Rae Yule Kim
Data is increasingly important for user experience (UX) designers. One example of a data-driven UX strategy is the personalized recommendation algorithm. As one of the leaders in this technology, Amazon collects customer data about their shopping and search patterns to derive personalized recommendations, which, in turn, optimizes user experiences and maximizes the chances of sales. Similarly, Netflix and Zillow have been two of the biggest beneficiaries of personalized recommendation algorithms for growing their businesses.
→ Data is a valuable resource for UX design.
→ Data provides insights into behavioral patterns that are not easily observable.
→ UX designers can use data to make informed decisions.
Customer data provides valuable insights that can inform decision making and improve operations. About 89 percent of businesses say that they compete on customer experience (CX) . Data-driven UX design can improve sales and customer satisfaction.
By analyzing data, UX designers can make more-accurate predictions about what users want based on past trends. Sometimes, they can learn about unforeseen competitive advantages from customer data, which can be used to enhance customer experiences by tailoring the products and services to meet their needs. In addition, UX designers can better mitigate an unforeseen risk by detecting an anomaly in user satisfaction before the damaging trend becomes too big to manage. Overall, when used adequately, data can help UX designers make better decisions, optimize operations, and manage risks.
Due to the benefits of utilizing data for UX optimization, many organizations look to develop a data-driven culture. To enable data-driven UX design such as personalized recommendation algorithms, multiple business departments should collaborate to generate and analyze data.
A data-driven culture indicates a strategy where organizations use data to optimize business operations, including UX design. In order to find patterns, trends, and insights that can guide decision making, multiple departments within an organization should collaborate to gather and analyze data, including customer data and point-of-sale data. Data-driven organizations utilize various types of machine learning techniques to deliver insights from patterns in customer behavior and business performance.
The first step in creating a data-driven culture is to convince business stakeholders of the benefits of using data. Poor change management, often because of business stakeholders who are unmotivated to rearrange business processes to generate and analyze data, is one of the main reasons why some organizations fail to become data driven . Here are several benefits of becoming a data-driven organization:
Data-driven organizations perform better. The main theme in making data-driven decisions is "learning from the past." Data-driven organizations can make informed decisions based on past mistakes and successes.
For example, companies that personalize customer reach and interaction based on customer data grow 40 percent faster on average compared with those that do not . Analyzing customer demographic data and common patterns in customer interactions can provide insights into the wants and needs of a specific customer demographic, which can then be used to optimize customer experience and also sales.
Data-driven organizations understand their strengths better. Data-driven organizations can gain detailed insights into customer preferences and behaviors. Analyzing customer characteristics and behavior patterns can help UX designers discover unforeseen patterns and exploit them to optimize the user experience.
For example, Walmart data scientists have observed that sales of beers and diapers are highly correlated on Friday evenings. It turns out that men go to Walmart on Friday evenings to get beer and often get diapers too as per their wives' requests . Data can provide insights into unforeseen behavioral patterns. It is up to UX designers to figure out why patterns occur and use the insight to improve user experiences.
Data-driven organizations are resilient. Collecting and analyzing data continuously enables detecting anomalies in UX key performance indicators (KPIs) and managing risk before it becomes a more significant threat to the organization. UX designers can better detect the cause behind negative trends in user experiences by tracking KPIs based on customer demographics or user interfaces.
The next thing is to lay out deliverables by implementing the data-driven process. The main goal of implementing the data-driven process is to generate ready-to-use datasets that can be used to derive insights and improve UX/CX. To do so, an organization should identify the KPIs clearly.
Customer data. One of the most common types of data that UX designers use is customer data. It includes demographics, purchasing transactions, search history, browsing activities, preferences given in surveys, and more.
UX designers can utilize behavioral data, such as click-through rates (CTR) and conversion rates, and survey data, such as customer satisfaction ratings, as the KPIs to examine the effectiveness of UX strategies. The most common method to examine customer data is A/B testing. UX designers can compare changes in KPIs before and after they implement a new UX strategy.
Behavioral data is the most controversial type of customer data, as issues with privacy and data security have been raised around the collection and exploitation of people's behavioral patterns. Many organizations are now making efforts to ensure that they are gathering consumer data ethically and using it responsibly and transparently.
Finance data. Finance data records macro-level business performance metrics. It includes a wide range of financial transactions and activities, such as revenue, market share, stock prices, earnings, and more.
Finance data has been largely irrelevant to UX design; however, it may provide clues to important behavioral patterns. For example, the correlation between beer and diaper sales on Friday evenings reveals an important shopping pattern that can be utilized to optimize the store layout .
Business data can be further classified into qualitative data and quantitative data. Qualitative data, such as customer interview scripts, has not been as popular as quantitative data because it is challenging to draw measurable KPIs from text data. Text data can be quantified into sentiment scores by applying text analytics algorithms. One key benefit of using qualitative data to track customer sentiment is that sentiment is less vulnerable to extremity bias compared with survey scales. For example, online review ratings tend to be skewed to extremely negative and extremely positive values. Online review sentiment, however, is often normally distributed .
Data that is collected by business organizations is often quantitative. UX/CX KPIs are often recorded in numbered scales. Also, certain behavioral data such as whether a customer has made a conversion (1) or not (0) or clicked on the ad (1) or not (0) can be quantified in a binary format. The main reason why KPIs are quantitative is because quantitative data can be easily tracked and compared.
Quantitative data can be further classified into three broad categories: continuous data, discrete data, and categorical data.
- Continuous data. Continuous data refers to numerical, ordinal data. Finance data such as revenue, market share, and stock prices and scalable survey responses such as UX/CX KPIs are examples of continuous data. This type of data can be compared and used to track changes and trends.
- Discrete data. Discrete data refers to numerical, ordinal data that takes on certain, distinct values. For example, whether a customer has made a purchase (1) or not (0) can be recorded in numerical values that are distinct in binary formats—zero or one—and thus not continuous. Noncontinuous survey scales, such as customer engagement levels, can be discrete data. Customer engagement levels might be ranked and measured in three distinct values: 1) have never heard of the product, 2) have considered purchasing the product, or 3) have purchased the product. Due to the lack of variance and continuity in outcome measures, discrete data should be evaluated differently from continuous data, such as by using discrete choice models .
- Categorical data. Categorical data can be numerical, but it is not ordinal and simply indicates a categorical value. Zip codes, race, and gender are examples of categorical data. Zip codes are numeric, but they are not ordinal and do not indicate trends. Categorical data can be used to classify data and examine patterns by comparing the KPIs by demographic category such as average expenditure, gender, and neighborhood.
Data-driven organizations enable UX designers to discover behavioral patterns and make more-informed decisions to improve user experiences. This article discussed several benefits of data-driven UX design and types of data that UX designers can use to derive insights and improve their customers' experiences.
1. Arora, N. et al. The value of getting personalization right—or wrong—is multiplying, McKinsey and Company, Nov. 12, 2021; https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/the-value-of-getting-personalization-right-or-wrong-is-multiplying
2. Dykes, B. 10 reasons why your organization still isn't data-driven. Forbes. Jun. 1, 2021; https://www.forbes.com/sites/brentdykes/2021/06/01/10-reasons-why-your-organization-still-isnt-data-driven/?sh=4718e6d47d80
3. Salehi, B. UX and data science teams need to work together. Built-in. Feb. 23, 2023; https://builtin.com/design-ux/ux-data-science-work-together
5. Douglas, A., Segmentation and discrete choice models from market research data. Medium. Feb. 26, 2020; https://medium.com/fulcrumanalytics/segmentation-and-discrete-choice-models-from-market-research-data-a84e46c67d26
Rae Yule Kim is a professor of marketing at Montclair State University. His research provides insights for marketers looking to improve the digital presence of their business. [email protected]
Copyright held by author. Publication rights licensed to ACM.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2023 ACM, Inc.