serve auto-reports
Telco · BI BONC · China Unicom
Telco data governance & BI
Rebuilt the analytics infrastructure for a provincial telco from the ground up — 30+ source systems, 10M+ subscribers, full data governance, and zero-manual-effort reporting.
30+sources unified
10M+subscribers
30+
Source systems
10M+
Subscribers covered
0
Manual daily reports
The governance foundation
Ran cross-department alignment sessions (marketing, customer service, network ops) to agree on all key metric definitions. One definition for "active user". One for "churn". No more metric disagreements.
Built automated data quality monitoring across 4 dimensions: completeness (missing fields), accuracy (cross-system consistency), consistency (same data, same answer), and timeliness (update latency). Threshold breaches automatically generate tickets to the responsible data owner.
Designed data cleaning pipeline: standardised field formats, resolved naming conflicts across systems, mapped 30+ sources into a unified schema.
The analytics layer
Management dashboards: ARPU trends, churn rate by segment, Net Adds, convergence rate, store rankings — updated daily, no manual assembly.
Self-serve query tools for analysts — no more waiting days for a data request to be fulfilled by the engineering team.
Automated scheduled reports delivered to stakeholders on time, every time.
high value churn risk mid tier
Telco · Analytics BONC · Segmentation
User segmentation & prediction
Clustering-based segmentation across 4 dimensions, feeding churn prediction, upgrade propensity, and convergence selling models — all translating into targeted campaigns with measurable ARPU impact.
92%annual retention
+18%avg ARPU
92%+
Annual retention
+18%
Avg monthly ARPU
5–8
Actionable segments
Segmentation system
RFM value model: ARPU, payment frequency, recency of activity — identifying high-value, mid-tier, and low-value groups.
Behavioural patterns: data usage intensity, service mix, peak usage times, content preferences.
Lifecycle stage: new, active, at-risk, dormant (zero usage 30+ days but not yet churned), and churned.
Preference characteristics: price-sensitive, data-heavy, voice-primary, elderly users, enterprise contacts.
Deliberately constrained to 5–8 clusters — analytically meaningful but operationally executable. More segments = less execution.
Predictive models
Churn prediction: spend decline trend + complaint frequency + usage anomalies + detected competitor touchpoints → risk score and intervention timing.
Upgrade propensity: data cap hit rate + 5G device capability + video subscription status → who is ready for a higher-tier plan.
Convergence selling: location data + family call graph → mobile-only users with high household broadband potential.
Campaigns executed across SMS, app push, outbound calls, and co-marketing (e-commerce and bank partners) — channel matched to segment behaviour.
0 6 months 100K launch MVP
eCommerce · 0-to-1 UCAR · Marketplace B2C + C2C
Used-car marketplace — concept to 100K listings in 6 months
Built from scratch: user-facing web, PIM, OMS, and CRM. Supported both B2C dealerships and C2C private sellers. The hardest product decision was knowing what not to build in version one.
6moconcept to launch
100K+peak listings
70%dealer supply
6
Months to launch
100K+
Peak active listings
2
Supply models: B2C + C2C
Strategic context
Parent company operated China's largest car rental fleet — generating a steady pipeline of retired vehicles. This became the seed supply. Combined with a fast-growing used-car market, the opportunity was clear. I owned the product side: web experience, PIM for unique per-vehicle listings, OMS for orders, and CRM for dealer relationships.
MVP decisions under time pressure
Framework: get the core transaction flow working end-to-end. Everything else is post-launch. No advanced personalisation, no full inspection infrastructure, no complex analytics in v1.
Prioritised dealer onboarding first: 70% of supply came from B2C dealers. Without a smooth bulk-listing tool and dealer portal, there was no marketplace. C2C individual sellers came second.
Used-car PIM was uniquely complex — every vehicle is a unique SKU with its own condition, mileage, and history. No standard product catalogue to inherit.
The hardest trade-off: vehicle verification and inspection infrastructure would build trust but couldn't be delivered in the timeline. Launched with a lighter disclosure model and clear buyer communication, with full inspection infrastructure planned for post-launch.
Skills

What I bring.

Product management
Product strategy & roadmap
Agile / Squad model
Stakeholder management
Cross-functional execution
Requirements & specs
Figma / Axure / Miro
Analytics & optimisation
Funnel analysis & conversion
A/B testing & experimentation
Attribution modelling
User segmentation & clustering
Churn prediction & retention
Cohort & lifetime value analysis
Data & engineering
SQL — advanced
Python — data analysis
Data pipeline design (ETL)
Dashboard design & BI
SPSS / statistical modelling
Snowflake / Power BI
Systems built
Web / App
PIM
OMS
CRM
ERP
Inventory
Data Products
Credit Engine
BI Platform
Dispatch Engine
SSO / Identity
System Integration
Tools
GA4 Adobe Analytics GTM / Tealium Jira Figma Axure Miro WordPress
Education & Certifications

The foundation.

Master of Information Sciences
Massey University · New Zealand
Feb 2024 – Jul 2025
Data Science — Python, SQL, data wrangling, visualisation
Advanced Machine Learning — deep learning algorithms, LLM
Big Data — multivariable analysis, statistics, SAS
GIS — geographic information systems, QGIS, Postgres
Bachelor of Computer Science
Central South University · China · QS #431
Sep 2005 – Jul 2009
Information Security specialisation
C/C++, Java, Software Engineering
IBM AI Product Manager Specialization
IBM · March 2025 · Credential ID: OJ2WFN4HUDNZ
Advanced Product Manager
Colab Cohort NZ · October 2025
Google Analytics Certification (GA4)
Google Skillshop · 2025
Security Operations Center (SOC)
CISCO · November 2025
Contact

Let's build something
that moves the numbers.

Open to product management, digital optimisation, and analytics roles in New Zealand.

LinkedIn → Email →