Skip to content
ale@portfolio:~/portfolio$ cat README.md

portfolio

Every case study here answers one question: measurable outcome.

lost-and-found-challenge.mdZapier

Lost and Found Challenge

ShippedTeam: Data
TL;DR

Turned ownership assignment for 500+ events and 3,000+ tables into a contest, taking ownership coverage from under 10% to fully tracked.

Outcome

  • 10%→100%ownership coverage across 500+ events and 3,000+ tables
  • 2 dayscatalog downtime during metadata recovery
Saw It
  • Fewer than 10% of tables and events had an identified owner, and only 40% of tables had been queried in six months.
  • Without ownership, monitoring, cleanup, and self-serve all stalled at the root.
↓
Figured It Out
  • Recognized the task itself was genuinely boring, so instead of mandating it, built a leaderboard contest with a prize.
  • Needed to make a tedious annotation task feel worth doing.
  • Looped in the whole data org through a monthly Data Governance call presentation.
↓
Did It
  • Shipped the contest with daily leaderboard updates and a celebration raffle at the end.
  • A filter change accidentally deleted all user-generated metadata mid-challenge, and the catalog was down for two days during recovery. Daily S3 backups meant nothing was permanently lost, but the recovery itself was stressful.

My Role

Owned the contest design, the rules, the roadmap prioritization, and the automation that now captures ownership by default going forward.

Full breakdown

Problem & Risk

Value Risk

No ownership meant no accountability for data quality, monitoring, or safe deprecation. Risk was continued erosion of trust in the data as unowned assets piled up.

Customer Discovery & Validation

Evidence

Under 10% ownership, 40% query rate, under 10% monitored, concrete numbers that made the scale of the gap undeniable and justified the contest format.

Context

Fewer than 10% of tables and events had an owner. Ownership had to exist before any of the governance work could stick.

Timeline

git log --oneline⏱ contest format
kickoff: <10% of tables/events had an ownerWeek 0
discovery: catalog ETL audited, contest rules scopedWeek 1
ship v1: contest launched, leaderboard and daily updates liveWeek 2
feedback: filter bug wiped metadata mid-challenge, 2 days downWeek 3
ship v2: automated ownership capture on every new table/eventNext

What We Did

  • Audited the Catalog ETL, pulled ownership from legacy sources
  • Wrote contest rules, presented at the monthly Data Governance call
  • Ran daily leaderboard updates, built feedback and bug-flagging forms
  • Hosted a celebration and prize raffle
  • Automated ongoing ownership capture for every new table or event
~ rough sketch, not an actual screenshot ~

<10% ownership

contest launched

leaderboard + prize

metadata wiped

2 days down

recovered from S3

ownership capture automated

going forward

Who Benefits

PMs, EMs, and non-data users can now self-serve the right data for their projects, because every asset has an accountable owner.

Next Steps

Would validate more broadly whether other teams wanted monitoring too, and tag tables by relevance upfront next time.