Takes one to know many

By IBM Watson Advertising

Finding the right audience for your brand is getting harder than ever. The upcoming changes to IDFA policies and cookies mean that the available data that can be used to identify potential customers will soon rapidly shrink.

But targeting is not becoming impossible. Data scientists excel at intelligently collecting and analyzing information – often first party data that will continue to be available – to understand and predict consumer behavior. But these teams are lean and their resources are limited, which often leads them to focus on initiatives that are more proven to drive revenue, growing existing customers rather than identifying new users.

Finding new prospects remains a top priority. Brands must discover how to do more with what they already have, utilizing their existing expertise and first party data to uncover new customers without adding workload for data science teams.

Better questions. Better results.

Sometimes the journey forward begins by stopping or even taking a step back to think. Instead of rushing to find alternatives to IDFA and cookies, brands should be asking more important questions:

  • How can we use our existing resources and first party data more intelligently and efficiently?
  • Instead of targeting on a mass scale, can we reduce wasted ad spend by focusing on the people who matter to our brand?
  • How can we achieve this goal with limited budget and resources?

Make your first party data actionable

As part of the suite of performance-based products from IBM Watson Advertising, IBM Predictive Audiences offers brands the opportunity to uncover new audiences more effectively by applying the industry’s leading AI to their first party data. This is achieved by using what you already know about your customers to identifying new potential users with a high likelihood to exhibit the same behaviors.

Your data science experts understand your first party data better than anybody. That’s why the first step is that your teams use LiveRamp to provide IBM with first party seed data that they have determined best represents your ideal audience. This data is combined with more than 15,000 user attributes and behaviors to understand what makes for high propensity potential consumers based on your goals and key performance indicators (KPIs).

IBM creates structured data sets based on the provided samples to generate and train models using IBM Watson Studio. Thousands of permutations are run across hundreds of various models to determine and train the best options for your brand. These training and testing processes involve traditional machine learning, deep learning and proprietary IBM models.   

The top performing models are used to score billions of RampIDs and predict which potential customers are most likely to behave like your client seed audience. Audiences are segmented into high, medium and low propensity to fit your objectives. These scored files are shared with your media team so your brand can connect with them through your preferred demand side platform (DSP).

Designed to grow revenue, not costs

Finding new audiences based off your existing first party data – and curated by your own experts – helps your brand improve efficiency by identifying prospects that are more likely to matter to your brand and accept your message. This can lead to significant revenue growth by converting high propensity prospects to actual customers without adding strain to your data science resources.

  • An outdoor retailer decreased its Cost Per Landing Page Visit by 44.6% among a Predictive Audiences-built segment based on its top customers.*
  • A major telecommunications provider’s Predictive Audiences segments led to a 45.7% conversion rate among users that engaged with the ad.*

To learn more about IBM Predictive Audiences – including how real businesses are driving conversion and reducing costs – read our new ebook here.

*Source: Results based on IBM Watson Advertising client campaigns.