Work and evidence

Case studies, labelled by what they prove.

Explore completed portfolio work and the methods behind it. Every entry is labelled so a demonstration is never presented as a client engagement or production result.

Evidence key

How work will be classified

These labels distinguish delivered engagements from demonstrations and set the evidence standard for future entries.

Client case study

A delivered client engagement with publication permission. Any stated outcome must be supported by agreed project evidence.

Anonymised client case

A real engagement with identifying details withheld. Scope and results remain factual, with sensitive figures removed or safely ranged.

Portfolio demonstration

A self-directed build using public or synthetic data. It demonstrates method and implementation quality, not a client result or production impact.

Methods behind the work

Explore the decisions behind the dashboard

These articles turn the portfolio build into reusable guidance on AI review, metric grain and refund-rate definition.

Why AI dashboard mockups still need human validation

A practical review of where a plausible mockup can hide metric, grain and decision-design errors before implementation.

Read the method

Power BI metric grain: from plausible numbers to trustworthy decisions

How to align session, order and order-item measures so a polished report does not silently compare incompatible grains.

Read the method

How to define refund rate without misleading the business

A worked method for choosing the right denominator, stating the limitation and keeping refund KPIs decision-safe.

Read the method

Work with HYDRADATA

Have a reporting challenge that needs clearer evidence?

Discuss a reporting project