DATA SCIENCE ● OCT 10, 2018
Supercharge your business with Data Science
AVP, Data Science
In a digital universe that is now constantly trying to sell some product or service to users, the ability to convert leads into customers is critically important. To do so requires businesses to understand how and why their customers behave the way that they do. Behavioral economics offers a ready rubric through which to measure their attempts at capturing user attention and the customer journey - and, as per evidence is an effective way to gain traction. Gallup’s analysis has found that companies using data-driven insights into customer behavior outperform their peers by
85% in sales growth and more than 25% in gross margins.
Types of data
But what kind of data should the company be looking at? Digital companies today can potentially collect enormous amounts of data about their customers -
Transactional data - When did the customer make a purchase? What did they buy? At what price?
Customer profile data - How old is the customer? Are they male or female? Where are they from?
Behavior data - What promotions do they respond to? How do they navigate the website?
Social media data - What posts do they like, share, and comment on?
Feedback or direct communication- How do they respond to surveys? Emails?
Companies can use this data to arrive at various conclusions, based on how they see and use it. For example, customers from a particular area might show a preference for a product category in a particular season. Companies can offer their other customers from the same region in that season targeted promotions and thus trigger higher conversions.
This is just scratching the surface of this data’s potential. This data can help companies make a variety of decisions based on behavioral insights which can help them get the best traction on critical KPIs - new customers, customer satisfaction, and customer retention, and profitability.
Getting new customers
One of the biggest challenges for high-growth companies is attracting new customers. These companies don’t have the large marketing budgets that large enterprises do. To be competitive enough to attract the attention of people unlikely to have heard of them before, they need to make sure that every cent they spend on customer acquisition is well spent.
This is where data and behavioral insights can help. If companies want to know, for instance, what would be the most efficient way to gain new customers in their target demographic, data can provide a clear and accurate answer.
Companies looking to increase the number of new customers they attract should studying their discovery funnel, i.e., the journey taken by people becoming their customers for the first time. Channel attribution models on top of your website’s Google Analytics data could help identify typical conversion paths and high RoI channels.
A Machine Learning model can discern patterns from historical campaign attributes and their response data to generate recommendations for existing and new campaigns. For example, you might understand what tags / content themes in campaigns could drive high virality. Companies can use such systems to make the right decisions to get the most out of their digital marketing budget and raise their rate of attracting new customers in a controlled and predictable data-driven fashion.
Catering to existing customers
In addition to growing their customer base, retaining and engaging current clientele is often a core differentiator for high-growth companies. To best utilize data to do this, we talk about a few areas they could pursue, ranging from deploying automated metrics tracking, feedback analysis using data for optimizing customer retention.
Tracking business KPIs using automation
Keeping track of critical business metrics changes as companies scale up and grow. At the early stages, this can be done through simple reports and can even be managed on spreadsheets. At some stage, the different variables which need to be looked become so numerous that finding an easy way to track all of them becomes difficult.
One of our clients, a beauty and wellness business with 23 treatment clinics across 3 different markets, was having trouble building performance reports from their laser facial skin clinic business. Collecting the data from their multiple locations which have catered to 150K+ customers was absorbing too much staff time and effort. When Atidiv started looking into this project, we realized that this was a repeatable process - and one that could be automated. We iteratively defined KPIs, built out a production system to automate reports by collecting data from the client’s ERP system, emails, and internal sheets. The automated reporting system gave the client management accurate daily, weekly, and monthly performance reporting on KPIs that mattered to their business. This initiative helped free up ~500+ hours of staff effort every year, translating to ~$300K+1 worth of time savings annually.
Analyzing customer feedback
Managing to engage with customers when they communicate with a business can also become a difficult task at scale. Feedback data - social media posts, emails, survey responses, reviews, etc. - is critical to a growing business, as it can let companies know what their customers want and where they should be focusing their attention. At scale, the sheer volume of such communications makes it impossible for anyone to manually go through everything to find which messages are important, and which ones are not.
Natural Language Processing (NLP) is an area of technology which can help. Companies can use these techniques to filter through amorphous data from across multiple sources, and group them under different themes. This way, an automated system can let companies know that they received lots of appreciative messages about one aspect of their service. It can also warn them that customers are sending negative messages about their customer support, for example. At Atidiv, we build these NLP feedback-analysis tools that provide you with interactive, daily updating dashboards. A company’s operations team needs to only keep track of such an integrated interface to understand feedback on how their products and services are moving.
Improving customer retention
Customer retention speaks to the long-term viability of the business, and is critical to a company’s ability to show its potential. Transactions and engagement behavior data from your existing customers helps determine whom to target and how to target them for better conversions.
But how can a company classify customers to slot them into convenient categories for personalized engagement? Customer scoring is a commonly used method. It uses recency, frequency, monetary value of purchases, and features like user longevity / loyalty to categorize customers.
For instance, if a customer just bought a company’s product, buys their products frequently, and buys premium products, they would have a high score in each of the three categories. Older customers who haven’t bought anything recently but buy frequently are probably better targets for discounts to incentivize them to fall back into the pattern of buying from a particular brand.
This is known as an RFML analysis. Customized versions of this scoring mechanism are extremely useful to help companies target their existing customer base for remarketing campaigns.
Determining how and why customers are choosing to disengage from a company’s products or services is difficult because of the sheer number of variables involved. A business could lose a customer to new competition, boredom, quality of service, change of preferences, or even a change in location. Knowing which customers are at risk of churn in the near-term could give you a great starting point to target specific retention schemes.
This becomes even more critical for subscription business. One of our clients, a $500M+ meal kit & food delivery business with 500K+ active customers wanted to know which users were likely to ‘churn’, i.e. cancel their subscription at the end of any week. We used predictive modeling to project the likelihood of churn for every user utilizing data from their prior purchases, behavior, ratings, reviews and customer profiles. As a result, we highlighted potential savings of $2-3M every year2 by helping the company know which users to target with promotions - and hence potentially improve customer retention.
1. Using average employee salary as a base assumption
2. Assuming a 2-5% conversion rate from retention schemes