‍

To access the infographic as a PDF, check out the below link:

The Marketing Landscape

• Privacy changes are forcing performance marketers to rethink their UA strategies. Take heart: change is good.
  
  
• As a result, marketers are turning to first-party data, as it is a much more accurate data source, and it’s fully permission based.
  
  
• First-party data will have to play a bigger role in advertising, and that, in turn, will dictate how performance marketers leverage deep learning technology.
  
  

How to Use First-Party Data for UA: 7 Key Steps

Step One: Ditch Pre-Built Audience Segments

• Personas limit the targeting pool, and prevent you from finding high value users at scale, wherever they are in the digital universe.
  
  
• Let deep machine learning find them for you, without any of the bias that’s inherent in traditional marketing.

Step Two: Gather Your First-Party Data

• Your first-party data is your building blocks for successful UA campaigns and strong ROAS.

• Look for your:
  
  Campaign data log/campaign clickstream
  
  Conversion data from your mobile measurement partner (MMP)
  
  In-app interactions

Step Three: Work With a DSP

• A demand-side platform (DSP) is critical for UA success. A good DSP can use your unattributed first-party data to train its UA models, enabling you to find, acquire, and retain your best users at scale.
  ‍
• Be sure to work with a DSP with a proven track record, ideally one with quality machine learning models in place.

Step Four: How to Scale Campaigns Using Your First-Party Data

• Deep machine learning can take your first-party data and then extrapolate accordingly. It looks at the relationships of inputs (user and channel characteristics) to outputs (actions taken).
  
  
• Use machine learning to translate inputs/outputs from one campaign to another. Do this prior to a campaign.

Step Five: Start Training Your Model Prior to Campaign

• Don’t wait until your campaign launches to begin optimizing your campaigns. Train your UA campaign model on past data to get a jump start. It will deliver ROAS much faster.
  

Step Six: Use Smart Inference Models to Focus on Installs By High Value Users

• App install optimization: predict the likelihood of a user installing an app as a result of seeing an ad.
  
  
• In-app event optimization: predict the likelihood of a given user to install an app and take desired action.
  
  
• ROAS optimization: optimize based on predicted business outcomes.

Step Seven: Update Model Based on Results Hourly

• Use campaign results to update models every hour on the hour, or in as real-time as possible.

‍

Get in Touch: Moloco Cloud DSP is designed for your campaign success. Our machine learning engine automatically powers your app’s growth and removes data risk so you can exceed even your most demanding goals and your team can focus on what it does best: creativity and scale.

‍

Get Started Today

‍

‍