BI Report Factory

A clear path from a messy reporting request to an approved prototype and a build-ready feature.

My role: Product Owner, workflow author, approval-gate designer
Portfolio status: Sanitized working system

01 / Business problem

The decision behind the work

Reporting requests often arrived as a mixture of emails, screenshots, inherited reports, partially defined metrics, and assumed data sources. Starting development from that material pushed unanswered questions onto engineering and created avoidable rework.

I started with the decision the report needed to support, checked the data, built something the requester could use, and only then wrote the engineering handoff.

02 / My product decisions

How the decision works

01
Separate discovery from design. I created a structured intake that keeps unclear terms, metrics, audience needs, and success criteria visible until someone resolves them.
02
Check the source before designing the screen. I identified what the team could reuse, what needed a trusted SQL source, and what still required data engineering.
03
Require two different approvals. Requesters approve the experience; development leads approve feasibility and scope. Neither sign-off substitutes for the other.
04
Create one clear build handoff. The approved HTML prototype travels with the Azure DevOps feature, calculations, and acceptance criteria.

Interactive evidence

Follow one request through the factory

Mid-Month Sales & Operations Scorecard · sanitized exemplar · human approval remains authoritative

ReviewerOutput: Structured intake

Intake

Normalizes emails, screenshots, notes, audience, metrics, filters, and open questions. Ambiguity is surfaced instead of silently converted into scope.

SELECTED EXCERPT

Decision enabled: pressure-test whether month-end volume, economics, and supply health can reach leadership’s target.

03 / Evidence

What the demonstration proves

The sample below follows one request from the business question through data design and acceptance criteria. Known gaps stay visible instead of being hidden behind a polished screen.

Open interactive scorecard →Open what-if simulator →

04 / Outcome

12h → 6h

Agent-assisted execution cut the report cycle time in half while preserving human approval gates and explicit data-quality checks.

05 / What I’d carry forward

A prototype is valuable because it makes interpretation reviewable. It is not permission to skip source verification, metric ratification, feasibility, or requester sign-off.