Companies invest more in data tools to help their marketing teams make better decisions.
But marketing needs more than just data and analytics tools to implement an effective marketing strategy. They also need to understand the type of data they collect and how to analyze it to gain meaningful insights.
This involves going back to basics and understanding ordinal data, one of the key marketing data types. This article will explore ordinal data and how it informs data-driven marketing decisions.
Ordinal data is quantitative data in which variables are organized in ordered categories, such as a ranking from 1 to 10. However, the variables lack a clear interval between them, and values in ordinal data don’t always have an even distribution.
The level of customer satisfaction is an example of ordinal data. Its variables could be:
Using ordinal data, you can calculate the frequency, distribution, mode, median, and range of variables.
Having defined ordinal data, you may wonder about other data types, such as nominal, interval, or ratio data. How do they differ from ordinal data? Here are some quick definitions:
Ordinal data occurs in different formats. Here are a few examples of ordinal data and how to synchronize it with your business strategy to improve your data management efforts.
Whether you've already launched your product into the market or introducing new features to your existing product, you’ll need to conduct market research to ask questions to gauge your target audience's interest.
Market research involves analyzing both qualitative and quantitative data to understand customer needs, their buying partners, and what motivates them to buy from you. These insights can help improve your marketing campaigns in the future.
For example, if you host conferences regularly, surveys can help you know how well you did and whether your attendees want to attend the conference again. Here's an example of interest-level data:
Source: SurveyMonkey
The questions you ask will reveal potential customers’ interest level in your product or service. Interest levels range from not interested, slightly interested, neutral, to very interested.
This type of ordinal data provides insights into your target audience's proficiency level.
Education level may inquire whether your target audience has acquired different levels of formal education, such as high school, college, and graduate school. You may collect this data by assigning numbers to each level, such as 1 for no formal education, 2 for primary schooling, and so on, until 10 for a doctoral university degree.
Education-level data comes in handy when using analytics in your recruitment process to help you evaluate the job applications of potential candidates.
Educational-level data can help you make powerful predictions about who to hire in the future to support company growth, where to focus your recruiting efforts, and find suitable candidates for specific positions.
If you run a sales team, assessing the education level of your team members enables you to know how to support their career development goals. This way, you can build a high-performing sales team and improve retention.
Understanding the socio-economic status of your target audience helps create and refine your customer segments based on their demographic and psychographic profiles.
You can then rely on these segments when running personalized marketing campaigns that meet their needs and wants. Ordinal data on socioeconomic status for a B2C target audience includes gender, location, household income, marital status, and age.
On the other hand, data for a B2B target audience includes gross annual revenue, stage of business growth, number of employees, market position, and type of industry.
The satisfaction level reflects how content your customers are with different brand interactions. For example, your customer onboarding process or how well you resolve different customer issues.
Customer satisfaction may be expressed as extremely satisfied, satisfied, unsatisfied, or extremely dissatisfied. Satisfaction level data helps you gauge customer service and sales handling satisfaction to identify areas for improvement.
Here’s an example of satisfaction level data from a product-market fit survey that Buffer conducted:
Source: Buffer
With this data, the company could tell how useful Buffer’s Power scheduler is to their customers, meaning that the product was the right fit for their users.
This involves asking questions that reveal the similarities or differences between two or more data points. Once you identify the similarities or differences, you can learn what characteristics are similar, which ones are different, and the degree to which they’re different or similar.
For example, you may want to compare revenue performance from 2021 to 2022. Your comparison will yield significantly less, about the same, more, and significantly more for each year's revenue.
With this, you can gauge macroeconomic and industry trends and adjust your strategy to fit your budgeting process to control spending. You may even decide to take this further and compare industry trends so that you can create reports and write thought leadership content to drive brand awareness.
If you asked someone to rank their level of satisfaction on a scale from 1-5, their response would be ordinal. You can collect this data through surveys or Likert scales using survey software.
Surveys are one of the oldest methods for collecting ordinal data. You can use surveys to determine your target audience's feelings about products, topics, or specific issues related to your brand, product, or service. You can survey with many methods, including in person, over the phone, or online.
With surveys, however, collecting accurate data from people who don't want to answer questions honestly or understand them is difficult. Surveys also require a lot of time on the researcher's part to create, validate, and analyze them.
A Likert scale is a survey that asks participants to agree or disagree with each statement on the survey, for example, “I strongly disagree.” Participants then assign themselves an answer based on their feelings toward the statement and their level of agreement with it.
Likert scales improve clarity during analysis because respondents rate themselves on an ordered scale with clearly defined intervals, for example, a scale of 1-7.
To collect ordinal data, you must run surveys with questions that rank answers using an implicit or explicit scale. For example, if you have lots of traffic coming to your company’s website, you can use an enterprise website feedback tool to collect feedback from your website. Ask:
“How content are you with the blog post you just read?’’
The possible answers could be:
You can conduct several tests on ordinal data to measure the difference between two or more groups. These tests include:
Ordinal data is a data type that ranks values from least to greatest. In other words, ordinal data is ranked or ordered.
The Kruskal-Wallis test is a non-parametric test used to compare the medians of three or more independent groups. It’s used when the data is not normally distributed, and the variance between groups is unequal. The Kruskal-Wallis test can also compare two dependent groups – before and after pictures of a website redesign.
The Mann-Whitney test is a non-parametric test used to compare the median of two independent samples. It can be used when there's ordinal data, such as ratings on a scale from 1 to 5, or when there are no clear groups in the data.
The Wilcoxon signed-rank test is a non-parametric test that can be used for data sets with or without normal distribution. It's an alternative to the t-test in cases where the data doesn't have a normal distribution.
When running a t-test, the assumption is that the underlying distribution of the data is normal, but this assumption can be wrong.
For example, when testing the difference in height between two groups, let's say one group has an average height of 180 cm and the other group has 170 cm. You won't see any significant difference in their heights.
However, using the Wilcoxon signed rank test, you can see beyond the regular difference in their heights.
The test is based on the premise that people's moods cluster around a median point, with some being more positive or negative than others. The Mood's median test often measures how individuals feel about an issue or idea, such as your customer’s opinion about your products or service. It can predict behavior based on their moods, such as whether your customers will buy from you or your competitors.
There are two ways to analyze ordinal data: inferential and descriptive statistics.
Descriptive statistics summarize the characteristics of a dataset and identify patterns. Here are the descriptive statistics for ordinal data:
Inferential statistics, on the other hand, predict what may happen in the future based on the data you have. You can use ordinal data to collect insights, create hypotheses, or even draw conclusions with the four tests described above.
The Kruskal-Wallis, Mann Whitney U, and Wilcoxon signed-rank sum tests all analyze ordinal data. They're all nonparametric tests, meaning they don't rely on any assumptions about data distribution.
Descriptive analytics collects, analyzes, and reports data about events that have already occurred. This differs from predictive analytics, which predicts future events based on historical data. Descriptive analytics helps businesses identify patterns in the past to improve their future decision-making.
In descriptive analytics, the goal is to find patterns in existing data, not predict what will happen in the future. Descriptive analytics aims to find cause-and-effect relationships between past events and use these relationships to predict future events.
Unlike other analysis methods, descriptive analytics can be used anytime with any data available. This makes it more accessible for smaller businesses with insufficient resources for predictive models or large datasets required by other methods.
Bars and graphs present data in a way that's easy to comprehend. They're beneficial when the data is too large or complicated to be displayed in a table. The type of graph you choose depends on the amount of information you want to convey, the data dimensions, and your audience.
Bar graphs display information as bars with lengths proportional to their values. They're used when the data is categorical, meaning it falls into specific groups. Here’s a bar graph for a call center, showing the time taken to respond each weekday:
Source: Datapine
They're a good choice when you want your audience to be able to compare values easily. Bar graphs are more intuitive and easier to understand as compared to numbers. It's also possible to use bars in combination with lines or other graphics, like scatter plots, histograms, or pie charts.
Line graphs are used when the data has an ordered value. These graphs use lines to connect points on two axes with the same scale on both sides. These lines can be solid or dotted and start at any point on either axis.
The lines represent change over time, such as how the stock market fluctuates daily or how the cost of energy changes year by year. Here’s an example of monthly inbound leads over 12 months visualized using a line graph:
Central tendency is the average of a set of numbers. It measures how closely the numbers in a data set are clustered around their mean.
Three main types of central tendency are mean, median, and mode. The most common measure of central tendency is the arithmetic mean, calculated by adding up all the values in the data set and dividing this sum by the number of values in that data set.
The median can also be used as an alternative to calculating central tendency, simply finding the middle value in a data set after arranging all numbers from low to high. The mode is the most frequent value in a set.
Ordinal data is more complex than nominal data and commonly used to gauge interest. The Likert scale is a popular ordinal data example.
Use some of the real examples provided here to inspire your own survey data collection. While at it, learn more about polling and how it helps collect data.
There was a time when marketing primarily consisted of guesswork and creativity. While...
by Atul Jindal
Every successful content strategy starts with data.
by Marta Szyndlar
Many companies are shifting from inbound to outbound ABM approaches – and for good reason.
by Kristina Jaramillo
