A quick start guide: Building lookalike audiences

Posted by Guo Zheng Ang on Jul 14, 2017 12:20:18 PM

Topics: Tips & How to, Product Features



REACHING OUT TO NEW AUDIENCES is always a part of every marketer’s top list. That’s why searching for lookalike audiences and targeting them should be included in your advertising arsenal.


Lookalike audiences are prospective customers who share a set of similar traits with your existing customers, thus the term “lookalike”. Comparable qualities such as age, income levels and interests, increase the probability that your new lookalike audiences will be interested in your product.


Lookalike targeting with Pocketmath

Pocketmath provides the opportunity for Advertisers who wants to DIY and experiment on lookalike targeting.


Getting Started

It all begins with your best performing audiences– customers who previously took action such as product purchase, form filling, or even by simply clicking on your ad. They will serve as seed to create new high-value target audiences who are more likely to engage and convert to your ad content.

Via Pocketmath ad macros, you can start mining valuable data as adverts are served to the seed audience. One vital set of data for lookalike targeting is the unique device identifier (UDID) object. UDIDs like Apple’s IFA or Google’s AAID function like fingerprints that identify individual users.


Customer Profiling

Once sufficient data on our seed audience is collected, we can move on to cluster audiences into segments. The objective is to gather insights on their characteristics, interests, traits and lifestyle choices. Armed with the UDIDs, Pocketmath can then match the seed audience with an existing database of audience segments using our **lookalike recommendation engine and the help of our partner DMP. The result is a list of insights about your audiences. Examples of these insights could be as follows:


  • 60% of these are aged between 16-23 and 23-31
  • 35% of these are male, 65% are female
  • 95% of these do not own a car
  • 80% of these are middle income families with 1 child or more
  • 40% of these do sports and 30% reads the paper

** This is not currently available in the system yet. However, we are working on it at the moment. Give us a nudge if you would like to be part of our initial trail.

Targeting multiple audience segments

Empowered with new customer insights, you can now expand your campaign’s reach. The insights reported will be designed in such a way that they match the audience segments readily available on the Pocketmath platform. By mixing and matching different segments, you can create new user personas that you can target based on the insights obtained from the prospecting results.




Six reasons to begin Lookalike Targeting 


#1 Maximizing results - Create more sophisticated yet accurate personas of your potential customers. When selecting multiple audience segments, you are no longer limited to demographics. Enjoy proven, sharper results with selections that include characteristics, traits and lifestyle choices.

#2 Increased Reach - Reach new and bigger audiences. Increase the potential for amplified conversion and revenue as your ad content connects to audiences who were never pinpointed by the original campaign.

#3 Creative Insights - Understand your customers better. Find out how ad content and the trafficking methodology affect different groups of customers. For example, promotion ad content might attract lower income consumers. Or premium ad content generates more engagement from middle to high income consumers.

#4 Choosing the right Audiences – Discover endless variations of new customer personas. Some of these personas might end up delivering better ROI than others. Hence, selecting the best performing persona would be essential in a KPI-driven campaign.

#5 Excluding users who have converted - Get attribution for engagement from existing customers. A UDID suppression list helps reduce the cost of ads served to existing customers, channeling most of the ad spend into new user acquisition.

#6 Frequency capping – Keep exposure at effective levels. If a user has not shown interest in the ad content after it appeared 5 times, repeating it another 20 times is futile. Stop at the fifth time when engagement looks unlikely.

Most platforms offering lookalike targeting solutions involve a blackbox. That and a lot of sheer trust that the algorithm is actually doing what it claims it will do. Pocketmath believes in empowering our Advertisers by giving them control and transparency in every step of the media buying process.


Advertisers could learn invaluable, new customer insights through lookalike targeting. More than improving their campaigns’ targeting, they will understand how different ad content impacts the return on their advertising investments.


Interested in DIY Lookalike Targeting? Try Pocketmath!




More stories to read.. 

Advanced RTB option: Pocketmath Mobile DSP Ad Macros

Ad Fraud 101: Understanding the different types of ad fraud

Rules made easy


Subscribe to Email Updates

Popular Posts

Recent Posts