LIMIT 10 doesn't make it cheaper: BigQuery bills the bytes it reads, not the rows it returns
A SELECT * with LIMIT 10 returns ten rows and can still bill the whole table. How BigQuery stores and reads data, and how to make every query read less.
Wrong With Data: digital analytics and GA4 mistakes, explained
Find it. Understand it. Fix it.
What's wrong with my data? Let's find out.
Most broken dashboards are not broken by the tool. They are broken by small setup mistakes that nobody notices. We find them, explain why they happen and show how to fix them.
How each mistake is explained
- find
The symptom: what you see in the reports, and how to spot it in your own account.
- understand
The cause: where the mistake is born (the site, the tag manager, the property settings) and why.
- fix
The fix, and how to check that it worked before you trust the data again.
From a 60-second video to a full article
Every short video we publish on Instagram, TikTok and YouTube has a post here. The post replays the same animation in landscape, step by step and without sound, and explains it calmly in writing: the context, the edge cases and how to verify the fix.
What we cover
The whole data stack, one mistake at a time. We started with Google Analytics 4, the Meta Conversions API and BigQuery.
- Analytics GA4, Adobe Analytics
- Tag managers Google Tag Manager, Tealium iQ
- Pixels Meta, TikTok, LinkedIn, Google Ads
- CDPs Segment, Tealium Audience Stream
- Warehouses BigQuery, Snowflake, dbt
Latest posts
-
-
Every day, every time: querying the GA4 export's events_* with no _TABLE_SUFFIX filter
Without a _TABLE_SUFFIX filter, events_* reads every day you've exported, plus intraday tables with streaming on. Higher bills and totals that don't match.
-
One purchase, six rows: CROSS JOIN UNNEST(event_params) before counting in the GA4 export
Unnesting event_params with a CROSS JOIN turns each event into one row per parameter, so COUNT and SUM multiply. How to read parameters without duplicating events.
-
Two users, one session: counting GA4 sessions in BigQuery with ga_session_id alone
ga_session_id only identifies a session within one user, so COUNT(DISTINCT ga_session_id) merges sessions of different users. How Google counts them instead.
Data not adding up?
Tell us what you see in your reports. Email us at hello@wrongwithdata.com