Gong (Johnny) Chen

Selected work

Wefluens · LA-based influencer marketing agency · Data & Growth Analyst Intern

Problem: Outreach ran across 12 channels with no view of which ones worked. Volume was allocated evenly, by habit rather than evidence.
Approach: Segmented reply performance by channel, creator profile, and send timing across a 3,987-creator funnel, then tested where the variance actually sat.
Result: Found a 5× reply-rate gap between best and worst channels and shifted volume to the top performers. The funnel produced 600+ replies (15%) and 43 qualified partnership opportunities.

Wefluens · Python, Streamlit, Gmail API

Problem: Acquisition performance lived in manual spreadsheets, so leadership saw numbers late and inconsistently.
Approach: Built a dashboard covering daily volume, deliverability, reply rate, follow-up coverage, and pipeline stage mix, refreshed automatically and scoped to exclude personal data.
Result: Used daily by the CMO and a 3-person growth team as the single view of acquisition performance, replacing manual reporting.

Wefluens · Scoring framework & decision tooling

Problem: Which brands and creators to pursue was decided case by case, so judgments were not comparable and downside risk surfaced late.
Approach: Designed a 10-factor model — five opportunity factors (commercial fit, audience match, revenue potential, deal economics, traffic fit) and five risk factors (viability, monetization, exit, integrity, audience decay) — with thresholds that force a no-go.
Result: Turned a subjective go/no-go into a consistent, comparable score that flags high-risk targets before resources are committed.

Wefluens · Contact enrichment & outreach pipeline

Problem: Contact data came from a paid vendor billed per credit, and sending was template-only at roughly 200 emails a day — capped on both cost and personalization.
Approach: Built an in-house pipeline: multi-platform contact enrichment with mandatory source attribution for every record, plus individually personalized outreach and automated follow-ups.
Result: 601 verified contacts, up to 589 personalized emails per day, and a $2.4K–6.0K annual subscription removed — roughly 3× the daily capacity at lower cost.

Course project · Washington University in St. Louis · 11/2025

Problem: Did starting background checks earlier in the hiring funnel improve downstream conversion, and which acquisition channel deserved the budget?
Approach: Defined CVR and CPA as success metrics, validated treatment impact with Z-score hypothesis testing across funnel stages, and compared cost efficiency across social, referral, and job-search channels.
Result: Identified the funnel bottleneck and recommended a budget reallocation toward the lowest-CPA channels.