Pillar 3

Predict: Becoming predictive and anticipating needs

Pillar 3 - Predict

Working with customer feedback is complex. While it always makes sense to ask customers what they want and what they think, acting on it directly is not always the right answer.

Today, most data analysts will look at the correlation between low NPS or CSAT scores and what individual customers are saying. They’ll draw conclusions based on a linear regression; i.e., what are the biggest drivers of low NPS? They’ll come up with statements such as, “20 percent of detractors rate speed of service as important to them.” Will addressing speed of service convert these detractors into promoters? Maybe.

But especially in the experiential context, there is interplay between these factors. What customers say they want isn’t always going to drive the highest return on investment. You might find that spending money to improve the speed of service doesn’t actually improve revenue or customer loyalty.

Take, for example, the experience of eating a meal in a nice restaurant. While almost all customers will rate the importance of visual appeal highly, it’s not what brings them back to the restaurant. If you get a meal that looks amazing, but doesn’t taste great, you are not going to rate the meal highly. Even though taste is influenced by visual appeal, ultimately, the taste of the food is one of the top defining factors in cuisine.

In CEM, interconnected data points can make it difficult to pinpoint exactly what’s driving customer behavior. A customer journey isn’t a series of separate, discrete data points. The experience a customer had online impacts the experience they have in the store. The speed of service impacts their opinion of the value of the product. There’s very few standalone moments when measuring customer experience.

This interplay is where it gets harder to separate which factors are truly relevant to the customer experience and what would actually move the needle to have an impact on your business.

And here’s where predictive analytics shines. At Cisco, we want to move beyond correlation analysis and start accurately predicting the drivers of long-term success.

Without the ability to accurately predict, CEM programs become a practice of randomly guessing at what will work. It’s an expensive hobby—and it won’t have the impact you want.

A predictive analytics platform identifies the drivers of customer loyalty, and pinpoints the touchpoints where taking action would have a measurable impact on achieving business goals.

In this section, we look at three advances in the field of predictive analytics that can give your data more power and increase the ROI of your CEM programs.

Tracking CEM success against financial metrics

Traditional closed-loop CEM programs often focus on increasing common survey metrics like NPS and CSAT through customer experience improvements. The typical cycle is: survey customers; identify drivers of poor customer experiences; implement changes; survey customers again; then wonder why you haven’t increased revenue.

This is because CEM needs to look at more than just what the customer is telling you.

Does decreasing the speed of wait time from five minutes to
two minutes make customers more satisfied? Potentially.
Does it increase their average transaction spend? Maybe not. Before making enormous investments trying to improve elements of the customer experience, it’s critical for businesses to model the impact on the business' finances.

Contextual predictive analysis pulls in financial metrics like
ARPU and LTV to help model how initiatives will affect your revenue. Customer experience management helps
businesses achieve their goals, so it’s important to tie
those goals into the analysis.

Predictions driven by machine learning

Given enough time, data analysts can manually perform calculations that are similar to a predictive algorithm.
But humans are fundamentally flawed when it comes to interpreting data. We make hypotheses and implement our own ideas when we look at a data set. Our brains seek out patterns that might not exist. It turns out, you can make data say almost anything you want it to, given enough conditions.
This bias expresses itself in leading questions and tenuous connections. When making strategic decisions, companies
need to base their investments on what the data actually says, not on instincts backed up by data.

Predictive analytics driven by machine learning are more accurate and (when supported by data analysts) illuminate areas of CEM that you weren’t looking at in earlier analysis. You don’t know what you don’t know, but machine learning can help you understand the trends you didn’t think to look for.

He uses statistics as a drunken man uses lamp posts—for
support rather than illumination.

Andrew Lang, Scottish literary critic

Insights at the speed of business

All this analysis is great, but it means nothing if it’s not delivered quickly enough for companies to take action on it. Collecting data, sending it out to a consultancy, and waiting for results can often take months.

But businesses don’t have months anymore. Customers won’t wait around for their concerns to be addressed because the competition is always only a Google ad away. Frustrated customers will take to social media or review sites to vent. Recovering from a few one-star reviews is possible, but if you wait a few months for data collection before acting, you’ll be swimming upstream to regain your market position. Customers have a world of choice, so you need to move quickly to address their concerns.

Instant analysis in the hands of your practitioners means getting on top of customer concerns before they become a trending topic on Twitter.

A CEM platform that combines a 360-degree view of the customer with machine learning allows businesses to make decisions quickly.

Summary of
Pillar 3 - Predict

Correlation and regression analysis can’t accurately
predict the drivers of customer experience. Teams need
to recognize the interplay between journey elements and
analyze accordingly.

Companies need to understand predictively how changes
to the customer experience will impact financial metrics.
Only contextual predictive analysis can deliver this
depth of insight.

Machine learning can help illuminate trends that human data analysts may not have thought to look for. Having these
insights empowers all stakeholders to make the right decisions around customer experience.

Businesses move fast, and only machine learning analysis
can deliver insights as quickly as leaders need them.

Bringing the three pillars together on one platform

At Cisco, we understand the importance of putting the power of a CEM platform into the hands of the people that will use it.

That means not waiting weeks for survey administrators to adjust and deploy new questions. It means integrations that don’t take a team of developers to set up. Finally, it means having access to instant insights that are easily understandable by the average user—no statistics degree necessary.

If what you’ve read resonates with you, and you’re interested to see these pillars in action, please request a demo.