So, with some guessing, would
select
distinct g.customer_id,
g.cnt
from
(
select
distinct customer_id,
count(customer_id) cnt
from
table
group by
customer_id
) g
inner join
(
select
max(s.cnt) max_cnt
from
(
select
distinct customer_id,
count(customer_id) cnt
from
table
group by
customer_id
) s
) m
on
m.max_cnt = g.cnt
do the trick?