Behavioral Intelligence for Automated Segmentation
Behavioral Intelligence is utilizing user behavior to influence actions to achieve desired outcomes, and this article focuses on how it relates to advanced and automated segmentation.
Join the DZone community and get the full member experience.Join For Free
Behavioral Intelligence is utilizing user behavior to influence actions to achieve desired outcomes.
Behavioral Intelligence enables developers, marketers and app publishers to grow their userbase by:
- Identifying usage patterns and anomalies, user behavior trends and shifts
- Creating a deep understanding of users
- Harvesting insights
- Making timely decisions
- Acting on those decisions
- Measuring the progress and impact of actions against the desired outcomes
- Influencing user behavior to achieve desired outcomes
Acting on insights in a timely manner is particularly important in environments where user behaviors change quickly as they do for mobile apps.
It is important for marketers and app publishers to build meaningful one-on-one relationships with users and over time attempt to affect behavior towards achieving desired outcomes. They can do so by personalizing interaction points and experience, including content feeds, user interfaces, communications, and promotion of overall usage, both new and under-utilized app features, and in-app purchases.
Building one-on-one relationships using traditional analytics is not possible in an evironment where an app or a service can have hundreds of thousands to millions of users. Identifying app users by segmenting them into niche groups of users who have common needs, interests, goals and priorities is useful to achieving effectiveness of promotions and communications for users within the same segment. However, traditional approaches are just too expensive, time consuming, and require significant expertise.
General Segmentation Techniques
General techniques used for segmentation include demographic, geographic, cohort groupings, technographic, psychographic, attitudinal, needs-based, outcome-based and behavioral segmentation.
Let’s start with a recap various segmentation and targeting techniques used today.
“Who They Are”
Segmenting users by who they are, for example, if a mobile commerce app wants to target to shoppers, you may think of women between 26 and 42 as a primary segment to target.
Marketers have known and incorporated Demographic (age, gender, income group, etc.) and Geopraphic segmentation in their strategy since the 1950s as it was easily collectable for radio advertising. Demographic segmentation’s effectiveness is questionable as it can be easily copied by competitors. See why relying on demographics alone is limiting in this well researched think with Google article.
Organizational Demographic Segmentation
“Who They Work For”
Organizational demographic is a segmentation technique that bring the enterprise into the mix.
Segmenting based on the industry, revenue, number of employees. For example, people who work for MasterCard.
“Where Are They From”
Segmenting users based on their geo-location, locale or region, for example, people from NY are more likely to sign up for weather alerts app in winter.
“When Did They Become a User”
Segmenting users by a monthly or weekly cohort, for example, users who joined last month.
“What Technology Do Your Users Use”
Segmenting based how users are using your app in terms of device model, device manufacturer, cellular carrier, size of screen etc. for example, people using Apple iPhone with AT&T versus people using a value phone over metroPCS
“What Users Say”
Segmenting based on user’s expressed beliefs. For example, people who responded to a survey saying they have a subscription to WSJ. On mobile, it is much harder to ask survey-like questions.
Polls and rating are common, but why ask if you can figure it out? An app should not ask anything it could work out using sensors available in a mobile device. Device sensors profoundly change what an app can know.
“What Are Their Values, Interests, Opinions and Lifestyles”
Segmenting based on users values, opinions and lifestyle. For example, people who use fitness apps, or have a huge music collection, or have multiple photography apps. Psychographic data is not always attitudinal, i.e. collected as an expressed belief. For instance, an app can collect psychographic data based on what users click on in a content feed.
“What Do They Need”
Needs-based segmentation involves segmenting users based on what users need or what product features and benefits were most appealing to them. For example, people from Boston who live in suburbs may indicate that they need snow removal tools. The needs of users can be collected using surveys or detected based on likes and visits to items of interest.
“What Outcomes You Want From Users”
Outcome-based segmentation is advantageous as it enables segmenting users based on what outcomes app publishers want to achieve from each segment. A group of users that may be okay spending (via in-app purchases) on filters and another segment may be perfect for sponsored filters. Segments are differentiated from feature-driven needs-based segmentation.
Motorola for example, used three market segments for its radio product: Discrete and covert communications (e.g. FBI, CIA), communications in life-threatening situations (e.g. Fire fighting), and communications for administrative tasks (e.g. organizing an event). Market segments may have overlapping needs (e.g. longer lasting battery).
“How Users Behave and What They Do, In Your App”
- Do they respond to specific push notifications and in-app messages?
- Do they research before buying in a mobile commerce app?
- What in-app purchases have they made in the past?
- How many positive and negative experiences have they had in the past?
- How many support tickets have the opened in the last 3 months?
For example, people who are engaged with your app and read at least 2 articles on WSJ app on a daily basis.
Generally, what users are doing your app is a much better way to determine meaningful segments than demographic, geographic, technographic or psychographic segmentation alone.
Relying on demographic, geographic, technographic or psychographic segmentation alone? Sorry, you missed the candy-crush-saga-playing 76yr old male & calculus-loving 10yr old girl.
Behavioral Intelligence & Automated Segmentation
Pyze supports all segmentation techniques automatically using the power of Intelligence Explorer, which is a key feature that enables app publishers to explore the entire user base through automated segmentation and real-time explorations across key behavioral attributes like Engagement, Loyalty, Form-Factor, App Starts or Sessions, Attrition Risk, Last Activity, Monthly Cohort, Recency, Relative Cohort, Weekly cohort, Weekly Loyalty or frequency, Revenue, Purchases, Long term value, Long term purchases, etc. in seconds. See Intelligence Explorer.
Further, app-defined dimensions make the exploratory intelligence offered by Intelligence Explorer even more powerful. They can be used in concert with the out-of-the-box dimensions to conduct fine grained explorations and precise targeting of specific groups.
App-defined Dimensions allow app publishers convert any behavior or categorization within an app, into an explorable dimension. For example, an app-defined dimension could be based on “Ticket holder Type” category for a sports team app to differentiate between season-ticket holders and discount-ticket buyers, or “Responsiveness to Political Alerts” for a news app to customize experience for users who are highly responsive to political alerts vs those who are not.
Marketers who try to reach their audience solely on demographics risk missing more than 70% of potential mobile shoppers – Mobile search & video behavior analysis, Millward Brown Digital, U.S., January-June 2015.
Also see my blog post: How games make money using behavioral intelligence.
Published at DZone with permission of Dickey Singh, DZone MVB. See the original article here.
Opinions expressed by DZone contributors are their own.