x 将值设置为两个变量时的 BigQuery 案例

Question

I am trying to identify session-level traffic sources in GA4 data within BigQuery. I initially used the traffic_source variables, until I realised they relate to the traffic source that first acquired the user: ie if I searched for Apple on Google then my traffic_source would be organic. if I then click on an email link from Apple my traffic_source remains organic (not email).

I can spot traffic from marketing channels with utm tags from the session start event, but am now trying to find traffic from channels that don't use utm tags: direct, organic search and referral. If the page_location does not have utm tags and the event_params contains a populated field called referral, then I know where the traffic came from.

SELECT
(SELECT value.string_value FROM UNNEST(event_params) WHERE key = 'page_location') AS url,
(SELECT value.string_value FROM UNNEST(event_params) WHERE key = 'page_referrer') AS referrer,
REPLACE(REGEXP_EXTRACT((SELECT value.string_value FROM UNNEST(event_params) WHERE 
 key = 'page_location'), 'utm_source=([^&]+)'), '%20', ' ') AS utm_source,
REPLACE(REGEXP_EXTRACT((SELECT value.string_value FROM UNNEST(event_params) WHERE 
 key = 'page_location'), 'utm_campaign=([^&]+)'), '%20', ' ') AS utm_campaign,
REPLACE(REGEXP_EXTRACT((SELECT value.string_value FROM UNNEST(event_params) WHERE 
 key = 'page_location'), 'utm_content=([^&]+)'), '%20', ' ') AS utm_content,
REPLACE(REGEXP_EXTRACT((SELECT value.string_value FROM UNNEST(event_params) WHERE 
 key = 'page_location'), 'utm_term=([^&]+)'), '%20', ' ') AS utm_term,
CASE
 WHEN 
REGEXP_CONTAINS((SELECT value.string_value FROM UNNEST(event_params) WHERE 
 key = 'page_location'), 'utm_') IS FALSE 
 AND (SELECT value.string_value FROM UNNEST(event_params) WHERE key = 'page_referrer') IS NOT NULL 
      THEN 'organic' 
      AND utm_medium = (SELECT value.string_value FROM UNNEST(event_params) WHERE key = 'page_referrer')
      ELSE REPLACE(REGEXP_EXTRACT((SELECT value.string_value FROM UNNEST(event_params) WHERE 
      key = 'page_location'), 'utm_medium=([^&]+)'), '%20', ' ')
  END AS utm_medium
FROM `bigquery***.analytics_***.events_***`
WHERE event_name = 'session_start'

so in my query I am extracting the utm values from the page location and placing them in new columns. I for organic search (which doesn't use utm tags) I am using a CASE statement. I want to set utm_medium as 'organic' AND utm_source as the referral site. Essentially trying to do this:

CASE 
 WHEN A = 1 THEN
  B = 'apple' AND C = 'fruit'
END 

Is this possible in BigQuery?

Answer 1

A CASE statement is an expression for creating a single column. The simplest thing to do is to simply apply the same expression twice:

...
    CASE WHEN a = 1 THEN 'apple' ELSE ... END AS b,
    CASE WHEN a = 1 THEN 'fruit' ELSE ... END AS c,
...

Alternatively, you could complicate things and use a STRUCT type column:

...
    CASE WHEN a = 1 THEN STRUCT('apple' AS B, 'fruit' AS C) ELSE ... END AS new_struct,
...

This accomplishes your goal in one step and allows you to do things like SELECT new_struct.B, new_struct.C in future.