The data, as said, is de-identified and aggregated. I understand that it doesn't mean a lot in and of itself. The level to which it is de-identified and aggregated, though, is quite significant.
Effectively, for one data line, we have a table called “percentage time at primary location”. This basically represents the percentage of time that a cellular device will be at one place. Wherever it spends the most time is the primary location. The dataset would say a date—a single day—and would give a province, would give a health region, or potentially go down to a census subdivision, which is effectively a municipality, and then we get a percentage of time at that location.
For example, for Manitoba in a health region, you might have 91%. We would know that for the aggregate, for the time or the day—it's aggregated for an unspecified number of potential users within that health region—91% of the time that device stayed at its primary location. We would use that as a proxy for movement and adherence with public health measures.