Why do CX initiatives fail, and how can AI help? (customer experience; artificial intelligence)

The past 10 years of customer experience management

Customer centricity is no longer simply a positive attitude; it is an operating model. Organisations treat customer experience (CX) as a core value, learn from their customers on an ongoing basis, and incorporate that learning into their decision-making processes. Over the past decade, this approach has delivered more stable growth in many places, as retention has finally come into focus alongside customer acquisition. Research, feedback and testing have replaced ‘we think this is right’. Omnichannel thinking has gained ground, reducing friction between channels and creating a more consistent experience. Programmes designed to improve customer experience have also raised baseline quality expectations, so in many industries today, a good experience is not an extra but the minimum requirement. It has had a cultural impact, too: empathy and the processing of customer feedback have become part of organisations’ everyday practice. (customer experience)

At the same time, CX has often stalled not because there was no intent, but because organisations did not make the shift at a systemic level. The most common issue was a lack of ownership: if no one is accountable for the experience at journey level, it becomes everyone’s responsibility and no one’s. In many places, measurement and reporting were implemented, but the insight–backlog–development process was not established. The operating rhythm was typically reactive: we learned in monthly or quarterly cycles, while experience problems emerged more quickly than that. Finally, short-term business pressure often overrode experience improvements that would have paid off over the longer term. The lesson of the decade is therefore that CX only creates lasting value when it is not a series of projects, but an integrated system that learns quickly and is founded on sound decisions. Seen this way, the CX story of the past decade is both a success story and an unfinished sentence: we have learned to do many things well, but repeatedly got stuck at critical points in the system. (customer experience)

With the arrival of artificial intelligence, we now have a technological toolkit that can make a meaningful difference precisely in these areas.

AI opportunities in customer experience (artificial intelligence)

1) Efficiency

AI can make an impact on CX where these systems have repeatedly fallen short: speed, uncovering connections and efficiently supporting decision-making. Over the past decade, many organisations have learned to measure customer experience, but AI now gives us the opportunity not to stop at recording what happens,but to quickly identify the patterns and causes behind it. In other words, we can see not only where the experience is deteriorating, but also what is driving it down and where it is worth intervening before problems become visible in KPIs. What used to involve weekly or monthly analysis cycles, manual categorisation and retrospective reporting can now move closer to real time. Not because we will have more data, but because AI can continuously process signals arriving from different channels, identify patterns and breaks in trends, and provide early warnings before KPIs deteriorate noticeably. (artificial intelligence) (customer experience) (key performance indicator)

2) A shared interpretive framework

The second major transformation takes place at the level of meaning. Information coming from CX is inherently heterogeneous: written complaints, tickets, survey scores, usage logs and qualitative interviews have often spoken different languages. AI delivers a step change when it creates a shared interpretive framework from this diversity: drawing on a unified taxonomy, it can gather evidence from multiple sources for the same issue and map cause-and-effect relationships more quickly. This also transforms the role of KPIs: they are no longer the endpoint, but ‘merely’ symptoms, potentially underpinned by causes at journey level. The emphasis subtly shifts from seeking correlations towards explanatory logic, bringing a new level of quality to CX analysis, even at an expert level. (customer experience) (artificial intelligence) (key performance indicator)

3) Decision integration

The third, and perhaps most important, change is decision integration. AI does not make decisions for organisations, but it redraws the path from insight to intervention: it offers prioritisation recommendations, runs ‘what if’ scenarios, and helps estimate business impact. This is critical because CX’s weakness to date has been that momentum disappeared between analysis and development decisions. If learning and decision-making move closer together, CX will no longer be about producing reports, but will become part of the development logic. In short, AI does not change the purpose of CX; it makes how it operates faster, more sensitive to connections and better equipped for decision-making. (artificial intelligence) (customer experience)

Where should you start?

1) Think small

If an organisation wants to integrate AI into its CX operations now, the biggest mistake is trying to build a ‘large system’ straight away. A better starting point is to select one or two areas where slowness or fragmentation already causes pain, and begin experimenting there with AI support on a small scale. Establishing a POC-based approach offers a good opportunity for this (POC = proof of concept). Typical examples include speeding up insight processing (such as consistently tagging and making complaints, tickets and research materials searchable), or introducing an early warning system (warnings) for critical journeys. (artificial intelligence) (customer experience)

2) Creating a shared language

Another important first step is creating a shared language. AI can be used effectively in CX when it is clear which customer experience dimensions we are monitoring, how we name the same problem across different channels, and who is responsible for turning learning into change. This is not a technology issue, but an operating model: taxonomy, ownership and decision-making rhythm. With this framework in place, AI does not simply sit on top of existing chaos; it begins to bring order to it. From there, it becomes much easier to expand the system step by step: with more and more data sources, increasingly precise cause-and-effect logic, and ever stronger decision support. (artificial intelligence) (customer experience)

The author is Ákos Maczinkó, Managing Director of Works.