[Source Block] Facebook Ads by Stitch

(Dillon Morrison) #1

##What insights does this block offer?

Evaluate campaign performance by age, gender, placement, and device. This block includes three dashboards which provide granular analysis on ad performance metrics. Each dashboard and insights they uncover are described below.

Data types and technical info

This model is based on the schema produced by Stitch for data from Facebook Ads.


####Overview dashboard

  • See high-level metrics on marketing performance
  • Evaluate individual ad performance at a granular level
  • Monitor ad spend, clicks, and impressions over time to understand how your ad effectiveness has
  • Understand the geographic popularity of campaigns

####Placement and device insights dashboard

  • Evaluate cost-per-click, click-through-rate, and frequency by device
  • Compare the cost per action to the number of actions generated for placement-device pairings
  • Compare reach and click performance for placement and device combinations

####Age and gender insights dashboard

  • Evaluate your reach, clicks, impressions, and spend by age group.
  • Filter metrics down by gender

How to implement this block

The LookML for this block can be found in this Github repo. You can either download the entire block into your Looker application by following the directions outlined here, or selectively migrate pieces of the block by simply copying + pasting the block LookML into your Looker instance.

If you don’t have a Github account, we encourage you to create one for easy access to this block. If you don’t have access to the repo, or cannot make a Github account, please contact a Looker Analyst or email blocks@looker.com and we’d be happy to help.

(Segah A. Mir) #2

This repo uses avg. calculations (such as avg_cpc). These unfortunately would only work at an event level. Since FB insights are exported at an adset-date level, any average should be weighted by the corresponding value (e.g. clicks).

(Sami Yabroudi) #3

Really sorry if this is a bad question, but the table structure in here in no way seems to mirror the table structure for our stitch integration with facebook. Am I missing something? As far as I’m aware literally every table and field is being replicated.


(Sami Yabroudi) #4

It might be that stitch with Redshift structures the data differently than with non-relational database. Not sure.

(Peggy Beard) #5

Hi Sami! I’d like to know more about what seems to be different. Could you please support@looker.com with some more details on what you’re seeing? Once we get it sorted out, I will make an update in a comment here for others who might be seeing the same thing.