Canada systematically under-reports inflation similar to the way they do in the US. But it's been that way for decades and their methods haven't changed recently so far as I know, so changes in the reported rate are still meaningful.

But it's reported monthly at an annualized rate. When they say 2.8% annual inflation, that doesn't mean it hasn't been much higher than that over the past year. Only that in that one month it was measured at a rate that would hypothetically add up to 2.8% if it continued for a whole year. When things are changing rapidly, one month's number might be less meaningful than usual.

It doesn't have to be all that realistic. All you are trying to determine is how the value of the currency has changed over time. Groceries tend to be especially volatile in their own value (fruits and vegetables can be less valuable during harvest, for example), which can mask the change in value on the other side of the equation. But at the same time groceries are great because everyone buys them regularly, which is exactly what you need to understand how their perception of the value of the currency is changing. Something people only buy once in their life would be completely useless. So, you seek some kind of balance in its inclusion.

It will never be perfect, but perfect need not be the enemy of good. And, in reality, economists will look at many different ways of calculating inflation when doing economisty things anyway. Headline inflation is just that – for the headlines.