Khoa Tran

Khoa Tran · Product Manager · Eindhoven, Netherlands

Product leadership across eCommerce, Growth, Data & AI.

I turn customer needs and complex workflows into products people use—from shopping and personalization across 36 European webshops to AI-assisted retail planning and enterprise data platforms.

6+years in product & digital delivery
36European webshops
70%first-month feature adoption

Projects

More projects

FireplaceDeals ecommerce homepage

Building a Magento webshop at scale

Delivering a commerce platform for 30,000+ SKUs and complex supplier and fulfillment requirements.

30,000+SKUs · 70+ modulesLaunched in 6 months
Read more
ACB mobile banking visual — data platform consultancy and implementation

Re-architecting banking data for scale

Separating analytics from core banking and building a foundation for real-time intelligence.

30%improvement in data-processing efficiency
Read more
Hoan My mobile app showing appointment booking and doctor availability HOAN MY · DIGITAL HEALTHCARE TRANSFORMATION

Driving digital healthcare transformation

From discovery to implementation of a connected digital healthcare system across a multi-facility network.

15 hospitals+ 6 clinicsDiscovery → implementation in 8 months
Read more
Power BI sales analysis dashboard from the project

Turning sales data into business decisions

Built a reusable Power BI model to diagnose performance across products, machines, locations, payments and time.

114.9Ktransactions · 3 months6+ analytical lenses
Read more

About

Khoa Tran

I connect customer needs, commercial priorities and technical constraints to decide what to build and how to deliver it.

Eindhoven, Netherlands 6+ years of experience B2C & B2B SaaS

How I make product decisions

I use the same operating logic across growth, data and AI products: make the problem explicit, reduce uncertainty with evidence, then scale only when the outcome is measurable.

Start with the decision, not the feature.

Frame the customer problem, commercial outcome and constraints before choosing the solution.

Use evidence to reduce uncertainty.

Define what needs to be true, test assumptions and separate engagement signals from business outcomes.

Turn complexity into focused delivery.

Connect design, engineering, data and operations around clear trade-offs, ownership and measurable success.

Experience at a glance

Product Owner Growth · vidaXL Dec 2025 — Present

Growth, recommendations and customer journeys across 36 European webshops.

Product Owner / Business Analyst · Conative AI 2024 — 2026

Retail planning, AI decision support, data platforms and integrations.

Business Analyst / Business Analyst Lead · CB/I Digital Jul 2020 — Jun 2023

Commerce delivery, banking data architecture and digital healthcare.

vidaXL · GrowthBack to projects
vidaXL logoITEMS PER ORDER

eCommerce & Growth · Product Owner · 36 webshops

Growing items per order through better product discovery

A growth program connecting shopper intent, relevant recommendations and easier basket building across 36 webshops.

1.35 to 1.55Items per order across the growth program
+0.20Additional items per order
36Webshops in scope

Challenge

  • Customers often bought a single item when their underlying need involved a complete project, room or usable product set.
  • Complementary discovery and basket guidance were inconsistent across the shopping journey.

My role

  • Led the IPO growth program, turning a broad set of ideas into focused workstreams with clear customer problems and success measures.
  • Defined customer-intent logic, data requirements and journey reporting with Data & Analytics.

Product strategy

  • Diagnose before designing: use order, product and category co-buy data to understand basket behaviour and where additional-item journeys break.
  • Match the intervention to the intent: distinguish completing a space, making a product usable, buying multiple units and seeking bundle value.
  • Connect the journey: organize recommendations, frequently bought together, cross-sell and guided selling around the moments when customers need help.

How I structured the growth problem

Instead of treating recommendations as a single UI feature, I mapped shopper intent to the right intervention and the evidence required before scaling.

1 · Shopper intentComplete a spaceMake the product usableBuy multiple unitsSeek bundle value
→
2 · Product interventionRecommendationsFrequently bought togetherCross-sellGuided selling
→
3 · Evidence to scaleRelevanceExposureEase of addingBasket impact
Key product decision
Treat relevance, exposure and ease of adding as separate problems—not one recommendation feature.

This made each assumption testable and prevented click-through rate from becoming the success metric. The program required measurable basket impact before broader scale-up.

Execution & trade-offs

  • Prioritized relevance, exposure and ease of adding items as separate problems to validate.
  • Defined recommendation and additional-item funnels across product pages, cart and checkout.
  • Required measurable basket impact before scaling; clicks alone were insufficient. Individual workstreams progressed through discovery, implementation and experimentation.

Measurement from exposure to business outcome

The measurement model connected module engagement to add-to-cart, orders and ultimately basket growth, while keeping conversion as a guardrail.

ExposureWas the right shopper shown the intervention?
ClickDid the recommendation earn attention?
Add to cartDid relevance translate into intent?
OrderDid the extra item survive checkout?
IPODid the program improve basket size?

Results & measurement

  • Items per order increased from 1.35 to 1.55 across the growth program—an increase of 0.20 items per order.
  • Evaluated progress through IPO, multi-item order rate, revenue per order and add-to-cart from relevant modules, with conversion guardrails.
  • This is a program-level outcome; it is not attributed to an individual recommendation or homepage change.
vidaXL · GrowthBack to projects
vidaXL logoLOYALTY & RETENTION

eCommerce & Growth · Product Owner · 36 webshops

Making loyalty easier to discover—and use

Expanded loyalty visibility in My Account to support member acquisition and reward-member activity across 36 webshops.

+19.9%Month-on-month new loyalty sign-ups
+34.24%Year-on-year orders from reward members
36Webshops with expanded exposure

Challenge

  • Existing loyalty value was underexposed in My Account, limiting discovery and new member sign-ups.
  • The opportunity was to make the existing program easier to find and connect visibility to meaningful member activity.

My role

  • Owned the expansion of loyalty exposure across 36 webshops.
  • Connected delivery to two outcomes: new loyalty sign-ups and orders from reward members.

Solution & decisions

  • Improve visibility at an existing touchpoint: expand loyalty entry points within My Account.
  • Scale the exposure: carry the changes across the 36-webshop estate.
  • Measure beyond enrolment: track reward-member orders alongside sign-ups to assess downstream activity.

Results

  • New loyalty sign-ups increased 19.9% month on month following expanded exposure.
  • Orders from reward members increased 34.24% year on year.
  • These measures capture different periods and outcomes: member acquisition and reward-member ordering activity.
Conative AI · Product case studyBack to projects
Conative product screens
Conative AI · B2B SaaS · AI-first analytics

From analytics to an AI-first decision loop

Conative had rich eCommerce data, but users still had to navigate reports, interpret signals and decide what to do next. I redesigned the launched analytics experience around recurring decisions and faster AI-assisted insight, then helped define the product direction toward a governed human–AI agentic workflow.

Product OwnerDiscovery · MVP · LaunchAI-assisted analyticsAgentic product direction
Problem statement

The product surfaced data, but the user still carried the cognitive load: finding the right signal, connecting context, understanding why it mattered, deciding what to do and leaving Conative to continue the work.

Business risk

Useful analytics without a strong frequency hook limited repeat usage and expansion within existing customers.

User friction

Collection · product · variant analysis was fragmented, with repeated search, filtering and exports.

Product opportunity

Create a decision loop where analytics and AI surface what needs attention, explain the evidence, support user judgment and create a reason to return.

My role

My role: discovery through launch

I led the product work from customer discovery to MVP definition: interviewed active users, mapped workflows and drop-off points, translated recurring questions into product requirements, prioritized the experience, and aligned Design, Engineering, Data/AI and customer-facing teams through launch.

Target personas

Lean teams making daily commercial decisions.

eCommerce managersMerchandisersInventory & demand plannersCommercial / marketing collaborators
The product

Helping teams identify what needs action

Before · Analytics-led

Search · Filter · Read · Export

  • Know where to look
  • Connect insights manually
  • Leave the product to collaborate
  • No strong reason to return
After · AI-assisted decision loop

Monitor · Detect · Understand · Decide · Return

  • Surface meaningful exceptions before users have to search
  • Use AI-assisted summaries to speed up interpretation
  • Keep context visible from signal to decision
  • Use insights and notifications as return hooks
Solution

Three product bets.

01 · REDUCE FRICTION

Task-based analytics

Simplified navigation and connected collection, product and variant analysis around recurring user questions.

02 · SPEED UP UNDERSTANDING

AI-assisted interpretation

Combined exception-first analytics with AI summaries so users could move faster from signal to explanation and decision.

03 · CREATE A RETURN LOOP

Reasons to come back

Used dashboards, proactive insights and notifications as recurring entry points while keeping decisions easier to share and follow up.

Metrics & impact

Did the new experience create usage?

70%of existing users used the new features within the first month after launch
+30%new active users observed after launch, alongside growth in returning users
1 monthfrom redesign to launch of the enhanced experience
Approach & methodology

Start with the decision. Work backward to the experience.

01 · Discover

Interview active users and observe how they actually evaluate products, collections and inventory.

02 · Diagnose

Map questions, touchpoints, friction, emotions and drop-off risks — not a feature wishlist.

03 · Prioritize

Turn recurring jobs into an MVP and distinguish reusable product problems from one-client requests.

04 · Measure

Define success around adoption, active usage and return behavior rather than feature delivery.

Customer workflow map
One artifact I keep: the customer workflow map that connected real user questions, touchpoints and pain points to the product decisions above.
AI-first product direction

From AI-assisted analytics to a governed agentic workflow

After the analytics redesign, I helped define how Conative could evolve AI Chat into a central work surface: a main agent coordinating bounded specialist analysis, using authorized tools, preserving context and returning evidence-backed recommendations to the user.

  • Goal or exception: the user asks a question or responds to a surfaced inventory, demand or performance risk.
  • Agentic analysis: the main agent coordinates the relevant specialist expertise and tools, then synthesizes evidence, assumptions and a recommended next step.
  • Human control: the user reviews the recommendation and retains authorization over consequential actions.
  • Continuity: state, follow-up and scheduled work create a path from one decision to the next instead of another one-off chatbot answer.

This section describes product and architecture definition for the agentic direction; some capabilities were proposed. The 70% adoption and +30% active-user results below refer to the launched analytics redesign, not the agent architecture.

Conative · Retail Planning · Data & AIBack to projects

Conative · Retail Planning · Data & AI

Making demand forecasts useful for planning

Turning forecast outputs into decisions about stock, purchasing and product lifecycle.

Demand contextSeasonality, stockouts and marketing signals
Lifecycle-awareCarry-over, new launches and end of life
Decision supportReplenishment and scenario planning

Challenge

  • Historical sales can understate demand during stockouts or overstate future demand when products approach end of life.
  • Planners needed forecasts they could interpret, challenge and apply to purchasing decisions.

My role

  • Translated customer planning problems into product requirements and business rules.
  • Investigated forecast behaviour with customers and connected outputs to replenishment, scenario planning and exception management.

Key product decisions

  • Account for demand context: frame requirements around stockouts, seasonality, marketing changes, one-off events and recent trends.
  • Separate actuals, forecasts and targets: let planners compare model output with performance and their own goals.
  • Treat lifecycle explicitly: distinguish established products, launches and discontinued items.
  • Connect prediction to action: bring forecasts together with stock, lead times and incoming purchase orders.

Validation & learning

  • Customer reviews exposed end-of-life over-forecasting, reinforcing the importance of lifecycle-aware requirements.
  • An Annmarie investigation found that a discrepancy came from manual conversion assumptions, helping clarify the forecast’s behaviour.
  • These reviews shaped requirements for explanation, traceability and planner trust.

Planning value

  • The broader Conative workflow helped Annmarie reduce quarterly finished-goods planning from roughly a week to an afternoon with spot checks, according to client feedback.
  • This is a workflow-level benefit across forecasting and planning—not an isolated measure of algorithm accuracy.
Fireplace Deals · eCommerceBack to projects

Fireplace Deals · eCommerce

Launching a Magento webshop at scale

Connecting a large product catalogue with the commercial and fulfilment logic needed to run an online retail business.

30,000+SKUs supported
70+Integrated modules
6 monthsTo launch

Challenge

  • Deliver a Magento commerce platform with a large catalogue and complex fulfilment requirements within a six-month delivery window.

My role

  • Business Analyst Lead at CB/I Digital: led discovery, requirements, wireframes and cross-functional delivery.
  • Translated storefront and operational needs into requirements that engineering teams could implement.

Solution & decisions

  • Connect shopping and fulfilment: define the storefront alongside the operational rules needed to fulfil orders.
  • Make supplier selection a commercial decision: shape logic that balances order profitability with shipping costs.
  • Coordinate platform delivery: bring the catalogue and integrated modules together into a launchable Magento webshop.

Results

  • Launched in six months, supporting 30,000+ SKUs and 70+ integrated modules.
  • Established the commerce platform and supplier-selection logic to support complex retail operations.

Explore

ACB Bank · Data Platform Back to projects

Fintech · Data Platforms · Business Analyst Lead

Data Platform Re-Architecture

Modernizing a legacy banking data ecosystem into a scalable platform for real-time analytics and future AI/ML use cases.

30% Improvement in data-processing efficiency
Phase 1 Landing Zone, Raw Data Vault and Business Vault established
AI/ML readiness Foundation for future use cases

Challenge

Reporting and analytics relied heavily on operational databases, creating performance risks and limiting real-time insights and scalability across Finance, Risk, Marketing and other teams.

My role

Business Analyst Lead

  • Analyzed customer, transaction and loan workflows and data.
  • Defined POC, business and technical requirements for data pipelines and ETL.
  • Bridged Business, Data and IT teams to shape the target architecture.

Solution

Helped shape a centralized data platform that:

  • Decoupled analytics from core banking systems.
  • Supported batch and real-time ingestion.
  • Combined Data Vault with Bronze, Silver and Gold processing layers.
  • Established MDM and golden records for consistent customer data.
  • Enabled BI, APIs and advanced analytics, with a foundation for future AI/ML use cases.

Target architecture

ACB target data platform architecture showing data sources, ingestion and hub, processing and storage, data access and intelligence, and business consumers.

EXAMPLE PROOF OF CONCEPT

Real-time personalized offers

Connected real-time card spending with customer and banking data to identify emerging interests and enable relevant offers through ACB’s digital channels.

  1. 01Customer event
  2. 02Real-time data
  3. 03Customer 360
  4. 04Behaviour model
  5. 05Relevant offer
  6. 06Mobile

Impact

  • 30% improvement in data-processing efficiency.
  • Phase 1 established across Landing Zone, Raw Data Vault and Business Vault.
  • Foundation created for real-time analytics and future ML-driven use cases.
Hoan My · Digital healthcare transformation Back to projects
Hoan My appointment booking interface Hoan My telemedicine and hospital appointment management interface

Healthcare · Mobile & Web · Data Integration

Driving digital healthcare transformation across 15 hospitals and 6 clinics

From discovery to implementation in eight months, connecting patient journeys, hospital workflows and shared data across a multi-facility healthcare network.

8 months Discovery to implementation
21 facilities 15 hospitals · 6 clinics
27% less Patient waiting time

Challenge

  • Long waits, fragmented medical records and inconvenient follow-up disrupted the patient journey.
  • Different hospital systems, limited API connectivity and uneven digital adoption complicated integration.
  • Eight months to move from discovery through implementation and rollout across 15 hospitals and 6 clinics, spanning data, backend, mobile and web applications.

My role

Business Analyst at CB/I Digital: led workflow and data discovery, defined product and integration requirements, and supported delivery and adoption.

Solution

  • Connect the care journey: bring booking, check-in, payments, telemedicine and follow-up into mobile and web experiences.
  • Unify the data foundation: consolidate HIS and LIS data in a lakehouse to support application services and synchronized records.
  • Coordinate delivery and adoption: run design, data and application work in parallel, with facility-level deployment and training.

One journey. Three connected stages.

Patient needs above; the hospital capabilities supporting them below.

01 Before the visit
PATIENT NEED

Arrive prepared

Book a doctor and time.
Provide details and link records.

HOSPITAL SUPPORT

Doctor schedules, booking capacity and patient-record matching.

02 During care
PATIENT NEED

Access care with less friction

Check in and pay digitally.
Consult remotely via telemedicine.

HOSPITAL SUPPORT

Queue codes, registration details, payments and consultation workflows.

03 After the visit
PATIENT NEED

Continue care

Access results and medical records.
Arrange a follow-up.

HOSPITAL SUPPORT

Digital results, synchronized records and follow-up reminders.

Shared data foundation Hospital information systems (HIS) + laboratory information systems (LIS) · Data lakehouse

Results

  • 27% reduction in patient waiting time: digital booking and integration contributed to this platform-level result.
  • Discovery, implementation and rollout completed across 15 hospitals and 6 clinics within eight months.
  • Patient reach expanded from 5–7 km to 80–100 km, as reported in the project summary.
  • App-based check-in, payments, results and follow-up reduced manual hospital administration.
Side Project · Power BI Sales AnalyticsBack to projects
Side · Data Analytics · Power BI

From raw sales data to clearer business decisions.

Built a reusable Power BI analytics model to explain sales performance across products, machines, locations, payments and time — and translate data gaps into concrete recommendations for better future analysis.

Data Analyst Power BI Dimensional modeling Exploratory & diagnostic analytics
Power BI product analysis dashboard
Problem statement
Business question

Where is performance strong or weak — and what explains the difference?

Data problem

Existing data answered basic reporting questions, but not deeper basket, customer or pricing questions.

Analytics opportunity

Create one consistent view for comparing products, machines, locations and sales patterns.

My role

Data Analyst — from business questions to Power BI analysis.

I translated business questions into KPIs, audited the data, designed the Power BI model and built descriptive and diagnostic analysis.

What I owned

Business questions · KPIs · data model · analysis · recommendations

Business decision framework

Move from reporting numbers to explaining performance.

MeasureSales, transactions, units and average value.
SegmentProduct, machine, location, payment and time.
DiagnoseCompare performance, trends, concentration and anomalies.
DecideIdentify assortment, location, pricing and operational opportunities.
Analytics foundation

A reusable Power BI model connecting sales to the dimensions behind performance.

I structured the analysis around Fact Sales and reusable dimensions for product, machine, location, payment and time.

ProductCategory · Subcategory · Price
TimeDate · Day · Month · Change over time
Central analytical layer

Fact Sales

Transaction · Product · Quantity · Sales · Refund · Machine · Location · Payment

Machine & LocationMachine type · Warehouse · Location type · Geography
PaymentCash · Mobile app · Other payment types
Recommended extensions

Customer identity · Basket structure · Price history · Promotion data — additional data needed to unlock cohort, affinity and price-sensitivity analysis.

Key data decisions

Data gaps became analytical recommendations.

Limitation

One transaction ID = one product.

Data-model decision

Capture the full basket under one transaction.

Capability unlocked

Basket affinity & cross-sell analysis.

Limitation

No reliable customer identity.

Data recommendation

Add customer identifiers where available.

Capability unlocked

Cohort & segment analysis.

Limitation

No price history.

Data recommendation

Track price and promotion changes.

Capability unlocked

Price & promotion sensitivity.

Scale & evidence

Broad enough to compare performance from multiple angles.

114.9Ktransactions analyzed
3 monthsMarch–May change-over-time analysis
6+ lensesproduct, machine, geography, location, payment and time

Analysis scope: transactions, time period and dimensions explored.

Power BI geography analysis dashboard
Example diagnostic view: geography performance and sales contribution.
What the model enabled

Reusable decision views

The same model supported product, machine, geography, location, payment, time, refund and warehouse analysis.

Approach & methodology

Start with the business logic. Build the model backward from the questions.

01 · Frame

Turn business questions into measurable drivers.

02 · Audit

Identify usable fields, relationships and data gaps.

03 · Model

Build reusable facts, dimensions and Power BI measures.

04 · Analyze

Compare performance and translate gaps into recommendations.

What this project demonstrates

Model the business question before building the chart.

That principle became a foundation for how I later approached data and product problems.