Hi, I'm looking for a way to improve loading time for a looker report on Meta ads (single source only) and found your article on Extract Data feature. My data connection is made with Windsor.ai. A rather simple raport with three charts (each with 12 months) and one metric, and it takes ages to load. Could you say a bit more on that: "if you’re date field is at a daily level and you choose to render it Year Month for instance, it won’t properly group" - does it mean that charts with year-month on X axis will not benefit from Extract Data?
Secondly, how does Looker "know" when to use Extract Data (after it is set up), and when to query the original source?
- It shouldn't take that long, how many accounts do you have behind, like many?
- In all cases, looks like you've got a use case to use Extract Data
- What I meant: If you extract data using Date as a dimension, Date = Day, and then use Date in a line chart, you won't be able to change this Date field and group it by Month in your line chart using Date type = Year Month.
- To do so, you would need to create a new dimension directly in the Extract (something like CAST(DATETIME_TRUNC(Date,MONTH) AS Date), and from there you would have a month dimension
- When you create your Extract Data, you can set up a scheduled refresh, and tell me: what is the timefrmae to look at (last 12 month, this month, between 2 fixed date), and when the refresh runs (Daily at 09:00, Weekly on Thursday, Monthly on the 22nd, ...)
In your case:
- Dimensions: Date, Month (as per the formula above)
- Metrics: anything you want to show
- Date range: Advanced Minus 12 Month, Today Minus 1 Day
Hi, I'm looking for a way to improve loading time for a looker report on Meta ads (single source only) and found your article on Extract Data feature. My data connection is made with Windsor.ai. A rather simple raport with three charts (each with 12 months) and one metric, and it takes ages to load. Could you say a bit more on that: "if you’re date field is at a daily level and you choose to render it Year Month for instance, it won’t properly group" - does it mean that charts with year-month on X axis will not benefit from Extract Data?
Secondly, how does Looker "know" when to use Extract Data (after it is set up), and when to query the original source?
Hi Janek. Thanks for your questions!
So:
- It shouldn't take that long, how many accounts do you have behind, like many?
- In all cases, looks like you've got a use case to use Extract Data
- What I meant: If you extract data using Date as a dimension, Date = Day, and then use Date in a line chart, you won't be able to change this Date field and group it by Month in your line chart using Date type = Year Month.
- To do so, you would need to create a new dimension directly in the Extract (something like CAST(DATETIME_TRUNC(Date,MONTH) AS Date), and from there you would have a month dimension
- When you create your Extract Data, you can set up a scheduled refresh, and tell me: what is the timefrmae to look at (last 12 month, this month, between 2 fixed date), and when the refresh runs (Daily at 09:00, Weekly on Thursday, Monthly on the 22nd, ...)
In your case:
- Dimensions: Date, Month (as per the formula above)
- Metrics: anything you want to show
- Date range: Advanced Minus 12 Month, Today Minus 1 Day
- Refresh: Daily at 09:00AM (for instance)
And everything should run smooth.
Hope it helps!