What the CGPI measures, and what it does not

The Canadian Grocery Price Index (CGPI) is a weekly measure of what a fixed basket of 50 everyday grocery items costs across major Canadian retailers. The headline index is a chained, matched-model weighted Jevons index: each week is compared with the week before over the items priced in both weeks, and those weekly links accumulate into the published level. The nine category sub-indices are, since August 10, 2026, computed on a fixed base: each week is compared directly with the week of June 1, 2026 (= 100), over the items priced in both. In either construction the comparison is same-composition, which is what lets us say a move is a real price change rather than a change in which products happened to be on the shelf.

As of the latest release the national index stands at 98.7, down 1.4% on the week and down 1.3% since the series began in July 2026.

What it can support

Week-over-week direction and magnitude for the national index, for each of the nine basket categories, and for individual cities and banners that clear the publication threshold. Because collection runs weekly, advertised-price movements and promotional cycles are visible as they happen.

What it does not support

Year-over-year comparisons: the published series begins in July 2026, so there is no prior-year base to compare against, and we do not publish a YoY figure rather than imply one. Cross-city comparisons of the index level: headline city and banner series are rebased to 100 in their own entry week, so those levels are not comparable between series; compare absolute basket cost in dollars instead, which is what our comparison tool does by default. (Category sub-indices are the exception: since 10 August 2026 they share a fixed base, the week of 13 July 2026 = 100, so category levels are comparable with each other.) And it is not a total-food-inflation measure: Walmart and Costco are not in the panel, and a cell only publishes once at least 60% of its basket slice is priced in both of the weeks being compared.

Every published figure carries its coverage and its methodology version, and the coverage page states the gaps in plain language. The point is not that the measure is perfect; it is that a reader can see exactly where it is strong and where it is thin.