5 Ways Brands Underutilize Data Analytics

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

Mauricio Vianna is the CEO of MJV Technology & Innovation, a global consulting firm advising the world’s largest companies on data strategy, business transformation, and design thinking. Their clients include Coca-Cola, BNP Paribas, Delta, and more.

Despite the widespread application of analytics dashboards and data-driven KPIs across the C-suite, most senior marketing teams are utilizing data primarily for backward-looking analysis to measure performance rather than building analytics dashboards that drive future initiatives and planning. For that reason, the majority of brands actually underutilize data that they already have at their fingertips. 

Without diminishing the importance of using data KPIs to evaluate past marketing performance, there’s often vast untapped value to be found in re-engineering some of your data points and what you can learn from them for better understanding your target audience and optimizing a campaign that speaks more effectively to their needs. By doing this, we’ve found marketing programs can not only become more effective in conversion of new customers, but also result in customers that have higher engagement with the brand and higher lifetime value. 

1. Knowing Your Target Audience

When we have broad audiences, it’s hard to deliver campaigns that are relevant to everyone. Therefore, one of the first steps to achieve better results in campaigns is to better know your target customer and segment the audience in a meaningful and actionable way, so you’re able to serve the direct needs of each small audience cluster. 

An effective way to perform audience segmentation and targeting is through a combination of design thinking methods and data science. The framework enables you to have a comprehensive understanding of different consumer profiles with different behaviors and needs, so as to convey the right message to the right audience and ensure that products and services are clearly communicated to meet their needs and help them achieve their jobs to be done.

Data science enables the analysis of large volumes of data, with the use of sophisticated statistical techniques, which allow for finding patterns among consumers. Thus, demographic characteristics, geographic information, product use, and behavioral characteristics, for example, can be used to analyze and segment consumers. Design thinking processes, on the other hand, allow us to analyze consumers in depth and, thus, identify the most relevant factors to segment them according to their needs as well as to create personas. When both are combined, it is possible to identify the patterns of similarity and dissimilarity, considering the most important factors, in addition to understanding the relevant characteristics that differentiate and describe them, which supports the creation of campaigns that resonate better with them. 

This design-driven data science framework is normally based on in-depth qualitative interviews with consumers to understand customer profiles and needs more deeply: on the analysis of large volumes of customer data for generating insights about customers behaviors, preferences and profiles; on advanced analytics techniques and machine learning (ML) to cluster customers and perform statistical analysis; and on the use of frameworks such as jobs to be done to capture consumers’ needs. This process is iterative and provides a means for testing hypotheses generated on the qualitative interview. Also, as some of the consumer insights generated from data analysis are based on correlations, which does not imply causation, data insights can also suggest some points to be explored more deeply on the qualitative interviews.

2. Optimizing acquisition cost by predicting lifetime customer value

Marketers are always under a strict budget. Therefore, it’s very important to optimize spending so as to derive maximum ROI from their allocated budgets in campaigns. Data analysis and machine learning can be great tools to improve customer acquisition processes and reduce its costs. Data can support the estimation of the customer acquisition cost (CAC) as well as the customer lifetime value (CLV), which tells the estimated profits the company can expect to have with each customer over his lifetime, starting with a new customer’s first purchase or contract and ending with the moment of churn.

By calculating the CLV, companies are able to evaluate how much to invest in a customer — based on the potential return — and evaluate the different strategies and levels of investment that are worth making for each customer profile, such as discounts, in order to acquire new customers with higher value. By segmenting customers according to their lifetime values, it’s possible to optimize acquisition costs by investing more in campaigns targeting the leads who will likely generate more revenue throughout the whole life cycle and perhaps carefully cut costs from the ones who will probably not generate financial value, when necessary. Naturally, you’re looking for an inverse relationship between your CAC and your CLV, with your CLV being the higher of the two numbers. The CLV could also be considered as the maximum possible spending to acquire that customer.

There are several different ways to calculate CLV, and it depends whether the business operates in a contractual (e.g., Netflix, credit cards, SaaS business), where the customer needs to cancel in order to churn or in a non-contractual setting (e.g., online retail, grocery stores) as well as if the transactions are discrete (e.g., monthly/yearly) or continuous (e.g., online retail where payments are less predictable). Many companies with recurring revenue end up using a simplified formula, which considers revenue per period, retention rate per period, and discount rate. The issue with this methodology is that it does not account for uncertainty, as it computes a single number expressing the expected CLV (or the average), and uses an aggregate retention rate among customers, instead of accounting for the different retention rates for different customers. 

A more complete CLV methodology uses probability models and requires advanced statistical knowledge in order to perform a more accurate estimation of the CLV of each customer, providing a more robust and dynamic metric. By segmenting customers with this metric, it’s possible to understand demographic and behavioral traits of the most valuable customers and even train a machine learning model to predict the CLV segment of new leads, which allows companies to optimize customer acquisition budget. Also, it’s possible to perform customer look-alike targeting to find new leads similar to the higher value ones.

See more: How FedEx, Pizza Hut, American Eagle Outfitters, NHL, and Members 1st Credit Union, Use Data Analytics: Business Case Studies

3. Building an effective propensity model

Even though marketers are always reinforcing the importance of sending the right messages to the right people at the right time, they still use one-size-fits-all approaches to engage leads. Data analytics can address this challenge driving greater personalization and business results by modeling consumer behavior through the use of propensity models. Proper use of these models helps to predict the likelihood that leads and consumers will perform certain actions, such as make a purchase or convert to the next step of the funnel. 

Many companies struggle with getting good outcomes with the use of propensity models. One of the reasons is that they use the propensity scores given by CRM tools or marketing automation platforms, which can be scalable but might not give accurate predictions, as they are not specifically designed to that company case and, therefore, do not take advantage of the richness of variables available. Another reason for the lack of success is that they might not act on the propensity scores efficiently, as they might target only the leads who are most likely to respond, for instance.

In order to be effective, a propensity model should be dynamic and be able to adapt to changes with time. This can be done through the automation of data pipelines and processes to retrain the model on a regular basis. Also, the model should be scalable so as not be abandoned after the first use in a single campaign. When building a propensity model, it’s also very important to be aligned with the business about the variables (e.g., demographics, product use, buying history, interactivity, behavior) that should work as good predictors and to use experimentation to validate the accuracy of propensity scores. Lastly, explainability is another factor that matters and by capturing the propensity drivers, which are the variables with the most predictive power to the model, business insights can be generated. 

With scalable and actionable propensity scores in hand, marketing teams can increase conversion rate, by defining the incremental impact of being targeted and target those people for whom the incremental response is better. Also, it’s possible to create strategies such as offering higher discounts for those who have a lower propensity score and need more incentives than for those who will buy anyway.

See more: Data Analytics in 2021: Key Trends

4. Monitoring and acting on consumer sentiments

In the digital world we live in, customers share a lot of information about their needs and the relationship with brands and products. Acquiring this information and analyzing it is very strategic for companies, as it provides a way to measure user satisfaction and loyalty. By leveraging approaches such as netnography and social listening, companies can understand customers’ emotions throughout their journey with products and services as well as their reactions to campaigns. This helps to understand points of improvement in order to achieve higher frequency of use and consumption, increase brand love, and build loyalty.

Data analytics through natural language processing (NLP) enables the analysis of large volumes of text data based on opinions and complaints left by consumers on social networks. Machine learning techniques can perform sentiment classification (e.g., negative, neutral or positive), using text as input data, and the results can be used to understand the opinion of customers: in relation to a particular product, service, or campaign; the feelings, pains, and desires of users when using a service in context from their point of view; and the relationship of different customer profiles with the brand. The information derived from this analysis can provide insights on how to improve the customer experience (CX) as a whole.

Sentiment analysis has become an essential tool for marketing campaigns, due to the scalability of analysis it enables in real-time, without losing accuracy. Therefore, based on this analysis, it’s possible to act on the consumers feedback, about what they want and feel, and personalize the marketing messages to attract the target audience.

5. Increasing lifetime value with personalized actions

As mentioned earlier, customer lifetime value is an important indicator that should guide customer acquisition and retention activities. In addition to tracking this metric to support the decision on which customers to invest more on campaigns, analytics can also help companies increase CLV through personalized offers and recommendations as well as predict the “next-best action” they can make to increase cross-selling and up-selling and capture more value from a client.

Research indicates that if a customer buys only once, there’s a 30% chance on average they will come back. However, it turns out that if the customer buys a second time, the chance of coming back increases significantly. Therefore, it’s important for marketers to act quickly to transform one-time buyers into two-time buyers. One of the ways to re-engage customers after a first buy is through a welcome campaign, which can also have personalized offers based on customer profile (or to which segment he belongs to). Those offers can also be triggered by a next-best action model, which uses predictive machine learning models to estimate the likelihood of the customer buying specific products and considers the one that has the highest chance of converting in that moment to create the offer. 

A next-best action model can prescribe content and messages relevant to the customer’s segment, stage of sales, and propensity and suggest the right sales opportunities, specific offerings, sales actions, and even actions to minimize churn. A good recommendation or next-best action model can increase conversion rate, CLV as well as consumer satisfaction, if the actions recommended meet the user needs. 

See more: Top Data Analytics Tools & Software 2021

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles