LOCATION LAB — PRESENTATION AUDIT — 1 OCTOBER 2026

Delivered: rebuilt phone-first six-screen workflow; nine actual-geography layers;
frozen free ABS, Overture, OSM, DTP and Vicmap inputs; current-run PDF/CSV/JSON;
12-page baseline report; notes appendix; saved-run restoration; source downloads.

Resolved during audit: PDF footer collisions (including the 5 km case), truncated
analyst notes, dropped draft form values, report business-category mismatch,
unstable configuration hashes due to key order, missing explicit basemap lineage,
source-pack omission of the actual frozen calculation bundle, legacy demo links,
and hidden mobile navigation actions.

Verified: 57 Python tests; 12 new browser/data acceptance groups; three legacy
browser suites; deterministic Python/JavaScript synthetic scoring parity. All
twelve final PDF pages inspected visually at print size. Current UI has no
browser exceptions or external provider requests during the tested walkthrough.

Important remaining evidence: actual inventory is zero. Three premises are
invented. No independent 30-business/ten-catchment audit has occurred. No validated
metro/category benchmark exists. Real market scores/ranks stay withheld. Property
use, access, rent, outgoings, availability and fit-out need manual verification.

Scope: relevant useful free sources were reviewed, not every dataset on the
internet. Overture is broader here than OSM, but this does not measure recall.
Providers overlap and are not summed. Population circles allocate 1 km cells
approximately; 2021 income is historical. Zoning does not approve a clinic and
station entries are not site footfall. Google is deferred; data/API spend is A$0.
