Using Big Data to Better Serve Customers, Donors and Prospects
You've completed an analysis of your customers or donors. You know which customers are profitable. Or, if you are a nonprofit, you know which donors are most supportive of your cause.
But you still have an incomplete picture of those individuals as people. Where do they live? Where do they shop? Where do they work? Who are their friends? How do they spend their free time? With what organizations are they affiliated? What other perhaps related products do they buy? In short, you still need an understanding of their demographics and behavioral traits. How else can you go out and find more of these profitable customers? It is on these occasions that Big Data becomes the marketer’s friend.
Here's how the process works. Take email addresses, phone numbers, names, social security numbers, or other identifying information. With permission, of course, use this information, you do know to reference more data in a data vendor's warehouse about what you don't know. Tap the database to find information about essential attributes like age, income, mailing address, location, driving record, credit, or assets. Select the variable based on what is useful to your business intelligence needs. Analyze and summarize those attributes to inform your marketing strategy and boost marketing performance.
By enabling this kind of analysis, Big Data allows marketers to understand customers deeply. Even better, with predictive modeling, marketers can even predict how likely customers or users are to buy or donate in the future. The Gartner Group surveyed its Gartner Research Circle members worldwide. The technology research firm found that most (73%) of those surveyed in its customer base planned to use Big Data in the future.[i] The bottom line? Smaller organizations are going to have to develop similar information to compete.
A Toe In The Water --- Social Media Custom Audiences
Facebook and LinkedIn provide small business advertisers alternatives to buying data about customers. Instead, Facebook advertisers can upload CSV files with email addresses of customers. If the emails uploaded match those of their members, Facebook will place advertisements in the members' timelines. Similarly, LinkedIn allows users to provide lists of company names. Marketers can advertise to LinkedIn, its members who work for those companies. See information on LinkedIn’s Service here.)
It gets better. With the Facebook custom audience tool, marketers develop ads for “lookalike” audiences. These audiences have similar demographics to those of existing customers. So, identify a list of profitable customers, donors, or B2B companies you currently serve. Upload them to the appropriate social media platform. Then, let their data scientists take care of the rest. Cool right?
The behind-the-scenes magic that has allowed social media marketing to work so well for Facebook has been a database of 98 attributes. Facebook worked with its data partners, companies like Experian, Epsilon, and Oracle Data Cloud. They helped Facebook complete its detailed knowledge of us. We members give Facebook our age, gender, location, education, relationship status, and employment. We do this to improve our connections with former classmates, colleagues, and friends. Our likes tell Facebook how we feel about many topics. Yes, Facebook knows our politics, our hobbies, our true loves, our family, and even our musical tastes.
To this, Facebook’s Data Partners program concatenated information from large data providers. How many credit cards do we have? What kinds of restaurants do we frequent? How likely are we to move? Are we likely to buy a new car?
The Facebook data and target market interface gave even the smallest businesses the targeting power of big companies. For small businesses, gaining affordable, easy access to this kind of data was revolutionary. It democratized Big Data. A small restaurant or hotel had access to the same information as so its national chain competitors across the road.
That was true until recently. The controversy over the Cambridge Analytica Affairdragged social media executives before Congress. Other challenges included the new Privacy Rules in the European Union (GDPR). Facebook announced it was phasing out access to its Data Partners’ this year. (See Facebook’s short Partner Categories Article here.)
Despite the change in policy, we think Facebook will continue to be an excellent advertising resource for small businesses targeting consumers. LinkedIn remains the most effective way to target B2B customers, by industry and position affordably. Both platforms will allow small companies to pinpoint critical segments in their customer base affordably.
Large Data Purchase - 100 Attributes for Under $1.00
Plan B is to buy your own demographic or propensity data. And there is lots of it available out there. A typical database vendor such as Acxiom, LexisNexis, Choice Point, or Trans Union has hundreds of pieces of information for sale. This information is gathered from many providers and sources, then cross-indexed and cross-referenced. Here is a link to one provider, USADATA (also branded as MyAcxiomPartner.com). USA DATA has a self-serve data mart targeting small and medium-sized businesses. We expect more small business-focused offerings from other large data providers in the future. Make no mistake. The market will find a way to fill the void left by Facebook’s policy changes.
Big Data - A Fundraiser's Life Blood?
Big Data is also crucial in the nonprofit fundraising sector. CRM vendors to nonprofits offer a series of services using Big Data. These services include data cleaning through the US Postal service relocation database. They also provide wealth indexing, and Social Security sweeps for recent deaths.
An example is Blackbaud Group's Raiser's Edge product.[ii] The Raiser's Edge Product and ResearchPoint databases are integrated. This integration allows fundraisers to access information about prospective donors such as net worth.
Want more? They can access even more granular information as well. What securities do they own? What real estate properties do they own? What are some of their other sources of wealth?[iii] The Research Point integrated data service also provides data about a prospect's past giving. It also can provide information on prospective donors' Board Relationships. The software even identifies people who live near the prospect who might be donors as well. For fundraisers seeking high net worth donors, these kinds of tools can be invaluable.
Computer, "Go find this customer’s VIN, please."
Insurers, banks, and other financial services institutions also use Big Data. Big Data allows them to make quick decisions about selling us insurance or lending us money. Insurance companies can buy underwriting data about prospective customers and can use that data, so the customer doesn't have to remember it or enter it themselves. Here’s an example of how consumer database information was applied to improve customer experience.
A client in the auto insurance industry was having difficulty keeping customers engaged. Customers had to fill out 100-question application forms to purchase auto insurance. Users couldn't for the life of them remember one piece of information needed to accurately quote the risk. Who do you know who can remember their vehicle identification number (VIN)? (The VIN is a 17-digit alphanumeric number appearing on the left of corner of the windshield.) Our client studied drop-off of users. Many who were requested to provide their vehicle identification numbers departed, never to return.
It was clear. The client could increase the number of completed applications. How? Get the VIN from data warehouse vendors. Don't force the customer to provide the information she doesn't have easily accessible. Instead, pull the data out of the cloud. Then ask customers to confirm a picture of their make and model. Large insurers are learning that the best way to pull customers through the online sequence is to purchase as much data as they can. Recent competition among auto insurance sellers to have the shortest quote time is an example of this trend.
Here are two more examples from a recent article in Inc. Magazine. Carvana, a mid-sized and entrepreneurial used car dealership, employs five data scientists. They held an online Kaggle competition to solve a complex analytical problem. Carvana asked the data science community for help in developing a better way to predict whether cars purchased at auction were lemons. The data model allowed Carvana to buy higher-quality cars "for $500 below what similar cars would sell for." The same article shared how a zoo in Tacoma, Washington, was able to save money on staffing. Zoo management correlated local climate data with attendance peaks and valleys. The resulting analysis allowed them to build a better staffing model, saving the organization thousands. The zoo later improved membership by 13 percent by upping its marketing spend in the zip codes of its most frequent guests.[iv] All of the preceding was thanks to the application of Big Data to small/medium-sized business problems.
Why Math Is Our Friend
So how do the professional data 'wranglers' do this? Does your company have large amounts of data about its customers? Can you afford to purchase more data about them? If so, here are some examples of techniques from a favorite article by two FICO data scientists in Analytics Magazine. Even if you aren't the one doing the analytics, you can benefit from understanding how these analytical techniques apply to your marketing. We summarize the methods in the below infographic; then explain them in more detail in the body of the article.
Clustering Algorithms offer the opportunity to create broad behavioral segments. Examples are animal lovers, people who enjoy traveling by RV, or people who buy only local farmer produce. These clusters bring together large numbers of customers who may be different in other ways, but who may have similar beliefs or tastes. Quite often, retail stores will use this kind of segment information to create decors. Marketers can also use it for product groupings, or in-store departments targeting this segment. They use this technique for cross-selling products. For nonprofits, this kind of segmentation offers obvious opportunities to identify like-minded individuals. These individuals may believe in the non-profit's mission. But they may not have had the chance to learn about the organization. Here's another use: translate these segments into keywords. Then, use the keywords to improve online advertising yields and organic search.
Propensity models are predictive models. These models bring together an extensive line of attributes which are predictive of buying or giving behaviors. For example, Scottish Terrier owners who live in urban environments may be more likely to donate to specific animal-related causes. Sometimes wacky, these predictive models have become a keystone analytic of direct-mail organizations. More recently, online marketers study clickstream data to try to identify when people are ready to engage in a valuable activity. For example, they may be prepared to make a significant purchase or perhaps, donation. In a world where constant marketing noise is the norm and in which client acquisition costs are soaring, these models can move the numbers. They do this by predicting who will buy or donate so that we can give them a nudge at that golden moment.
Collaborative Filtering is a way of taking data about existing customers and applying the data to prospective ones. Often called a "look-alike model," companies can use these models on the Internet to find new customers. Facebook does this with its lookalike audience feature discussed above. Amazon applies this technique with the "customers who bought this item also bought."[v] The method is also used by entertainment media companies. TiVo Suggestions. Netflix, Yahoo!, and Apple's Genius also reportedly use it to make their decisions.[vi]
A data scientist named Jonathan Goldman, a physics Ph.D. by training, joined LinkedIn in 2006. Legend has it he developed a module that allowed users to see the names of people to whom they were, "Not connected but were likely to know."[vii] The technique was an immediate success. The discovery reportedly threw LinkedIn into hyper-growth.
Flunked Stats in College? Here's Technical Help for English Majors!
Software as a service solutions (SaaS) offerings is emerging that incorporate many of these analytical capabilities. A Canadian company named Canopy Labs, for example, provides an online tool-set for marketers. The tools are designed to allow customers to consolidate consumer data and analyze the “customer journey.” The software includes different kinds of statistical analyses.
Another company, called Retention Science provides a similar online platform. Retention Science claims data cleaning, analysis, and marketing automation on its platform. The software boasts behavioral and transactional profiling. It also offers customer life cycle analysis, and analytics to measure campaign performance. Insight Squared has a product that integrates with QuickBooks, Google Analytics, or Zendesk. Even the mighty IBM’s talking artificial intelligence avatar, “Watson” has gotten into the act. An analytics package that starts free and grows with your knowledge, and the value created is available under the Watson brand. Thanks to IBM's big Watson branding advertising push, senior management understand what you are trying to achieve!
Solutions like these greatly simplify the analysis. Other solution providers are emerging daily, so shop around. They usually provide engagement teams with human beings who you can call and ask questions about what the data is telling you.
Professional Advisors – Ask them to focus on ‘Keeping It Real.’
Like everything else, there is the opportunity for a great deal of puffery in the analysis and interpretation of data. If a service provider overwhelms you with a long concatenation of buzzwords and hard-to-understand terms, find someone else. A plain-spoken expert can do the same work! Our recommendation is to keep it simple, go one step at a time, and get analytical help as you can. Quite often, a local university statistics professor or graduate student would be glad to help. They love a real-life problem on which to test their latest models. If your company's a little bigger and can afford to hire a data wrangling consultancy, don't be shy about hiring the right talent. Look for someone with a strong track record with clients like you. Make sure they provide references. Have them include Ph.D.’s as well as a few liberal arts-trained executives. Make sure they have led to real value-added for clients like you.
Do you know everything you should about your profitable customers? If not, it’s time to find out!
[iv] Kelleher, K. (2017). What 3 Small Businesses Learned from Big Data. Retrieved from https://www.inc.com/magazine/201407/kevin-kelleher/how-small-businesses-can-mine-big-data.html
[v] New Advances in Intelligent Decision Technologies: Results of the First KES International Symposium IDT'09 (Google eBook) Kazumi Nakamatsu, Gloria Phillips-Wren, Lakshmi. Jain, Robert J. Howlett Springer Science & Business Media, Apr 28, 2009 - Computers - page 103.
[vii] Davenport, Thomas H., and D. J. Patil. "Data Scientist: The Sexiest Job of the 21st Century." Harvard Business Review 90, no. 10 (October 2012): 70–76. See https://hbr.org/2012/10/data-scientist-the-sexiest-job-of-the-21st-century/