Read behavior
I look for the human pattern inside funnels, cohorts, retention, CPA, ROAS, and product engagement.
Chapter 01Introduction
Xiangxi (Sarah)
Chen.
I turn “so what?” into “now what?”
I study how people behave, decide what growth move should happen next, and build the data products and AI workflows that make decisions faster.
Signal · Stage to system
My edge started on stage: make complex ideas simple, useful, and worth acting on.2024 —
The memory point
Not just data. The next move.
Everybody talks about data and AI. I care about the part after: what to do, how to test it, and what to build so the decision gets easier next time.
What I want people to remember
Sarah turns behavior into growth decisions, then builds the system that helps teams move faster.
Choose the next move
I turn signals into a recommendation: what to test, what to change, and where the risk is hiding.
Make it repeatable
When the same decision keeps coming back, I build the workflow, model, dashboard, or product around it.
Impact
Evidence, but only the kind that changes a decision.
The thread across the work is practical: less manual reporting, earlier risk detection, stronger measurement, and faster team adoption.
73.5%
lower forecasting error on media delivery risk
<20 min
to prep MMM inputs that used to take about a week
300+
people trained on practical AI workflows
4x
first-place wins across WPP and Cannes programs
Selected builds
I make the next move real.
A small product wall: internal systems, AI workflows, and consumer apps built from repeated problems I wanted to make easier.
CapPilot
A forecasting system that spots media delivery risk early enough for teams to act.
Case study →Luxe IQ
An AI workflow that turns reporting inputs into client-ready dashboards and strategy outputs.
Case study →VizUp
A solo-launched Office English coach for non-native professionals who want to be heard at work.
App Store →Jaavo
A calmer job application tracker, launched with a partner for people managing messy searches.
App Store →Why it works
The edge is the combination.
User sense
Language learning, career anxiety, workplace visibility, campaign behavior — I care about what people actually do, not what decks say they do.
Analytical depth
Forecasting, MMM, attribution, lift, SQL, and business intelligence give me the rigor to separate signal from noise.
Shipping instinct
If the insight needs a product, workflow, or model to become useful, I do not wait for perfect conditions. I build the first version.
Looking for teams building around growth, learning, behavior change, and AI-native workflows.
Currently open to senior marketing analytics, data product, and AI startup conversations