July 2024 change in collected_traffic_source in GA4 Big Query

Somewhere around 2024-07-11 the GA4 Big Query export table schema changed:

• manual_source_platform, manual_creative_format and manual_marketing_tactic were added to the collected_traffic_source record
• new columns appeared: batch_event_index, batch_page_id and batch_ordering_id

A very visible change.

Then.. shortly after that, between around 2024-07-15 and 2024-07-24 a quiet, gradual change took place: it looks like the collected_traffic_source column is now only filled on the first batch of events.

Why? Not sure.

Is it documented? Nowhere I can find

Why is the change so gradual? You tell me.

How does it look? See image below

Is this on all properties?

Not sure, I have checked it on multiple properties and it follows the same pattern.

Not always with this big a “limbo” period though.

Want to check yourself? Here is a query. Let me know if you find something different!

SELECT
  parse_date('%Y%m%d', event_date) as date,
  countif(collected_traffic_source.gclid is null) as empty_record,
  countif(collected_traffic_source.gclid is not null) as filled_record,

  count(*) events_in_later_batches
FROM
  `production.analytics_31337.events_202407*` -- change this to your dataset
where _table_suffix between '10' and '31'
and (select value.string_value from unnest(event_params) where key = 'page_location') like '%&gclid=%' -- filter add hits
and (select value.int_value from unnest(event_params) where key = 'batch_ordering_id') > 1             -- and second batch or higher
group by all
order by 1 asc

Related posts:

How to make sure your GA4 events tables do not expire Analyze AB tests in GA4 via Big Query Google Analytics 4 truncates page location Making sense of Event Parameters in GA4