Showing irrelevant content is one of the most annoying things we face in today’s digital footprint. A business needs to understand how to show the content without disturbing the user experience. Having an incremental user base means you’re doing something right – people are engaging with your product. Businesses that aren’t in touch with their users’ ever-changing requirements will find it difficult to sustain them. With the emergence of data and AI around it, targeting a user is not an option, it is mandatory.
We may target users who won’t like the app but we can’t miss on the users who would have actually liked the app, which is a Type II error (false negative) in statistics.
User targeting is an effort to get promoted content to the most relevant users who can be potential customers. In a January 2018 survey of 450 advertisers, 62.0% of respondents said that improving audience segmentation to support better ad targeting was one of their top campaign management priorities.
One of our product highlights is that we don’t ask for any information about the user. And, hence the biggest challenge is knowing the user and getting relevant content in front of the right customers. Using AI in user targeting, one can facilitate the optimal app in front of the most receptive audience. Through the use of preferences, trails, and tracks we leave online with every click and query, we are paving the way for apps to be more individualized than ever before.
How to Target?
With the information amassed, we can then use predictive analytics to process the data and predict behaviors based on topography i.e. previous interactions with apps, demography, and physiography along with the user and app metadata. These predictions are then used to display app collections or send push notifications based on the specific personalities. Machine learning categorizes from multiple data sources and segments the users into audience clusters called ‘personas’.
One of the older ways is to find look-alike users to the target users, which helps in some cases but in most, it is better to segment the users on multiple personas, so that the analyst has the freedom to mix and match them to create the target user base.
User personas are representative users whose goals and characteristics represent the needs of a larger group of users. Further, campaign analysts can use qualitative research and exploratory data analysis skills to find the best combination of personas that can restrict or widen the audience cluster.
Some of the static personas like age, gender, occupation, etc. are predicted using a large-scale AI classification algorithm trained on synthetically labeled data based on app usage. And, then some are constantly changing personas like type of gamer, traveler, job seeker, etc. which works on both rule-based and Machine learning algorithms that take into consideration the recency. Another most important personas are location-based personas like the city, state, locality, etc. which are derived using geohashing techniques.
- Demographic segmentation: age, gender, education, marital status, device model, OS, etc.
- Psychographic segmentation: values, beliefs, interests, personality, lifestyle, etc.
- Behavioral segmentation: purchasing or spending habits, user status, brand interactions, etc.
- Geographic Areas: neighborhood, area code, city, region, country, etc.
This makes it easier for app development companies to pinpoint entire groups of people with similar interests and target apps to a much smaller, more focused demographic. Through the use of targeted ads, app development companies spend less time advertising to people who have no interest in their products and more time getting the material in front of people who are most likely to install and engage, resulting in a much higher return on investment (ROI).
Post user targeting, it is equally important to measure and learn how the user responds to an ad, which helps us to improve the next ad serving. Right avenue and the right time to show an ad is as important as showing the right app to users. Some feedback signals to measure are:
- Interest intent: Score users on the interest shown to ad within an avenue and app category.
- Black-listing users: Users below an interest intent threshold within an avenue and app category.
- Whitelisting users: Promote apps to users with high-interest intent within an avenue and app category.
- Interaction-based time slot distribution: The best time when a user interacts.
- Analyzing the user feedback is as important as targeting the user.
- Always keep a count and limit the number of times targeting a user.
- User targeting is a combination of art and science.
- Through extensive research, we also came across some use-cases that require the recency of targeting like travel related campaigns.
- Enrich your user personas with the feedback data.
“The customer expects you to have knowledge of their stuff, not just your stuff.”
— Jeffrey Gitomer