Using Google Cloud for Marketing

Subscribe to our monthly newsletter to get the latest updates in your inbox

Building an Audience Segmentation Pipeline Using GCP + GA360

How to start using Google Cloud for Marketing

Google Cloud offers a wide range of services that add value to your marketing technology stack. The serverless infrastructure of Cloud provides a framework to deploy repeatable analysis solutions that connect your marketing objectives directly to your customers in real time. Cloud Platform services help power real-time or batch analytics pipelines that consolidate, analyze and activate on user signals coming from multiple marketing data sources. Using platform tools like Pub/Sub + Dataflow we have a fully managed, auto-scaling event processing system that loads data directly into BigQuery for further analysis. Getting started is as simple as contacting Adswerve to discuss our Google Cloud Platform Reseller offering, or can be done using your own credit card at . You will receive a $300 welcome credit when you open a project and attach a credit card to billing.  

Building an Audience Segmentation Pipeline

Example Architecture for Activating ML Audiences using GMP + GCP

Marketers need to be able to activate on their insights. We will walk through the steps needed to build a machine learning powered audience segmentation pipeline. This pipeline runs continuously and is able to bucket users into audience segments, then activate those audiences using Google Ads, Optimize 360, Display and Video 360, and more.  

Begin with a Marketing Data Lake using Google BigQuery

The most obvious scenario for marketers to begin to leverage Google Cloud is in building the data lake that consolidates disparate marketing tools into a single database , knocking down the data silos that hinder your existing capabilities. BigQuery is a technology that gives us access to limitless compute and storage, without the overhead of maintaining and provisioning servers. If your enterprise uses Google Analytics 360, then you have an “easy button” for building the foundation of your marketing data lake using the BigQuery Export for Analytics . This service comes included with your GA360 license, and is offered at batch (daily) or streaming (near real time) intervals. This export gives you access to 100% of the data collected in your GA accounts across web, mobile/apps, and connected devices. We recommend loading as much of your marketing data into BigQuery as humanly possible! This includes everything from your paid media to the local weather. The more source data you bring into BigQuery the more opportunities the machine will find predictive power in the data. Google Cloud offers a marketplace that has pre-built connectors to make it easy to load your numerous data sources into BigQuery.  

Use BigQuery ML to Build Audiences with K-Means Clustering

Once the data is in BigQuery, we leverage built in machine learning capabilities using BigQuery ML . This new machine learning feature is incredibly transformative! Previously, this capability was limited to machine learning engineers, who are few in number... BigQuery ML offers these tools to anyone familiar with SQL, serving a population several orders of magnitude larger. We have found the most valuable use case of using cloud for marketers is to use unsupervised machine learning to automatically build highly-relevant audience segments. We do this using the k-means clustering algorithm , that is now natively supported in BigQuery ML. Our certified Data Engineers use Google Cloud to prepare our data for machine learning processing. Using feature engineering, they analyze the data's predictive power towards many business goals. Using BQML syntax, a clustering ML model is trained and stored directly in BigQuery. Be sure to check out our blog post on BigQuery ML when it was first released, that post walks through the syntax in detail. Training and serving models directly in BigQuery is a huge leap forward for marketers! The model is used to serve predictions on new data using traditional SQL syntax. Historically, this work was done using sophisticated Python or R code, requiring additional technical resourcing to support these projects.   

Getting Audience Data into Google Marketing Platform

Once we have a clustering model stored in BigQuery, we’re able to use it to continuously generate audience files that we upload directly into Google Analytics. A common scenario would be to generate a daily CSV file that has the user ID and audience label as columns in the file. This file is programmatically loaded into Google Analytics using the GA Data Import API. Our machine learning generated audience label is attached to our users in Google Analytics as a user scoped custom dimension. This new dimension is available in reporting APIs and directly in GA using secondary dimensions or building custom segments with the dimension. Once a segment is built, we’re able to convert that segment into a Google Analytics Audience . This audience is portable between many Google Marketing Platform and Google Ads technology stacks. In real time, these audiences are activated using Google Ads, Display and Video 360, and Optimize 360. A popular use case is to build a marketing automation pipeline for the “high value” audience segment that can remarket to those users at strategic moments to support business goals. Increasingly, we’re seeing real time personalization using Optimize 360 targeting the same audiences in the display remarketing budget. In addition to the direct integrations with the Google Marketing Platform, we now have 3rd party platforms, like Salesforce Marketing Cloud , integrating with those same audiences. Our simple machine learning pipeline is now central to our marketing technology stack, using real time decision making that manages exceptional user experiences.  

Conclusion and Summary

Google Cloud Platform gives our clients everything they need to power advanced analytics and level up their marketing automation and activation capabilities. We’re able to consolidate and warehouse important marketing data sources. We’re able to easily orchestrate the ETL process and ensure our data is warehoused following OLAP best practices and ready for BI processing. We use machine learning processing with the language of SQL, a language many of our business stakeholders are comfortable with. Finally, we’re able to continuously activate and personalize user experiences using the power of the Google Marketing Platform. If you’re wanting to level up your marketing capabilities, be sure to reach out to our sales and account management teams and we can walk you through the steps to achieve your goals.