AI Use Cases in Action

AI Use Cases in Action

5th September 2018

With Artificial Intelligence (AI) technologies accelerating at rapid pace so too is the potential to make customer experience programs more effective and efficient. Customer experience (CX) professionals who neglect the potential of AI risk getting left behind as the rest of the market leverage the true potential of rapidly evolving vendor offerings. Leading independent research analyst Forrester recently released their latest report, “The AI Revolution in CX Measurement”, in which they detail core uses cases to better track and action CX, below we explore in detail some of these core competencies.

AI driving action

The report centres around the benefits of predicting CX quality instead of surveying and reacting to situations. Key to advancement is the ability to mine insights across multiple data sources and action these in real-time. These core competencies are geared at helping stakeholders engage with CX metrics in a proactive way to facilitate change.

While the principle of utilising multiple data sources is far from new; the practicality of tying these channels together for greatest gain can be tricky. AI holds the potential to address these issues, with a few key examples:

1)Mine text data more effectively, at scale
Text feedback is a useful yet largely untapped resource for most CX professionals. Meaningful insight is there for extraction however mining this is often a resource intensive and difficult process. Typically, CX professionals will adopt a rule-based approach to extract useful data from text. Whilst beneficial, this method remains labour intensive and also loses effectiveness over time as fundamentals, terminology and patterns change. Literal meanings and interpretations of context are also substantial factors which can impact discoveries.

By utilising machine learning, text analytics vendors are now able to identify topics or sentiment that rules-based engines miss. Deep learning not only enables text to be analysed for patterns, trends and insight, but crucially, it gets it right in relation to sentiment and context. Google saw a 60% reduction in translation errors when it switched from traditional phrase-based machine translation to an AI centred algorithm.

2) Find better insights in speech data
Hidden signals and nuances are common in any form of communication. This is none truer than with the spoken word. The phrase “it’s not what you say but how you say it” carries weight in the world of Customer Experience.

Until recently, the process of analysing speech has proved difficult to extract value in a customer experience context. Events such as Customer Service calls have the potential to unearth valuable insights into customers struggles, wants and needs. While transcripts are available, these critically miss context and often lack quality. Similarly call scripts and logs are time consuming to analyse and can miss audio queues such as a raised voice or different intonation.

To combat this, emerging tools such as Cogito use AI to analyze behavioural signals and dynamics in live conversations. Everything from pitch and tone to pace and mimicry is detected and analysed. AI models such as this make it far easier to identify and implement new language patterns, helping to identify intents, topics and relationships in conversation.

3) Surface anomalies and emerging issues
Most organisations monitor and detect known journeys with alerts and events around pre-selected metrics. This works for surfacing recognised problems as they occur but so often the smaller segments that go unnoticed can have the largest impact to conversions and the bottom line.

Using AI to parse through data helps flag CX patterns professionals aren’t actively looking for. For example, within UserReplay, behavioural data is collected, and machine learning algorithms applied to identify customer struggles, no matter how subtle they may be.

For an online retail client, UserReplay found that some customers got stuck while trying to pay when one item in the basket had just sold out. This is because the error message for customers did not indicate which item was no longer available. Customers then had to remove and re-add items, which frustrated them and led to cart abandonment. UserReplay estimated the annualised opportunity of adding a clearer error message at £986,000.

UserReplay Dashboard

Start preparing now for AI success with CX measurement

Whilst AI creates new challenges for CX professionals there is also a wealth of opportunity waiting to be unlocked. Vendors continue to be progressive and innovate new solutions requiring professionals to do the same.

In order to succeed Forrester outlines that CX pros should partner with analytics and privacy peers to ensure quality and unbiased insights. They should also piggyback on existing AI-enabled tools in their firms. Rooms To Go do this to tremendous effect aligning tools and teams to deliver in-store service levels of service online, driving up conversions. Finally, as with any improvement programme, CX measurements and KPIs require updating to measure the true impact of AI on the business. Relying on AI alone is not enough, CX pros need to exercise vigilance to ensure that algorithms do not engender bad Customer Experience.

Image Source –

Related Stories

Digital Disruption
22nd September 2016

The Second Wave of Digital Disruption

24th January 2018

The key to customer conversion

8th September 2016

Machine Learning Basics