I tried COALESCE() function as JHH suggested and didn't see much difference performance wise. it seems COALESCE() is internally a CASE statement.
However when I switched to:

isnull(percentile_cont(0.5) within group (order by transamt) over 
(partition by customer),0)

the computation time dropped by a factor of 2.
I don't know if this can be generalized to any scenarios or is it merely my particular query.

According to "...percentile_cont returns NULL if there's no transaction for a particular customer...", it could be a data issue or by design there could be entries for a customer without any transactions could have trans_date in the past 52 weeks but transamt is null. If that's the case, maybe this work for you by changing order by transamt to the following:

select distinct 
       customer,
       percentile_cont(0.5) within group 
       (order by case when transamt is null then 0 else transamt end) 
       over (partition by customer) as median_amt
  from trans_table;

Second guess: if percentile_cont() return NULL then show 0. Using coalesce().

select distinct customer,
       coalesce(percentile_cont(0.5) within group (order by transamt) over (partition by customer),0) as median_amt
  from trans_table;