SQL & PL/SQL :: De-normalizing A Log Table And Counting Distinct Rows?

I have a table in which I can see via which communication channel we have previously communicated with our customers on different dates.

customer_interaction_log (I removed the date column)

customer_id channel
------------------------
1 SMS
1SMS
1SMS
1Email
1Email
2SMS

My goal is to have a de-normalized summary table in which total communication volume for each distinct channel is displayed per customer.Briefly, I would like to reach the following output

customer_interaction_summary

customer_id sms_countemail_count
-----------------------------------
132
210 (or null)

When I run the sql below I am not able to de-normalize by customer id; counts are accurate.

SELECT distinct cil.customer_id,
(CASE channel WHEN 'SMS'
THEN (SELECT COUNT (channel) FROM customer_interaction_log cil1
where cil1.channel='SMS' and cil1.customer_id=cil.customer_id) END) SMS_COUNT,
(CASE channel WHEN 'Email'
THEN (SELECT COUNT (channel) FROM customer_interaction_log cil2
where cil2.channel='Email' and cil2.customer_id=cil.customer_id) END) EMAIL_COUNT
FROM customer_interaction_log cil;

Output:

customer_id sms_countemail_count
-----------------------------------
1 3
21
1 2

I am trying to have one row per customer, however as you can see above it is not the case.

CREATE TABLE CUSTOMER_INTERACTION_LOG (
CUSTOMER_ID NUMBER(2),
CHANNEL VARCHAR2(10 BYTE)
);
Insert into customer_interaction_log (CUSTOMER_ID, CHANNEL) Values (1, 'SMS');
Insert into customer_interaction_log (CUSTOMER_ID, CHANNEL) Values (1, 'SMS');

[code].....