Filter turning characters into numeric

A reproducible example would be useful here:
FAQ: What's a reproducible example (reprex) and how do I create one?

Without having access to the columns you're using, I can only say that one uses dplyr::filter() with unquoted columns. So your script may look something like:

cwm %>%
    group_by(ordernumber, description, customer) %>%
    filter(description != "Local Delivery")

Is it possible you aren't using dplyr's filter somehow? stats::filter() returns something similar to your data frame of V1, V2, etc.

If you explicity call dplyr::filter() do you get the same issue?