A comparison between standard psychographic segmentation and geosocial segmentation.
Understanding the people in a community. This is at the heart of nearly every location decision. Whether you are buying a home, selecting a new site for your business, urban planning, or deciding where to invest in real estate, understanding the people in that area is crucial.
The foundation for understanding the people in a community is demographic data. However, demographics alone do not adequately describe a community, as we demonstrate in the previous post, Why Demographics Don’t Tell the Whole Story.
So what type of information can businesses turn to when demographic data isn’t enough? Traditionally, companies have used “Psychographics” to try and understand, as Experian describes it, the “hearts and minds” of the consumer. More recently, companies have begun turning to Geosocial data (which is what we provide at Spatial.ai) to measure the behaviors and personalities expressed in communities.
When using any type of data or information for key decisions, it’s important to understand where that information comes from and what use cases are appropriate. In this post, we will go through the key differences between the data that is used to create both Geosocial data and Psychographics and their different use cases.
Psychographics were created as a personality lens into demographic data. Providers in the real estate world include ESRI, Claritas, and Experian. Typically, households are assigned a classification such as “Metro Renters” (example of segments from Tapestry). This classification is defined by a certain demographic profile, including their age, income, and whether they live in a rural or urban area. The descriptions will also include additional information, like the type of television they are likely to watch.
The key element to understand this information is that the base data being used is demographic and geographic data. Third party datasets, like the Simmons Survey of 25,000 people, are then associated with the demographic data. These associations are then projected onto all other households that fit the same demographics. So surveyed people living in rural areas with high income tell Simmons that they like to watch romantic comedies, their answer will be projected on everyone else with the same demographics. Since there are many different datasets compared, it is difficult to understand exactly the level of coverage for each source. When interpreting psychographics it’s important to remember that the real data is mainly demographics, and the associations and personality information are projections that may or may not be true for that household.
Using surveys as a key source of information can be problematic. Not only because the results are extrapolated, but also due to the types of questions that are asked. There is a significant difference in designed data versus organic data. The U.S. Census Bureau discusses in this article the advantages and disadvantages of designed versus organic data. Designed data is useful when there is no other way to get the desired information. Organic data is preferred when available. The Census notes that designed data had been the only option historically, but with technology and the internet, the amount of organic data available dwarfs the size and impact of designed data.
When a survey asks if you agree with the statement “I’m no good at saving money” the learnings are far different from observing the natural attitudes being expressed by consumers around saving. It is also required that the survey designer decides what information is useful and important. This leads to insights like, “millennials use their cell phones in many different ways to get information.”
"72% use their cell phones in many different ways to get the information they need (123)"
The foundation for Geosocial data is location-based social media data. When people are posting publicly onto social media and opting to share their location, that data can be captured and analyzed. Since this is such a massive raw dataset with billions of data points, at Spatial.ai we have condensed this into 70+ segments. These segments are based on the most common patterns of organic behavior found in the raw data across the entire U.S. and Canada. It is not based on demographics or surveys in any way. However, our team has found that there are some statistically significant relationships with demographics when we analyze the top neighborhoods for each segment. We report those relationships in our taxonomy. Since they are not directly tied together, they are demographic tendencies of the areas that usually demonstrate a behavior. The primary quality of Geosocial data to remember is that when an area is scored for a segment (such as being an 82 for “Green Thumb”) it is measuring the organic data in that location.
In this simple example of an area in Mesquite, TX, we see that the zip code is summarized by psychographics as primarily made up of "Soccer Moms." According to the segmentation, nearly half of all households in the area fit into the Soccer Moms demographics. On the right, we see a heatmap of the the Geosocial segment, Connected Motherhood. Geosocial data shows the areas where this behavior is common, and also areas where it is quite uncommon in Mesquite. Geosocial data explains how common many different behaviors are in this area, and has identified "Green Thumb" as the most common behavior in the area. This is just one illustration of how different the information can be that these two datasets communicate.
Psychographics is demographic data with additional insights (like from surveys). Geosocial data measures the real activities of that community, and uses demographic tendencies as an additional insight.
Psychographics for location intelligence is very valuable when you are trying to better understand demographic information. If you know an area’s demographics, or you have a demographic profile that tends to be a good customer for you, psychographics can help you understand the common tendencies of people that fit the same demographics. Psychographics can also be a much simpler way to understand or digest demographic information, because it packages the many variables of demographics into easy to understand segments.
Geosocial data is only correlated with demographics, so if you are trying to better understand specific demographics, Geosocial data isn’t a great option. Correlations can be interesting and informative, but since they don’t have direct relationships, these correlations should be analyzed with a grain of salt.
A common situation we hear about is that companies will try to use psychographics in a statistical analysis that already incorporates demographic data. This is not a good use case for psychographics, because they won’t add signal to the statistical model. Data-driven tools for decision-making have to rely on unique datasets. Adding another dataset that is based on demographics won’t make them more accurate.
Geosocial data is a very large dataset that is difficult to use in its raw form for statistical analysis. Spatial.ai’s core focus is to make this data usable. If the data is made usable, it is very powerful in any data-driven or statistical modeling environment. Our blog has many case studies on this specific subject. The signal is very different from the signal demographic data provides. Geosocial data often provides more than a 20% improvement to models that have already incorporated demographics.
Because psychographics are rooted in demographics, they tend to not be as recent of a dataset. Surveys and projections of the census can be updated, but at their core any demographic information will rely on the U.S. Census (most recent Census was in 2010). For many use cases this is up-to-date enough. But if for some reason a more recent dataset is needed, like if you are trying to understand how an area has changed in the last year, psychographics is not the best solution.
Geosocial data is being created every single day. If you are trying to decide where to locate your business for the next 10 years, you don’t want to use data that covers a small time window. We typically recommend using a 2 year time window (that’s what we use) to analyze the data and update the data every 3-6 months.
Demographic data is based on households. If you have a use case where understanding qualities of where people live is important, psychographics can provide an easy way to understand the projected interests of those households.
Geosocial data is created where the behavior or conversation is actually taking place. That’s part of what makes it so useful in statistical modeling, it does a better job than household data at capturing the real behaviors of a community. However, if you are trying to understand information at the household level, like for a direct mail campaign, it is less appropriate.