AI explained: how to cut through the noise and use it successfully in your Customer Management

So much has been written about artificial intelligence (AI) and what it can do for businesses, their customers and employees (as well as other parts of society). But unfortunately, too much of the writing about AI veers between the hugely over-hyped, the hugely negative and the hugely confusing (or confused). 

It’s understandable there’s such a mix of views, but also regrettable. Over-promising or underplaying what AI can do for businesses will hamper them from using it most effectively to serve customers better and create value. To help with this, we wanted to bring some much-needed clarity to how AI can be used in Customer Management to benefit both businesses and their customers.



The first thing to understand is that AI is here; it’s not just some prediction of future magic. Almost a quarter of CIOs say they have already deployed it, according to Gartner, and some surveys put the figure higher. 

We’re also going to see much more of it. Nine out of ten tech executives told Bain they view AI and machine learning as priorities they should be incorporating into their product lines and businesses, and Covid has persuaded many companies to accelerate this adoption.



Clearly, the uses of AI extend far, far beyond Customer Management, but that’s what we’ll focus on here. Even in this specific space, the possible applications are numerous, and we’ll give just a few examples below. All of them create clear opportunities to serve customers better, from resolving their queries or issues more quickly to personalizing sales offers.

  • Cost reduction: AI driven chatbots and the use of natural language processing (NLP) can be used in automated customer care applications to control costs while providing high first-time resolution. According to the latest McKinsey Global Survey on AI, contact center automation is one of the most common use cases across all business activities (along with talent optimization and warehouse automation) when it comes to reducing operational costs through the deployment of AI.
  • More effective call routing: Using AI tools and different data sources (from demographic data to customer-specific data), Customer Management providers can predict which agents, types of agents or emotional tone are most likely to, eg, offer first-time and fast resolution, convert a sale, or generate high customer satisfaction or low customer effort scores. The potential uses cover a range of business processes including customer care, sales and technical assistance.
  • Personalized offers and service: AI and machine learning – including the use of speech analytics – can produce a range of predictive insights that empower businesses and/or individual agents to make personalized offers. Again, the applications span a range of different processes from customer care to sales to credit collection.
  • Better leads scoring and leads generation for sales or cross-sales: Using hundreds of variables from both general data (eg household data or demographic data) and client-specific data, brands can identify trends and use them to sell proactively or finetune their sales approach to different customers – whether that’s what time to call or what offer to make. Adding AI tools that analyze speech can add to brands’ options here, offering the capability to predict in real time which customer care callers might be open to a cross-selling or upselling approach.
  • Reducing customer churn: Through AI-driven predictive analytics, brands can build powerful insights into churn propensity. Variables used in these analytics could include demographic data (eg age), client histories and agent analysis, enabling business to better understand customer attributes, behaviors or warning signs, among other things, relating to high churn rates.
  • Enhancing agent performance and reducing staff churn: Similar approaches to all those above can also be used to enrich staff training and coaching, providing employees with better insights into customers and trends, or skills to deal with different scenarios, or the tools to personalize customer offers in real time. This in turn can feed into work satisfaction, performance and engagement, helping to reduce churn rates. Managers can also use agent performance insights from AI to evolve retention and rewards approaches.



When AI applications like those above have been clearly mapped and proven, it’s perhaps surprising to see some of the skepticism and ambiguity around AI that’s so common in research and media comment. For example, the same Bain research that said 90% of tech executives viewed AI as a priority also found that nearly nine in ten respondents were not satisfied with their organization’s current AI approach.

There could be many reasons for people’s caution or disenchantment with current AI approaches. One is that they expect too much from AI and data: they expect the algorithms to reveal hidden truths and solutions when in fact they are more about modelling. To quote (and translate) a recent article in Mind Fintech “A model created on the basis of machine learning can be efficient in prediction or categorization  … but will not explain the phenomenon it is supposed to predict or realize.”

Businesses that understand this distinction are better placed to harness the power of AI – seeing it as an enabler that supports strategy rather than being a substitute for strategy. Let’s take the example of reducing customer churn above: AI can help you identify and analyze churn propensity but won’t necessarily provide an innovative strategy to address it successfully. That’s where humans still come in.



Another issue that businesses still need to understand better is that the quality of results from AI depends on the quality of data put in. “Garbage in, garbage out,” as they say. Issues could relate to data hygiene (i.e. whether there are errors in the data), data ethics (e.g. whether there is bias in the algorithms or data sets), or timing (eg pre-Covid or mid-Covid data may no longer be appropriate in the Next Normal). 

There could also be issues relating to organizational silos. We talked earlier about using client-specific and customer-specific data for predictive analytics. Ideally, this data will give a holistic view of customers’ interactions with the brand – not just their activity with Customer Management teams, but with other parts of the business: sales, marketing, finance, front-line stores or offices. Businesses that lag on back-end and front-end integration and don’t view customer data holistically are missing an opportunity to do AI better.



Two other common reasons for disappointment with AI results are around process and people. 

On process, businesses may be adopting on too small a scale (eg to try to automate an existing single process) or too large a scale (eg to reinvent their whole business). “Neither of these routes … can deliver the level or speed of change companies require to grow and thrive in the digital age”, is the McKinsey verdict.

On people, businesses may lack the in-house capabilities to implement AI successfully. To quote Bain again: “AI is complex to mobilize, and it can be challenging to get it to actually do what companies need it to do. Many executives don’t believe they have the talent on hand to make it happen”.

The solution to these issues is threefold:

  • ask the experts, those Customer Management BPO providers that are already using AI-driven tools and processes
  • be clear and targeted in your goals, such as whether you are looking to optimize costs, reduce customer churn, generate better leads or another specific goal 
  • be prepared to back your AI processes with the necessary organizational change, such as better integration of back-end and front-end or data from across your operations.


The Comdata team is here to help you with all three of these points.


Join us and let's shape tomorrow together

we make things happen

Please, don't be shy. We would be pleased to help you!

What do you want to talk about ?
Scroll to Top