3 Ways CloudCherry Improves Customer Experiences Through Predictive Analytics
Predictive analytics offers a unique benefit in that they allow a company to preemptively know what a customer is going to want, feel, or do within a product or application. In order to actually improve the customer experience, CX teams must act on the information they receive from predictive insights. A closed-loop customer experience management program involves five steps:
* Measure – create and execute surveys to receive customer feedback.
* Align – deliver insights across the organization, democratizing the data.
* Respond – close the loop on customer feedback and improve customer experience.
* Improve – implement tactical wins across the journey.
* Evolve – revisit the customer journey map frequently to ensure it accurately encompasses the customer’s perspective.
Predictive analytics help align and improve the customer experience. Using the data from the Measure stage, organizations can align around potential action items that are surfaced from the analytics engine. Putting these action items in play closes the loop on customer feedback and drives improvements across the journey.
There are three different types of detection that CloudCherry accomplishes: sentiment analysis, impact analysis, and path analysis. Each of these can be used to benefit your customer experience in different ways, whether it be for providing proactive experiences, feature request quantification, or just benefiting the overall CX of your product. Here’s a breakdown on what each of those is, and how they can be used to your customers’ benefit.
Detection of Sentiment
Understanding how your customers feel can be a bit of a nebulous exercise. We can ask them about their level of satisfaction and track their actions, but bucketing customers into negative, positive and neutral groups tells you very little about how they actually feel. How positive do customers feel about their website experience? Why is it positive, and what drivers are impacting these feelings? While sentiment analysis isn‘t truly a predictive tool, it does use machine learning to provide useful context around why customers do the things they do.
Rather than just asking customers how they feel, sentiment can also be derived from the language customers use in customer service interactions, survey responses, and reviews. Using machine learning and deep learning, organizations can analyze large amounts of data in order to understand not just how customers report feeling, but their overall temperament in every company interaction. CloudCherry also uses the “bag-of-words” model to classify text blocks based on the frequency of words in the document. These two approaches provide a number of benefits above and beyond simple surveys.
First, the amount of data is multiplied. Companies that are able to listen to their customers across a multitude of channels are able to identify trends more broadly. Unstructured data analysis can help uncover trends that you didn’t even know to ask about—whereas surveys require the use of specific questions to pinpoint feelings and feedback. It also allows for a deeper level of analysis. While understanding that customers have negative feelings about going to your retail store is important, deep learning can offer a closer look into the data and identify problems associated with checkout speed, ambiance, and staff friendliness – without even needing to survey customers. Sentiment analysis derives quantitative insights from unstructured, feelings-based data.
Secondly, using advanced machine learning is massively faster than previous methods of sentiment analysis. Insights surface in real-time, rather than waiting for data analysts to comb through tagged data to determine what drives sentiment. Anyone customer-facing is granted immediate context into every customer situation and sentiment as they are providing service. With more understanding of customer sentiment and the journey that the customer has taken with your product or service, you allow them to exert the least effort possible while still receiving an excellent experience.
Detection of Impact
What makes your customers come back again and again? Is it the selection? The quality of your product, or service? What do they tell their friends about when they recommend you?
These critical aspects of your customer experience are called “key drivers”. They are what drives your customers to spend more and recommend you to their friends and family. Uncovering your key drivers is critical to capitalizing on opportunities both in marketing and operations. Additionally, knowing what key drivers your organization aren’t consistently delivering on can inform your customer experience management program.
In order to understand why customers are or are not loyal, we need to dig into the data with Impact Analysis. This statistical model looks at the impact of drivers on customer’s loyalty to identify what actually moves the needle. Impact Analysis uses data from across the customer journey, including customer surveys designed with specific questions to get to the bottom of customers’ opinions on a variety of rating questions.
For example, along with asking for their likeliness to recommend, each NPS survey also offers one or two driver questions for more depth and insight into the customer’s experience. Customers might see a question on the cleanliness of the branch, or the friendliness of the customer service rep. Because of CloudCherry’s single question approach to data collection, all NPS results are stored in the same dataset, regardless of which survey they are collected from. Results are tagged with demographic and driver responses so that customer experience teams can filter and digest NPS data on multiple levels. Each survey response is connected to the customer’s previous responses so you can see how their experience changes over time. If Bob, a 55-year-old male, answers three NPS surveys over the course of a year, along with three different driver questions about store cleanliness, product quality, and employee friendliness – CloudCherry collects all of this information together for better data analysis.
All of this happens in real time without the need for complicated survey administrators and bureaucracy. This means CX teams can act quickly on the trends and insights uncovered by Impact Analysis – and never miss an opportunity to capitalize on a key driver. These types of data-driven insights can be earth-shattering for product teams and customer experience teams alike: they help inform both decisions and recommendations that your customers will love and appreciate.
Detection of Path
Customer experience is messy to measure because no two customers’ journeys are the same. Each interaction contains a myriad of factors that positively or negatively shape the customer’s experience with your brand. What are these factors? Are some factors more important than others? Accurately representing the impact of these factors is much more difficult than a simple regression analysis.
This is all complicated by the fact that customers don’t always know what they want themselves, so survey results are often misleading. We can only be sure that customers will do two things: Customers will tell you what they think, to the best of their knowledge, and customers will act. The difference between what they say and what they do is due to the messy tangle of experiential factors that drives customer actions.
The difference between what customers say, and what customers do is due to the messy tangle of experiential factors that drives customer actions.
In order to detangle these factors and discover what is truly driving customers to stay loyal and recommend your company to others, we use Path Analysis. Path Analysis is an extension of multiple regression. This model uses machine learning to understand the impact drivers of CX by creating separate paths for individual CX drivers (product quality, visual appeal and more) and connecting them to KPIs like NPS and Average Revenue per Customer. Path analysis involves working with multiple variables and solving for multiple predictions – not something that is easily done manually.
Why is this important? As most CX teams know, if you only focused on what customers were saying is important to them, you’ll never move the needle on the metrics that matter. It’s like what Henry Ford said about introducing the Model T: “If you asked customers what they wanted, they would have said a faster horse.” Pay attention instead to what your customers don’t know that they want or need, and you’ll be better off.
Path analysis cuts through what customers are saying they need to uncover what drivers actually make them come back and spend more money. Knowing the degree to which different drivers impact CX makes it easier to predict the financial returns of investments in CX.
Operationalizing improvements has always been a struggle for CX leaders requiring cross-departmental cooperation and a clearly established end goal to bring everyone on board. Predictive analytics helps support operational leaders by reducing the uncertainty of new initiatives. Continually updating frontline employees on the results of implementation can create an upward spiral of improvement, and allows for even more insight to be gleaned and pumped to the rest of your company. Continue to use predictive analytics to better your customers’ experience so that you can level-up the insights that you gain, and keep getting better.