Big data, big insights?
61% of companies state that Big Data is driving revenue because it is able to deliver deep insights into customer behavior. For most businesses, this means gaining a 360° of their customers, by analyzing and integrating existing data.
Well, not always. Big data doesn’t always equate to big insights, but, increasingly, we see big data driving action. And, big data-driven problems.
Currently, you’ll find debates all over websites and social media — some arguing that big data is causing big problems, not the least of which is perceived manipulation of consumers by using psychometrics against them. Others see big data ushering in and era where businesses have the analytics they need to optimize their returns on investment (ROI).
Big data, big insights
I recently wrote a post defending big data, so you know what side of the debate I’m gonna take. Data is never the problem, in my opinion, it’s the way you use big data that causes big problems.
Here are the major problems cited related to big data that we’ll discuss today:
- Manipulation of consumers
- Violating the privacy of consumers
- Overwhelming volume, velocity, and variation
- Poor decisions
Let’s chat about these problems and how to fix them.
Manipulation of consumers
This week, the social world was in an uproar over allegations that the Trump campaign used psychological profiles of INDIVIDUAL users to target messaging — if you missed the discussion, here’s the link to a great article summarizing how using psychographics changed (?) the election outcome.
Now, let’s take this discussion out of the political arena and into the business world. Does using psychological profiles to craft individualized messages to prospective buyers violate some ethical compact?
We had this discussion in my digital marketing class yesterday and students seemed split on this question.
First, let’s understand that using psychographics (likes, values, attitudes, etc) isn’t a new thing — it’s been around for a while. What’s changed is the amount of information we have because social media records every interaction — like, post, share, etc. The article above makes the point that, with just a few of these, you can categorize the personality of users. More information and you can determine more detailed psychological traits, like view of authoritarians. Again, useful information.
In fact, if you’ve ever bought Facebook ads, you’re using one of the same databases used by the Trump team.
One-on-one marketing also isn’t new. With technological advances, we can reach increasingly smaller groups with individualized messaging. And, that’s valuable, according the HBR.
Practiced correctly, one-to-one marketing can increase the value of your customer base. The idea is simple: one-to-one marketing (also called relationship marketing or customer-relationship management) means being willing and able to change your behavior toward an individual customer based on what the customer tells you and what else you know about that customer.
One-on-one marketing is just an extension of segmentation and this article highlights how using personality segmentation aids marketing efforts. It reflects that not all consumers want the same thing from products, so helping them see factors they value in your products improves market performance.
The ethical problem arises with the messages you send. If they are untruthful or misrepresent the true impact of you product, then it’s unethical regardless of whether you used psychometric evaluations to get you there. For instance, if you know a consumer is particularly interested in sustainability, it’s perfectly fine to send them messaging promoting your record on sustainability. It’s not ok to lie when you’re not, in fact, sustainability driven, or to highlight one positive action among a host of negative actions.
For the most part, marketers aren’t hacking into sources of information, but accumulating mountains of publicly available information. The problem occurs when these sources indiscriminately sell PII (personally identifiable information). Here’s a partial list of where this data comes from:
- Government sources such as DMV and property records
- Educational institutions — such as schools
- Biometric devices and other smart devices, such as fitbits, Nest
- Social networks
- Credit card transactions
- Gmail and other online sources, including browsers
The breach of ethics occurs when this information is obtained fraudulently by companies who aren’t vetted by the provider of the information or are using it for a purpose different from the one stated when obtaining the information.
As a consumer, there’s little you can do to protect your privacy other than appeal to congress for new laws. You can attempt to stay “off the grid” but you’ll have limited success. That said, you should still close your social profiles to only connections, use search engines like Duck Duck Go that don’t share information, and avoid devices that share PII internally or with other companies.
Overwhelming volume, velocity, and variation
This is almost the definition of big data, so using big data, big insights is challenging. A few years ago, businesses dealt with the 3V’s of big data by data mining to detect insights. Unfortunately, data mining turns up too many spurious correlations to be of much use.
Courtesy of Information Management
Visualizations, like the one on the right, are much better tools for dealing with big data and don’t rely so much on spurious correlations. And, here’s a list of some of the best tools for data visualization.
People just can’t evaluate numbers well. That’s why creating an image that depicts the data is so valuable. Here are some other great data visualizations.
Image courtesy of Data Mentors
Color really helps a lot in helping generate big data, big insights, especially when you use colors that have an inherent meaning like red for bad, green for good, or to draw attention to the most important features of a graphical design.
Big data, big insights doesn’t always happen. Sometimes marketers make poor decisions despite using big data to guide the decisions.
Sometimes, data isn’t able to predict something because your assumptions of the attitudes and behaviors underlying the data are wrong. For instance, Google thought it could predict flu outbreaks by mapping searches for terms related to the flu. They were wrong. Likely, what they were mapping was more a function of neuroticism than actual flu outbreaks.
Sometimes, you make poor decisions because you don’t have the right data or a complete set of variables to predict the most appropriate course of action.