Discussion: Exploring explores ?

[er1k]

Hey everyone, just wanted to discuss something that I have had in my mind for a while.

What would you think if it would be possible for regular users to select fields under the explore (Example: customer_id, count_of_orders, avg_order_size) and then they would be allowed to use this dataset as a new dynamic explore?

Meaning the selected fields will be moved to subquery and all selected fields would become available as a dimension for the new explore. Then the user would be able to build smth like average_order_size, count_of_customers (as custom measure), sum_of_orders (as custom measure).

I know it goes a bit against Looker rule of having everything written down in LookML, but it would create many more discovery options for endusers without developers needing to define all the potential slicings (which could be super ridiculose some times, like: “users who activated in the last 30 days during the weekend, had campaign code, spent more than x and used payment method y vs Others who activated in the last 30 days during the weekend”

What do you think of this idea?

[michael_zearn]

I like this idea a lot. I often get requests from our team that could be answered this way. Usually things like “How many [DIMENSION] had 100+ [MEASURE] in [DATE WINDOW]?”

Obviously one can create summary tables and stuff like that, but those can’t answer every query, and they get unwieldy. We’ve set up a model using templated filters and liquid parameters that can handle some of these requests, but not all of them. Oftentimes our business users just download results from Looker to pivot in Excel, which I don’t think is a great solution.

[er1k]

Exactly! I’m also trying to avoid moving away from Looker. Currently it’s annoying two step task from Looker to Excel. This also makes enduser to want even more raw data, so they could to every calculation outside Looker.