map(x, f) calls f on each element of x and returns a list. The typed versions return a plain vector instead, and stop if a result has the wrong type: map_dbl() for numbers, map_int() for whole numbers, map_chr() for text and map_lgl() for TRUE and FALSE.
A data frame is a list of columns, so map_int(customers, \(x) sum(is.na(x))) counts the gaps in every column, with each count named after its column.
To run the same step on each group, split first: split(cafes$revenue, cafes$store) makes a named list with one vector per cafe, ready for map_dbl(max).
map2(x, y, f) walks two vectors side by side, calling f(x[1], y[1]), then f(x[2], y[2]). When each call returns a small table, list_rbind() stacks the results into one.
Typed map functions over groups, columns and two vectors.