Want to prevent customer churn?
As the business world has become increasingly volatile due to the impacts of COVID-19, there’s no better time to focus on taking the best care possible of your existing customers.
With mobile usage at an all-time high, providing a seamless experience is the best way to ensure that you’re doing everything you can to deliver a consistently high-quality experience.
Many still think this means manually tracking and analyzing key metrics from both a frontend and backend standpoint. But who has the time? Doing it yourself is a sure-fire way to have incidents slip through the crack. And if you’re relying on business intelligence methods, you likely understand the dashboards and static thresholds don’t provide the granular visibility and context needed to support high-fidelity alerting.
Many data-driven companies have instead embraced machine learning to automate the process, reduce human effort and false positives, and monitor at scale.
But first, it’s important to grasp what makes CX visibility too complex for traditional methods.
Understanding the importance of delivering a seamless customer experience is very intuitive, as we’re all customers, and we’ve experienced the frustration associated with things like long page loading times or an app crashing while you’re using it.
Your customers will inevitably interact with your product and company across multiple touchpoints, customer journeys and devices. The result of this multi-layered experience is that your customers are constantly generating huge amounts of data that’s waiting to be analyzed.
Customer experience monitoring is the process of making sense of all this data and adapting to changes in real time in order to deliver the most seamless experience possible.
From customer logins and conversion rates, to churn rates — each of these business KPIs has their own uniquely complex data patterns. The following factors contribute to the complexity of the customer experience:
- A multi-layered experience: Apps are built from multiple layers including the infrastructure, application, interaction and audience. As you can see in the image below, each one of these layers is susceptible to their own issues that affect the overall customer experience, and as such must be monitored at all times.
- Data is volatile and context-sensitive: Another feature that is unique to the customer experience is that, since we’re dealing with humans, the data is extremely topical and irregular. As a result of this dynamic data, each one of the customer experience metrics must be evaluated in relation to a set of changing conditions, instead of in absolute terms.
- CX has an unknown business impact: In contrast to machine data where we know the relationship between machines, the business impact from changes in the data cannot be predicted with certainty due to its dynamic nature.
In researching monitoring solutions for customer experience, you’ll find that many companies have tried to feed customer experience metrics into traditional IT or APM monitoring solutions. As these companies have realized, however, monitoring these metrics in practice requires a solution that can account for the unique nature of customer data discussed above.
To solve the challenges related to the customer experience, companies have now realized that machine learning is the only means of achieving the accuracy and granularity needed to monitor effectively.
Leveraging AI for customer experience monitoring has proved a viable solution to deal with its unique data. Not only can AI constantly monitor each metric on its own, unlike traditional monitoring systems, it can also find correlations between all the metrics.
In particular, an AI-based customer experience monitoring solution employs the following characteristics:
- Unsupervised learning is used to identify anomalies: Unsupervised learning refers to a branch of machine learning algorithms that can take in unlabeled data and derive patterns and structure from it. In practice, this means that the AI can learn the normal behavior of each CX metric on its own in order to identify anomalous behavior.
- Real-time monitoring of all data streams: With the scale and granularity of customer experience data, identifying potentially revenue-impactful events requires you to monitor 100% of the data at all times. This means monitoring live data streams for every product, country, device and so on. Only an AI-based solution can find the multi-layered correlations between this data to detect significant anomalies when you need it most.
- Seasonality detection: As mentioned, since we’re dealing with human generated data, this means we need to account for seasonality to construct a true understanding of each metric. By detecting seasonality in data, you’re able to free your technical team from false positives, false negatives, and alert storms.
- Root cause analysis: As incidents inevitably arise in the customer experience, AI can then perform a root cause analysis and identify all the events and contributing factors that led to the issue so that you have the fastest possible time to resolution.
- Anomaly scoring: As machine learning algorithms are employed for complex analysis of many different CX metrics, the data is filtered into a single, scored metric that anyone can understand. The significance of each anomaly is then evaluated so that only critical incidents are detected and flagged.
Now that we know what customer experience monitoring is and how AI is applied to the data, let’s review several real-world use cases.
First-Time User Experience
As you start to scale and acquire new users, one of the first questions you should ask is whether or not these users are having a good experience with the app. As you can see below, in this example, a gaming company was alerted whenever there were significant drops in first time user experience flow completions:
Detecting App Crashes for New Version Releases
If you’re rapidly developing and pushing new versions of your app, it’s almost inevitable that there will be issues. One such issue is how each version will perform on various platforms. For example, in the image below you can see a company was alerted when there was a significant drop in daily active users, which was due to an increase in app crashes after releasing a new version on iOS:
Monitoring Traffic & Page Load Times
As mentioned at the beginning of this article, one of the most frustrating customer experiences you can have is a long page loading time. In the example below, you can see an eCommerce company experienced a spike in page load speed, and as a result shoppers chose not to wait and traffic dropped significantly:
The one common denominator between all these use-cases is that the faster you can fix them, the better experience customers will have, and as a result you’ll often generate significantly more revenue.
As we’ve discussed, during times of uncertainty, one of the best investments you can make is taking care of your existing customers.
Customer experience monitoring allows you to stay up-to-date on exactly how your customers are interacting with your product. Since we’re dealing with human behavior, however, the data generated from customer experience monitoring is incredibly volatile, irregular, and seasonal.
This dynamic nature means that traditional monitoring methods such as IT or APM monitoring simply can’t keep up the volume of data or number of metrics you have at your disposal.
Instead, AI and machine learning solutions offer companies the granularity and scalability of being able to monitor thousands of customer experience metrics simultaneously and find correlations that would have otherwise gone unnoticed.
There are many use cases for AI-based customer experience monitoring, although they all have the same underlying goal: finding and detecting issues as fast as possible so that your customers have the best experience possible.