From a public feed
to a useful decision.

THE PROBLEM

A chart is only useful when you know what it means.

A public feed can contain timestamps, units, missing fields and records from different periods. Putting those numbers on a screen is only part of the job. A reader needs to know where they came from, whether they describe the same interval and what the data cannot establish.

Electricity Pulse: compare the same interval.

We built a regional explorer for National Electricity Market dispatch observations. A visitor chooses a region, changes between wholesale price and scheduled demand, and moves through observations while retaining the source context.

Wholesale prices are shown in Australian dollars per megawatt-hour, excluding GST. Scheduled demand is measured in megawatts. The display identifies AEST as UTC+10 so a market interval is not silently shifted by daylight saving.

The saved snapshot makes the page usable immediately. A refresh checks for a newer retrieval. If that check fails, the interface retains the prior observations and shows the limitation; an older response does not silently replace newer data.

Inspect the electricity explorer

Government Buying Radar: follow the award to its source.

We built keyword and buyer filters around Commonwealth contract-award notices from AusTender. A visitor can inspect published values and dates, open the official record and download the evidence behind the view.

An award notice is not an open opportunity. Its published value is not a measure of payments or actual spending. The product labels its publication window, retrieval time and any partial retrieval. Missing dates, amounts and currencies remain missing.

Inspect Government Buying Radar

The engineering decisions you can check.

• Source links and definitions sit beside the information they qualify.
• Filters and chart controls change the view without removing the underlying context.
• Evidence downloads make it possible to inspect the selected data outside the interface.
• Empty search results have an explicit state and a reset action.
• Refresh failures retain the previous observations with a visible warning.

What this demonstrates—and what to measure next.

The delivered result is two public tools with working source connections, controls and evidence downloads. We have tested their mobile and desktop journeys, including filtering, reset and downloads. We have not published a measured user-productivity or revenue uplift for these tools.

For a commissioned product, we would agree the user's decision and baseline first, then test whether the interface reduces the work needed to find and verify the relevant evidence. The same disciplines apply to internal reporting, operational search and a source layer for an AI workflow.