In this post, I want to explore an unorthodox way of making demographic-based advertising more effective using some well established behavioral economic and psychological findings.
Demographics as identity
So how can you make more effective use of your demographic data? Make it visible to your customers.
In most cases, demographics operate in the background, and simply serve as a tool to determine what ads to show. But as these ads don’t tend to be very effective, a way to build on this model is to also show customers the demographic information itself. Done correctly, this may serve to motivate customers by leveraging who they are and how they perceive themselves (what the psychology literature might call their “identity”).
The goal of this approach is to move away from “You might prefer X” and arrive at “People like you prefer X.” With surprisingly few exceptions, people like to feel like they belong, and they tend to prefer to behave as typical members of whatever their relevant social group may be. There are a couple different reasons this may happen, but typically it involves a desire to have a coherent identity or a desire to appear acceptable to others (sometimes even if no one may be watching). This is why, for example, public service advertisements that make claims like “4 out of 5 people do X” tend to be relatively successful.
Demographic data allows you to construct similar appeals: “4 out of 5 men prefer X” or “2 out of 3 San Franciscans like Y” will be much more persuasive than simply targeting men from San Francisco with canned ads for X and Y. Rather than simply targeting a demographic, this tactic is now targeting customers’ identities.
A nice example of this comes from a 2008 study by Noah Goldstein, Robert Cialdini(author of Influence), and Vladas Griskevicus. They set up a series of experiments in hotel rooms in which they aimed to promote bathroom towel reuse. You may have noticed signs in hotel bathrooms encouraging you to reuse your towels in order to save water and energy. The authors manipulated what these signs read in order to appeal to different levels of guests’ identities. As their baseline, they used the standard pro-environmental message that appears in so many bathrooms (i.e. “Help save the environment!”), but in addition, they appealed to:
- Patriotic identities: “Join your fellow citizens in helping to save the environment!”
- Gender identities: “Join the men and women who are helping to save the environment!”
- Customer identities: “Join your fellow guests in helping to save the environment!”
- Targeted identities: “Join the guests who stayed in room #xxx in helping to save the environment!”
Each of these was accompanied with a message that “75% of guests reuse their towels,” with a notable tweak that the most targeted message specified “75% of guests who stayed in room #xxx reuse their towels,” where this number always matched the room in which the sign was displayed. The graph shows the effectiveness in terms of the percent of guests who reused their towels:
We can draw a number of interesting conclusions from their data, but there are two big takeaways that I think are most relevant. The first and most important result is that the very specific identity, namely “people who stayed in the same room,” was the most impactful message, with 49% of these guests reusing their towels as compared to 37% who received a generic call-to-action. On one hand, this demonstrates that demographic appeals will be more persuasive when they are more specific and more relevant to the situation. On the other hand, it also suggests that the effect was not hurt by guests’ realization that their own behavior was being monitored, a concern which pervades many people’s online behaviors.
The second takeaway, which may be less obvious just eyeballing the chart, is that while the patriotic, gender, and guest identities all produced better outcomes than the generic message, these three demographics were not statistically different from each other. Thus, it would be premature to look at this data and think that patriotic appeals (e.g. “I’m a San Franciscan!”) are more effective than gender appeals.
Reinforcing the identity
You may have already noticed two big real-world challenges to using demographics in the way I have suggested. The first is that it requires having accurate data. If a customer is a twenty-something woman, you don’t want your demographic appeals thinking she’s a 35 year-old man (a mistake Google has been known to make). Assuming your data is accurate, there is still another problem: which demographic identities do you use?
To get the strongest effect, your customers need to feel that the comparison group you are showing them is relevant. On one hand, this could mean determining what your customers value: am I more motivated to be “like a man” or “like a San Franciscan?” On the other hand, you could take a more behavioral economics approach and tell themwhich group is important.
The idea here is pretty simple: if you can get a customer to embrace the fact that they belong to a given group, their affinity for that group increases (at least in the short run). If a man endorses the idea that he is a man, he will likely behave in a more masculine way. When San Franciscans acknowledge the fact that they are from San Francisco, they will likely feel worse about the 49ers not making it to the Super Bowl. They will also be more interested in products that other, like-minded San Franciscans prefer.
An interesting example comes from research on the psychology of multiculturalism and globalization. A number of studies (see here for an overview) have found that bicultural individuals think and act differently depending on which of their cultural identities are “active.” For example, Chinese Americans who were shown images or symbols emblematic of China (such as the dragon below) expressed a greater appreciation for what they considered “Chinese values,” but when they were instead exposed to symbols of the United States, they endorsed “American values” more strongly.
Appealing to the demographics
Bringing these ideas together suggests a two-step approach towards using your demographic information to persuade customers:
- Activate or strengthen your customers’ affinity for a given demographic.
- Use that demographic to endorse a product, promotion, or advertisement.
Point #2 is perhaps the more straightforward of the two: targeted messages such as “X% of San Francisco men prefer Y” can fulfill this purpose, but the key is satisfying requirement #1 by making sure your target customer identifies strongly enough as a San Francisco man.
The first strategy is to simply pay attention to what you are selling and how people think about it. If one demographic category is particularly relevant, then that is the one you should probably be using!
Another strategy, taken from the bicultural studies, is to use priming, which is the jargon for using subtle cues to encourage customers to adopt a desired mindset. Whereas the bicultural participants were shown Chinese or American flags, a website might offer images emblematic of a customer’s location (e.g. the Golden Gate Bridge for San Francisco or the Statue of Liberty for New York) or of a past decade if the target demographic is age. These symbols will get your customers thinking about their relationships to the images (often without realizing it), which in turn can bring the desired identity to the forefront.
Disclaimer: priming is a widely used and generally effective way to influence how people think, but it is definitely far from foolproof. If you want to try it out, the key thing to remember is that you want to activate an idea in your customers’ minds to which they can attribute meaning. Try to keep it simple, and remember that images are not the only (or even necessarily the best) way to do this.
A third and perhaps stronger method is to have your customers actively endorse a demographic identity. This can be more difficult because it is more overt than the other two strategies. Displaying a question such as “How long have you lived in San Francisco?” serves as priming, but if you can entice your customers to actually answer it, their commitment to being a San Franciscan on your site will be much stronger. Answering questions, endorsing statements, or even the process of filling out demographic information to make an account may all be able to play into this approach.
One of my favorite examples of this kind of idea comes from a 2011 paper by Christopher Bryan, Greg Walton, Todd Rogers, and Carol Dweck. In one study, they recruited registered New Jersey voters for an “election survey” during New Jersey’s 2009 gubernatorial election. They either asked voters to endorse statements that were focused on behaviors (e.g. “I am going to vote”) or statements that focused on an identity (e.g. “I am a voter”). After the election, they looked at the votes and found that whereas the average voter turnout was 47%, if people had endorsed voting behavior then their turnout rate was 79%, and if they had endorsed a voter identity then their turnout rate was 90%. Just talking about voting was enough to raise the probability, but internalizing it was even better.
- Targeted advertisements based on demographics alone are not very effective.
- The effectiveness of demographic advertising could be improved by making the demographic known to your target customer.
- This effect can be further strengthened by taking steps to increase customers’ affiliation with the demographic in question.