← All projects
Analytics & GrowthComplete

Travel Product & Supply Analytics

Four equal travel-marketplace cases: content triage, funnel experiment, weekly ops packaging, and public-source readiness operations.

Evidence labelPublic listings + synthetic clickstream · Proxy and simulated impact

MY ROLE

Sole analyst across content scoring, funnel experimentation, weekly ops packaging, and the Bangkok content-operations platform case.

THE DECISION

Which content gaps, experiment readout, weekly exceptions, and readiness tickets should a Product Ops analyst act on first—under explicit evidence labels?

FOUR EQUAL CASES

Each case carries its own decision, evidence label, finding, action, and limitation.

1

Content Quality Audit

Decision. Which public-listing content gaps should be addressed first?

Evidence label. Public source + derived metrics; proxy opportunity estimate

Finding. 11,480 of 23,233 listings are below provisional score 70

Action. Prioritize the ranked backlog; validate outcome with an experiment

Limitation. The 8,908 annual-review estimate is a proxy, not booking, revenue, or causal impact

2

Conversion Funnel & Experiment

Decision. Should the PDP intervention ship?

Evidence label. Synthetic clickstream + simulated experiment

Finding. 25,000 sessions, 77,096 events; simulated result is DO NOT SHIP

Action. Keep the control and redesign before retesting

Limitation. No real user behavior or realized lift is represented

3

Weekly Product Operations

Decision. What exceptions need an owner in the next weekly cadence?

Evidence label. Derived portfolio package + simulated workflow

Finding. Reconciles P1 content and P2 funnel/experiment outputs

Action. Assign owners to the action queue and review exceptions weekly

Limitation. Illustrative WoW and workflow; no verified time-saving claim

4

Bangkok Content Operations Platform

Decision. Which public-source readiness issues should enter the ticket queue?

Evidence label. Public-source analogue + derived metrics + simulated workflow

Finding. 14,551 of 31,069 listings are below provisional readiness score 70

Action. Triage the simulated ticket queue by priority and SLA

Limitation. Airbnb/OSM analogue, provisional threshold, and simulated ticket/SLA fields; not Agoda data or workflow

APPROACH

  1. Scored public Bangkok listings and ranked a content backlog with demand proxies
  2. Built synthetic funnel SQL and a simulated PDP experiment readout
  3. Packaged P1/P2 outputs into a derived weekly ops demonstration
  4. Modeled public-source readiness issues with a simulated ticket/SLA layer

EVIDENCE & DELIVERABLES

  • 11,480 of 23,233 listings below provisional content score 70 (derived)
  • 25,000 sessions / 77,096 events; simulated experiment decision DO NOT SHIP
  • Weekly ops package reconciles P1 content and P2 funnel/experiment outputs
  • 14,551 of 31,069 listings below provisional readiness score 70 (derived)

SELECTED WORK FILES

HTML

Travel Analytics Case Library

Four equal cases with linked dashboards, executive briefs, methodology, limitations, and SQL where applicable.

8 KB
Open ↗
PDF

Travel Analytics Portfolio Case

A 12-slide portfolio overview that connects the four cases without elevating any single case.

216 KB
Open ↗
HTML

Weekly Ops Diagrams

Clickable diagram previews: open each full diagram or its Mermaid source file. Sample/process only — not Agoda production.

20 KB
Open ↗
HTML

Travel Funnel Dashboard

Browser-ready conversion funnel and experiment dashboard built from synthetic clickstream data.

5 KB
Open ↗
HTML

Weekly Product Ops Brief

Slack-ready weekly brief combining content health, funnel metrics, experiment readout, and ops flags.

4 KB
Open ↗
SQL

Content Quality Prioritization SQL

DuckDB queries that size content gaps and rank public-listing improvement opportunities; no listing data is bundled.

4 KB
Open ↗
SQL

Funnel & Experiment Analysis SQL

DuckDB funnel, device, cohort, and experiment-population queries for the synthetic clickstream case; no event data is bundled.

7 KB
Open ↗

EVIDENCE INTEGRITY

This case is presented as public listings + synthetic clickstream · proxy and simulated impact. The label distinguishes observed work from simulated impact, proxies, historical comparisons, or proposed architecture.

NEXT CASE

Analytics & GrowthE-commerce Growth