Measuring time in funnel steps - SQL

1_filterRows.sql

SELECT
    CONCAT(post_visid_high, POST_VISID_LOW) as adobe_id
    , visit_num
    , DATE_TIME as hit_time
    CASE
      WHEN (instr(POST_EVENT_LIST, '211') > 0) THEN 'start' 
      WHEN (instr(POST_EVENT_LIST, '236') > 0) THEN 'end'
      ELSE POST_PROP1 
    END as registration_event_type 
  from 
    [table name] 
  WHERE 
    DATE_TIME >= TO_DATE('01.07.2019 00:00:00', 'dd.mm.yyyy hh24:mi:ss') 
    AND DATE_TIME <= TO_DATE('01.07.2019 00:30:00', 'dd.mm.yyyy hh24:mi:ss') 
    AND ((instr(POST_EVENT_LIST, '211') > 0) OR -- Registration Start
      (instr(POST_EVENT_LIST, '236') > 0) OR -- Registration End
      (POST_PROP1 is not null))

2_calculateStepStart.sql

  select 
    adobe_id 
    , visit_num
    , registration_event_type
    , min(hit_time) as timing 
  from 
    t 
  group by 
    adobe_id, 
    visit_num, 
    registration_event_type

3_calculateTimeDifferences.sql

  select 
    adobe_id 
    , visit_num
    , registration_event_type
    , trunc(TIMING) AS dt
    , round((timing - (min(timing) OVER (partition by adobe_id, visit_num order by timing asc))) * 24 * 60 * 60) AS time_diff_secs_start 
  from 
    t2

4_calculateMeasurements.sql

SELECT 
  registration_event_type
  , round(AVG(time_diff_secs_previous_step),0) AS avg_step
  , MEDIAN(time_diff_secs_previous_step) AS median_step
  , round(AVG(time_diff_secs_start), 0) AS avg_start
  , MEDIAN(time_diff_secs_start) AS median_start 
FROM 
  t3 
GROUP BY 
  registration_event_type 
ORDER BY 
  MEDIAN(time_diff_secs_start) asc