Display & Video 360 Feature Adoption

This is not an officially supported Google product. It is a reference implementation.

Overview

This is a reference implementation for a solution that will pull the necessary SDF and reports needed from Display & Video 360 into BigQuery to create a feature adoption dataset.

The idea is to deploy a set of Cloud Functions and schedule jobs for individual Display & Video 360 advertisers to generate a feature adoption dataset every 24 hours for analysis.

Prerequisites

• Git
• Node.js
• Cloud SDK

Deployment

1. Clone this repository:

   git clone https://github.com/google/dv360_feature_adoption.git
   

2. Change into the repository's directory:

   cd dv360_feature_adoption
   

3. Install the dependencies:

   yarn
   

4. Create a Google Cloud project (skip if using an existing project):

   gcloud projects create dv360-feature-adoption
   

5. Enable the Cloud Build, Cloud Functions, Display & Video 360, BigQuery, and Cloud Storage APIs:

   gcloud services enable cloudbuild.googleapis.com cloudfunctions.googleapis.com doubleclickbidmanager.googleapis.com displayvideo.googleapis.com storage.googleapis.com bigquery.googleapis.com
   

6. Create a Service Account for authenticating:

   gcloud iam service-accounts create dv360-feature-adoption \
       --display-name="Display & Video 360 Feature Adoption"
   

   You may need to grant this new service account the BigQuery Admin role in the Cloud console.

7. Create a Display & Video 360 User Profile using the service account's email address in the UI and give the user access to the Advertiser IDs you plan on scheduling jobs for.

8. Deploy the provided Cloud Functions to your Google Cloud project:

   gcloud functions deploy DisplayVideo360FeatureAdoptionSdf \
       --runtime nodejs14 \
       --memory 1GB \
       --timeout 540s \
       --trigger-http \
       --service-account "[SERVICE_ACCOUNT_EMAIL]"
   
   gcloud functions deploy DisplayVideo360FeatureAdoptionReport \
       --runtime nodejs14 \
       --memory 1GB \
       --timeout 540s \
       --trigger-http \
       --service-account "[SERVICE_ACCOUNT_EMAIL]"
   

   Replace [SERVICE_ACCOUNT_EMAIL] with the email generated while creating the Service Account.

9. Add the roles needed to run the functions:

   gcloud functions add-iam-policy-binding DisplayVideo360FeatureAdoptionSdf --region=us-central1 --member=allUsers --role=roles/cloudscheduler.jobRunner
   gcloud functions add-iam-policy-binding DisplayVideo360FeatureAdoptionSdf --region=us-central1 --member=allUsers --role=roles/cloudfunctions.invoker
   gcloud functions add-iam-policy-binding DisplayVideo360FeatureAdoptionSdf --region=us-central1 --member=allUsers --role=roles/bigquery.admin
   
   gcloud functions add-iam-policy-binding DisplayVideo360FeatureAdoptionReport --region=us-central1 --member=allUsers --role=roles/cloudscheduler.jobRunner
   gcloud functions add-iam-policy-binding DisplayVideo360FeatureAdoptionReport --region=us-central1 --member=allUsers --role=roles/cloudfunctions.invoker
   gcloud functions add-iam-policy-binding DisplayVideo360FeatureAdoptionReport --region=us-central1 --member=allUsers --role=roles/bigquery.admin
   

Scheduling

1. Create a Scheduler Job for both Cloud Functions with:

   gcloud scheduler jobs create http dv360-feature-adoption-job-[ADVERTISER_ID] \
       --schedule "0 0 * * *" \
       --uri "[CLOUD_FUNCTION_URI]" \
       --http-method POST \
       --message-body='{ "advertiserId": [ADVERTISER_ID] }'
   

   Replace [CLOUD_FUNCTION_URI] with the deployed URI for the Cloud Functions and replace all instances of [ADVERTISER_ID] with the Display & Video 360 Advertiser ID.

Testing

1. Run the function locally:

   yarn run-report
   yarn run-sdf
   

2. Call the functions:

   curl -X "POST" "http://localhost:8080" \
       -H 'Content-Type: application/json; charset=utf-8' \
       -d $'{ "advertiserId": [ADVERTISER_ID] }'
   
   curl -X "POST" "http://localhost:8081" \
       -H 'Content-Type: application/json; charset=utf-8' \
       -d $'{ "advertiserId": [ADVERTISER_ID] }'
   

Authors

• Tony Coconate ([email protected]) – Google