Data flow · PostHog SaaS analytics
TaskFlow: seeded event pipeline to report
TaskFlow: seeded event pipeline to report
A data-flow diagram of the TaskFlow demo SaaS: a seeded event simulator (2000 users, 30 days, seed 42) writes the committed data/raw/events.jsonl snapshot and can optionally send the same events to PostHog, the demo app captures browser and server events into PostHog, the analysis layer reads the committed snapshot or pulls the same events from PostHog's Query API, computes the activation funnel, day-N retention, the onboarding A/B test with a sample-ratio-mismatch check and revenue, and the shared chart builders render that one metric set into a Markdown report and a Streamlit dashboard: the A/B effect is set by the seeded generator, not observed in production.
01
SIMULATE
02
CAPTURE
03
ANALYZE
04
FIGURES
05
PUBLISH
EVENT
SOURCES
LIVE
PLATFORM
ANALYSIS
PIPELINE
RENDER
OUTPUT
SEEDED EVENTS
SIM
Seeded Simulator
scripts/simulate_events.py
seed 42 · sets the effect
LS
APP
Demo SaaS App
app/main.py · posthog-js
browser + server capture
LS
PLAT
PostHog Project
Query API · flags · errors
live ingest · --send
LS
LS
EVT
Event Source
data/raw/events.jsonl
live API · else snapshot
LS
DB
ANL
Analysis Suite
funnel · retention · a/b
chi-sq · SRM · LTV · ttc
DB
TB
DSH
Streamlit Dashboard
dashboard/app.py
cohort selector · 6 tabs
FL
CHT
Chart Builders
analysis/charts.py
matplotlib · shared figures
TB
FL
RPT
Markdown Report
posthog_analysis_report.md
analysis/report.py
FL
LEGEND
pipeline node
focal node
LS
event
DB
dataset
TB
table
FL
file
left chip = input · right chip = output · empty cell = no role in that step
data handoff
seeded source
API pull
optional --send
the A/B effect is set by the seeded generator, not observed in production