A running total adds each row to everything before it: year-to-date revenue, members so far, stock used this week. cumsum(x) does it, in the current row order, so sort first and group by store when each store needs its own total.
cumsum() has no na.rm: one missing value makes every total after it NA. Decide what the gap means. If nothing happened that day, count it as zero with coalesce(x, 0L) or replace_na(); if it is unknown, say so in the report.
The same family includes cummax() (the highest value so far, useful for records), cummin() and cummean(). A cumulative share, cumsum(x) / sum(x) after sorting from largest to smallest, shows how few categories make most of the total.
Running totals and records per cafe, and a missing value counted as zero.