Unique visitors
5
fixtureeligible users
Analytics engineering data product
DBT FIXTURE • 20 modeled events · loading tested output
Behavior, modeled.
Decisions, explained.
A reproducible measurement case study: one small synthetic event fixture, transformed and tested in dbt, then exported to this static dashboard.
Fixture window · loading
Engagement rate
100.0%
fixtureengaged sessions
Project view rate
83.3%
fixtureof eligible sessions
High-intent rate
83.3%
fixtureresume, LinkedIn, contact
RECRUITER JOURNEY
Sessions reaching each step
Landing6
100%Project viewed5
83.3%Case study1
16.7%Resume3
50.0%Contact1
16.7%Fixture interpretation only
These six sessions demonstrate the calculation path; the sample is too small for behavioral conclusions.
ACQUISITION
Traffic mix
6sessions
- Organic search50.0%
- Direct16.7%
- GitHub16.7%
- LinkedIn16.7%
01
Source distribution is present to prove attribution modeling—not to infer channel performance from six sessions.
Behavioral analytics
What visitors evaluate
Project interest joined with journey depth and downstream action—not vanity pageviews.
PROJECT PERFORMANCE
Content that moves decisions
ProjectViewsDeep readAction rate
JOURNEY MODEL
Fixture paths to high intent
Eligible sessions6
Project viewed5
Resume action3
Case study viewed1
Contact action1
Measurement system
The metrics modern teams actually need
Acquisition, engagement, conversion, retention, experience, and collection health—defined at the right grain and connected to action.
Page experience
Not collectedCore Web Vitals, p75
LCP—collector implemented
INP—collector implemented
CLS—collector implemented
TTFB—collector implemented
FCP—no RUM provider activated
Retention
Insufficient historyReturn cohort
CohortW0W1W2W3W4
Fixture5 users————
The committed fixture spans four days. Retention remains intentionally unavailable until a multi-week cohort exists.
Collection quality
CI verifiedCan this fixture be reproduced?
1 bot excluded1 internal excluded1 synthetic event excluded
Metric governance
Every number has a trail
Select a metric to inspect its definition, owner, transformation path, and freshness contract.
METRIC CONTRACT
Resume conversion rate
sessions_with_resume_action / eligible_sessions
VALUE50.0%
OWNERAnalytics Engineering
GRAINSession × fixture
FRESHNESSCommit-bound
→→→
Three of six eligible fixture sessions contain a resume view or download. This proves the formula and lineage, not a production conversion benchmark.
Reliability & DataOps
Built to be trusted
Tests, contracts, incremental processing, and transparent demo boundaries are part of the product—not footnotes.
15/ 15
DATA TESTS PASSING IN ROOT CI
Fixture contract verified
The root workflow rebuilds all models, reruns every assertion, regenerates the JSON, and fails if the committed dashboard data drifts.
✓
Generic dbt assertionsUnique, not-null, accepted values
13/13✓
Singular business rulesSession order and conversion bounds
2/2✓
Generated-data contractCI rejects dashboard JSON drift
ENFORCEDIMPLEMENTED FIXTURE + REFERENCE TARGETS
Free-tier and reproducibleREFERENCE ADAPTERS · NOT ACTIVATED
→
GA4PostHogWeb Vitals
REFERENCE TARGET · NOT RUNBigQueryprofile only
→
IMPLEMENTED + RUNDuckDB + dbt Coreseed → staging → marts
→
CONSUMEData productmetrics / lineage
GitHub Actionsroot CI: dbt build + export + drift gate
Privacy by design
Consent-aware activation, anonymous identifiers, least-data collection, and no public individual-level data.
Real and demo stay separate
This showcase uses a labeled synthetic dataset. Production models exclude synthetic traffic through explicit flags.
Cost-aware engineering
Partition pruning, incremental merges, rolling late-arrival windows, and aggregate exports minimize scan volume.
Portfolio intelligence platform
A transparent analytics engineering case study.
Versioned fixture → tested transformations → generated metrics → explicit evidence limits.