Ale CabreraHubSpot
Enrichment Data Accuracy Index
ShippedTeam: Data
TL;DR
Built a Snowflake accuracy index and self-serve dashboards, replacing manual data requests from Support, execs, and data team.
Outcome
- ↓duplicated engineering effort across teams
- →single source of truth, no extra engineering support needed
Saw It
- Support, execs, and the data team were each requesting adoption, stickiness, and churn data individually.
- Repeat requests for the same underlying data, with no shared source, meant duplicated engineering effort and inconsistent numbers across teams.
↓→
Figured It Out
- Built the index in Snowflake specifically so it could serve as one shared source rather than three separate pipelines.
- Needed self-serve dashboards on top, not just a backend table, or the same request pattern would continue.
↓→
Did It
- Shipped the accuracy index and self-serve dashboards.
My Role
Owned the index design and the dashboard build.
Full breakdown
Problem & Risk
Repeat, uncoordinated requests for the same data from three different teams, creating engineering drag and inconsistent numbers depending on who pulled what.
Context
Support, execs, and the data team each had to request adoption, stickiness, and churn data individually.
Timeline
git log --oneline⏱ single source of truth
kickoff: support, execs, data team requesting data individuallyPhase 0
discovery: accuracy index requirements scoped in SnowflakePhase 1
ship v1: accuracy index built in SnowflakePhase 2
feedback: self-serve dashboards addedPhase 2
next: single source of truth adopted, no extra eng support neededNext
What We Did
Built an accuracy index in Snowflake, backed by self-serve dashboards.
~ rough sketch, not an actual screenshot ~
3 teams requesting same data separately
accuracy index built in Snowflake
self-serve dashboards, single source
Who Benefits
Single source of truth for Support, execs, and the data team, with no extra engineering support required to get an answer.