You could encode "?" as missing value:

import pandas as pd

data = {"x": [1, 2, "?"], 'y': [3, "?", 5]}
df = pd.DataFrame(data)

print(df.isnull().sum())
# x    0
# y    0

df = df.replace("?", pd.NA)
print(df.isnull().sum())
# x    1
# y    1