I spent the morning at a workshop hosted by SPSS (an IBM company) — thanks a bunch for the invitation. Marketing analytics with SPSS sure have changed in the last 20 years since I started using them — sure there’s still regression and logit and discriminant, but there are lots of new tools, too.
Today, I’ll give you an overview of SPSS and the cool new modules and features of SPSS.
What is SPSS?
SPSS stands for the statistical package for the social sciences and is one of two commonly employed methods used for quantitative data analysis. The other is SAP. Back in the day, SPSS and SAP were hard to use as your analysis was coded using Fortran commands so it took a long time to learn the software. But, for a decade or two, SPSS existed as a point-and-click analysis software that reduced the requisite of learning some basic Fortran commands. While this user-friendly interface was a great advancement, it also meant it was subject to abuse from analysts who didn’t know what they were doing. Plus, interpreting the output still requires some skill and you still face the age-old issue of garbage in/ garbage out resulting from poor questionnaire construction, sample selection, and bias.
A relatively more recent entrant into the analytics software game is R, which is open-source. However, learning R still relies on learning the syntax so it takes a significant time to learn how to use R, in contrast to SPSS or SAP. The most recent edition of SPSS, version 29 offers some very cool tools, which we’ll discuss at the bottom of this post.
Through development or acquisition, SPSS has a full line of tools for predictive analysis from data planning and data collection through analysis, reporting, and deployment of solutions. You can enhance SPSS with additional modules such as R and Python (another open-source software program designed to do task automation and analysis — Python runs Microsoft Excel in the background, for instance). This diversity of tools and the ability to buy modules individually allow customizable solutions for a variety of business and non-profit problems. A valuable feature of all the SPSS software is that the various modules fit together seamlessly and share a similar interface, making it easy to move between modules and reducing the learning curve. The products also interface with other software products such as Excel.
To increase affordability, SPSS is now available through a subscription that’s much less expensive than the initial investment of buying the software if you don’t need to use it routinely. Another big advantage of choosing SPSS for your business analytics requirements is that many students train on this software in their undergraduate statistics classes (SPSS provides a lower cost version of their 3rd tier product to schools at a reduced cost), so you’re likely to find folks skilled in this tool.
Marketing analytics with SPSS
The standard version of SPSS includes the following:
- descriptive statistics
- T-tests, chi-squared, etc
- cluster analysis
- linear regression
- some other basic types of analysis
With the advanced statistical extension, you also get:
- multivariate analysis of variance
- non-linear regression
- logic and probit regression
- weighted least squares analysis
- some other more complex statistical analysis types
The complex testing model allows researchers to complete the following:
- conjoint analysis
- neural network analysis
- time-series analysis
- decision trees
- and more complex analysis types
IBM also offers a more sophisticated version of marketing analytics with SPSS to help researchers understand marketing trends.
SPSS offers a host of descriptive statistics to help the researcher understand their data through traditional statistical tools such as mean, standard deviation, median, and variance. The tool reports these statistics as tables and graphs you can easily import into your reports. Most researchers start with descriptive statistics as a tool to aid in data cleaning and then follow up to express the composition of their respondents.
While relatively simple, descriptive statistics are powerful. For instance, if you want to understand which groups represent your best customers, you can collect customer data to see how the demographics and geography break down. If you discover that most of your customers are between 25-35, make between $30 and $50K, are married, and live within 10 miles of your location, you can adjust your marketing campaigns to appeal directly to these consumers to improve the efficiency and ROI of your advertising.
Of course, this is really correlation analysis rather than true causal analysis but these tools allow researchers to predict outcomes to make better decisions. For instance, you can use regression analysis to forecast sales for the next quarter based on a mixture of variables that impact sales including economic variables such as inflation and interest rates, past sales figures, competitive intensity, and other variables found to impact sales based on past performance.
In causal analysis, the researcher commonly starts with an assumption regarding an important relationship, called a hypothesis. For instance, you might posit that spending $10,000 on a particular marketing campaign will result in an additional $50,000 in sales. You propose variables, called independent variables, that might impact the variable you want to predict, called the independent variable. Then, using an analytic tool, such as regression, you collect historical data from either internal sources, secondary data, or surveys for both independent and dependent variables.
HBR, for example, used causal research to discover that a 5% increase in retention by a firm increased profits by 25% to 95% depending on the firm. Your chosen analytic tool depends on the type of relationship you expect, your research question, and your data. For a linear relationship, regression might provide the best results. If you want to understand how much more a customer is willing to pay for a particular feature, conjoint analysis is the right choice. If you want to understand how groups vary from each other, ANOVA is your likely choice.
Of course, there’s a downside to causal analysis. Spurious correlations (correlations that have no real meaning) lead to bad decisions. An example is women’s hem lengths which vary based on the economy. There’s no logic that relates these two variables and you can’t improve the economy by lowering or raising the length of women’s dresses. The relationship uncovered might result from a 3rd variable, consumer confidence.
Marketing analytics with SPSS: new analytics features
Competing with more established products like nVivo and Hyperresearch, marketing analytics with SPSS developed a text analysis program that uses native language to search a variety of textual materials to find patterns in word use. Since MIT estimates that between 80 and 90% of the world’s data is unstructured (ie. text, image, etc rather than quantitative), businesses need powerful tools to evaluate this data. Forbes estimates that the amount of unstructured data grows at a rate between 55 and 65% year over year. For marketing analytics with SPSS, a text analysis tool is invaluable in understanding social media insights such as unmet needs that suggest innovations, sentiment analysis, and much more.
According to IBM, their product for text mining,
Text mining, also known as text data mining, is the process of transforming unstructured text into a structured format to identify meaningful patterns and new insights. By applying advanced analytical techniques, such as Naïve Bayes, Support Vector Machines (SVM), and other deep learning algorithms, companies are able to explore and discover hidden relationships within their unstructured data.
Text mining and text analytics use natural language processing (NLP) to understand the nuances of textual material, such as social media, transcripts of customer service calls, etc. Of course, the English language is very nuanced and printed text lacks the non-verbal parts of communication that enhance a researcher’s ability to interpret meaning. Other languages are even harder to evaluate as they depend not only on non-verbal cues but context. An example comes from an Asian culture where harmony is important. In this culture, yes might signify agreement but it might also simply show understanding in an effort to maintain harmony.
Basically, the software breaks down text into categories using NLP into predefined topical groups. The categories are then collapsed into thematic patterns that are then used for marketing analytics such as sentiment analysis. Although a single attempt at sentiment analysis is fraught with problems due to the limitations of NLP, conducting sentiment analysis over time does provide insights as it takes the problems out of the equation to an extent.
Some of the uses for text analytics beyond sentiment analysis include:
- Identifying serious pain points to prioritize service and reduce dissatisfaction
- Discover industry trends by sifting through industry reports and similar documents
- Uncover the root causes behind operational problems that recur over time by revealing patterns leading to failures
- Spam filtering to improve the overall customer experience
By combining the IBM text analysis module with Watson’s trained NLP, you can derive insights faster and more efficiently.
Modeling software (in this case the SPSS Modeler module) allows users to analyze vast amounts of data (big data) searching for unexpected patterns hidden within the data. How does Modeler differ from basic SPSS data analysis such as regression? The answer is that it makes no assumptions regarding the relationships between variables while basic analysis requires the researcher to make at least a superficial assumption as to the relationship between variables, for instance, whether it’s linear or non-linear.
Big data is characterized not only by its volume but the velocity associated, which means your data isn’t stagnant. It comes at you in real-time, fast. Big data also takes many forms including structured, unstructured, and semi-structured data, as you can see in the graphic below.
Thus, we talk about the 5 V’s of big data:
- variety in that data comes from a variety of sources in a variety of formats. Google Analytics recently changed its tool, GA4, to collect data from a variety of Google products, including websites, the app store, Google Ads, and more into a single platform to aid in analysis across these data sources.
- value implies that not all data is valuable to the organization. In fact, unless the organization can derive insights from the data, it represents a cost rather than value.
It’s this last element that can bite you in the butt because not all the data collected is true. You might have missing data, data entered incorrectly, or data mistakes across relational databases, such as transposing from different lines. We’ll discuss this issue in a few minutes.
Companies use data mining for a variety of tasks, including:
- predicting which customers are most likely to buy a particular product
- improve ROI by detecting fraudulent transactions to decline them
- implement dynamic pricing to gain sales and profits
- recommendation engines to increase sales and average order size
- optimize routes to deliver products more quickly
- optimize product offerings and inventory to match customer needs
- create marketing campaigns that resonate with customers
Of course, big data can also obscure events to reduce insights when used indiscriminately. For instance, analyzing a customer’s credit card purchases might confound their purchases with gift buying rather than reflecting their own needs. You must ensure you account for these confounding factors to get clear insights.
Additional analytics tools
IBM recently acquired Cognos, which provides business intelligence (BI) to track a firm’s Key Performance Indicators (KPIs). Cognos software provides highly customizable marketing dashboards to increase understanding of the data and display it graphically.
In addition, SPSS offers new modules to handle direct marketing, which allows customized offers to customers based on past behavior, as well as calculates Customer Lifetime Value (CLV), which is important for grouping customers and determining how to market to them. There are new analytics tools to handle missing values and other data preparation tools to remove outliers, etc. SPSS has also acquired AMOS, one of the top software packages for Structural Equation Modeling (what used to be called Causal Modeling) used for simultaneously evaluating constructs and relationships between constructs to test complex models.
Need marketing help to support business growth?
We welcome the opportunity to show you how we can make your marketing SIZZLE with our data-driven, results-oriented marketing strategies. Sign up for our FREE newsletter, get our FREE guide to creating an awesome website, or contact us for more information on hiring us.
Hausman and Associates, the publisher of MKT Maven, is a full-service marketing agency operating at the intersection of marketing and digital media. Check out our full range of services.