Spark illustration

Identifying gaps in your CX data

Emily CleaverContent Editor30 Sept 2022

The places current customer experience methodologies leave you guessing

Your CX analytics toolbox has some powerful options for building a picture of your customer experience. NPS, CSAT, CES, digital analytics, purchase behaviour metrics, footfall, churn rates; these tools generate bucketloads of data and deliver useful insight. But the picture they build of customer preference has gaping holes in it.

In this article, we look at the state of current CX analytics and identify where those insight gaps lie. And we look towards a new kind of metric that can fill the gaps in our understanding of CX performance.

CX impacts tend to be measured across a mix of three types of analytics:

  • Customer perception feedback
  • Channel engagement metrics
  • Customer lifecycle measurement

Customer perception feedback

The mainstay of customer analytics for decades, customer perception feedback relies on customers sharing what they felt about an interaction with a brand or retailer.

Net Promotor Score (NPS), Customer Satisfaction Score (CSAT), Customer Effort Score (CES) and product review analysis are perception feedback metrics.

These metrics require real-time, responsive, multichannel activation of large volumes of surveys completed ideally by identifiable consumers to ensure that samples are representative of the buyer market.

Customer perception metrics rely on feedback coming from genuine purchasers

Challenges of customer perception feedback

High data volume
For customer perception feedback to deliver meaningful insight, large volumes of data are needed. Survey engagement volumes need to be high to be representative of a potential market and often need to be from verified customers.

Customer surveys are often linked to staff bonuses, with financial rewards for high scores and negative consequences for low scores. This leads to gaming of scores by staff, for example asking customers to return high scores.

Negative bias
Engagement with customer surveys is typically dominated by the disgruntled and ignored by the indifferent or satisfied.

Lack of actionable detail
Data aggregated for a simple score doesn’t necessarily provide the granularity required to be confident of what changes to make to improve things. Returned feedback is subject to sector and category-specific nuances, making it unreliable as a basis for a predictive model in the field.

Input is questionable
Interpretation of the data from customer surveys doesn’t consider subjectivity. A customer may give a different response to a survey in different contexts and mindsets.

Customer surveys record a customer’s opinion at one interaction point, but don’t effectively predict a customer’s long-term loyalty.

The ‘one-number’ problem
The basic structure of most customer surveys is open to bias from the affect heuristic. Surveys often ask customers to pick one number to rate their experience. But a choosing a keystone number is a system-1 thinking trigger. System-1 thinking is fast and instinctive compared to more considered logical system-2 thinking. This means the conclusions you draw will be based a non-rational decision on the part of the consumer in that moment, in other words risky and unreliable data.

The connection between perception insight and commercial application remains tentative and costly to analyse or engineer in a robust way.

Customer perception insight tends to concentrate on data from within an organisation, ignoring competitive influences in market.

Channel engagement metrics

A diverse set customer engagement measures; digital metrics such as email engagement, page views and digital conversion funnels; contact centre KPIs such as first contact resolution ratios, average resolution times, complaint volumes; in-store engagement metrics such as footfall, sales volume and mystery shop audits.

Digital engagement metrics can be developed to a high degree of granularity with a well-integrated and comprehensive tech stack.

Reliance on internally available data means that bespoke predictive models can be made, which will more accurately reflect the specifics of an organization’s market, sales and services. The line between commercials and engagement can be more robustly modelled and tested.

Challenges of channel engagement metrics

Data overwhelm
Rapidly improving data-capture technology has led to an ocean of data being available to companies. But identifying the metrics that matter — which data is significant — can be a challenge.

Siloed data
Different technology is used for different metrics; Google Analytics for web views, EMS for email engagement and so on. But linking these channels of engagement is still superficial. A customer who appears not to be engaging with emails might be responding to social media advertising, but this link can be invisible to companies. Following a customer through their journey to purchase is challenging.

Invisible customers
With physical store captured metrics, the chief challenge is understanding what led to a buyer walking out without purchasing when they cannot be profiled or identified directly.

Channel engagement metrics suffer from lost customer gaps

Customer lifecycle measurement

Tracking the complete customer journey from initial awareness through to repeat custom includes purchase behaviour metrics, time series metrics and transactional KPIs such as RFM, retention and churn rates.

These metrics are predictive at cohort level and great for generating transaction-based target segments geared towards retention. This is particularly valuable in organisations where multiple variants exist at the SKU level.

Challenges of customer lifecycle measurement

Only predictive at macro level
Customer lifecycle is predictive to an extent at the cohort level, depending on volumes of purchasers relative to potential market size, but not at an individual level.

Information gaps
Purchase behaviour data leaves the root cause for purchase unknown.

Limited to known customers
CRM and Martech stacks generally gather data from customers who have raised a flag to make themselves known. But while this technology is well designed for this purpose, it raises challenges when you need to move to wider view. To achieve growth, you need to work beyond the bounds of the identifiable customer, appealing to people who are not yet customers. But you can only hazard a guess on how to do this based on the data you have on known customers.

Limited acquisition strategy
These metrics are not suitable for generating acquisition hypotheses, particularly in unpredictable markets or where a lot of competitive force exists.

Effect of competitive entrants
Customer lifecycle measurement can’t predict or compensate for the effect of new competitive entrants to market.

Doesn’t account for changes in context
The underlying hypothesis for growth with these metrics relies on past performance being indicative of future behaviour. It requires constancy in purchase context and customer needstate over time, which may not exist.

Additional analytics required
For creative activations, transaction data needs to be heavily supplemented with cohort research, emotional engagement and attitudinal research.

A new metric to plug the gaps

Why do customers choose one retailer over another? What really influences their choices at the moment of decision? These are the questions where current CX metrics leave you guessing.

Relative Attractiveness (RA), Uncrowd’s new analytics metric, measures how retailers or brands perform at the point of customer decision, observing and analysing the customer experience environment itself. RA does not rely on canvassing customer opinion; instead, it uses objective observation of the CX environment, bypassing the bias inherent in customer surveys.

As a result, RA is unlike all established tools, plugging large gaps across all measures. RA delivers an overall market view, describing how different competitors show up at decision points.

RA can help you understand why your customers choose you, or a competitor.