Benchmark methodology

A narrow, labelled comparison—not a market claim.

Every available benchmark is specific to an item, unit, cuisine, geography and time window. The method favours honest suppression over a weak number.

01

Make unlike invoices comparable

Approved lines are mapped to a canonical item and unit. Pack quantity and GST treatment must be clear. A 25 kg rice bag becomes an excl.-GST price per kilogram; an ambiguous pack stays out.

02

Select one eligible observation per site

For the item in question, PriceGrid takes each site’s latest approved, non-duplicate, non-quarantined observation inside the window. Multiple invoices from one site never create extra weight.

03

Use the smallest safe area

The cascade checks suburb, city and Auckland region for 30 days, then repeats over 90 days. It returns the first cohort with at least five eligible sites and labels that choice.

04

Calculate and timestamp the signal

The current cohort uses continuous percentiles for P25, median and P75. The interface shows its window, scope, cuisine, unit, GST basis and as-of time.

05

Suppress without leaking

If no cascade level reaches five sites, the operator response contains no count and no percentile values. It says only that there are not enough comparable sites yet.

Demo data · available

Boneless chicken thigh

Auckland region · Indian cuisine · last 30 days · 6 sites · NZD/kg excl. GST · as of 18 May 2026

P25$7.93Median$8.06P75$8.14Your price$8.18
Demo data · suppressed

Ginger paste

Auckland region · Indian cuisine · last 30 days · NZD/kg excl. GST · as of 18 May 2026. Your own history remains visible.

Not enough comparable sites yet