Can Gen AI finally enable the one-to-one future?

Looking back on the last 12 months, I think many of us will remember it as the time when AI reached a critical inflection point and seemed to creep into almost every conversation we have! 

Having lived and worked through the dot-com bubble of the early 2000s, the shift to mobile following the launch of the iPhone in 2007 and the subsequent shift to social media in the early 2010s, Gen AI feels like another one of these seismic shifts. Judging by the number of client requests I’m getting at the moment, it’s certainly a topic that every client wants to discuss. 

Of course, there's a hype cycle (are we at peak hype?) and there are all sorts of practical issues and barriers to overcome (AI hallucination, ethics, regulation, copyright, bias, etc.). But I've seen and read enough to convince me that Gen AI will profoundly impact CX, marketing, sales and service over the coming years, creating both opportunities and threats.

At the Cannes Lions International Festival of Creativity, where EY teams had a number of events and meetings with industry leaders, Gen AI and its impact on the creative industry was a big topic. The EY “Better Question” campaign and poll on LinkedIn summed up the debate perfectly:

One of the biggest opportunity areas for Gen AI within CX, marketing, sales and service is the potential to deliver on the long-time promise of hyper-personalisation.

Hyper-personalisation (also known as "segment of one") is not a new concept. It was first publicised through the book One to One Future, written by Don Peppers and Martha Rogers in 1993. The theory was pretty simple: a local shopkeeper knows the wants and needs of their customers and tailors conversations and offers to each individual customer. Customers enjoy the personalised service, so they return and recommend the shop to their friends, driving loyalty and customer lifetime value. 

The original promise of CRM software and ecommerce was to systematise that 1:1 personalisation and try to scale it beyond the “local shopkeeper” to everyone who interacted with customers (sales reps, call centre agents, marketers, etc.). More often than not, this worked in the short term as revenues increased (e.g., "people who bought nappies also bought baby wipes") but, the sugar rush of short term revenue growth meant that the original goals of true customer-centricity were often taken over by an inside-out mindset of "let's just try to sell customers more stuff.” 

At its worst, one-to-one became the sole domain of marketing campaigns and turned into industrialised spam and surveillance capitalism. That, in my opinion, was never the intent of The One to One Future. The book described the concept of relationship marketing, enabled through customer empathy and trust: “Empathy is the ultimate form of customer insight… new technologies make it possible for even the mass marketer to assume the role of a small proprietor, doing business again with individuals, one at a time.” The One to One Future was first and foremost a mindset change towards empathy and customer-centricity.

Aside from mindset, there were other significant blockers to the one-to-one future, including practical and technical constraints. It was difficult to aggregate customer data across all touchpoints, let alone build a customer insight function and enable a two-way mechanism to automatically spot an insight and act on it by (orchestrating the customer journey, changing pricing, customer service processes, content, etc.). For most companies, the one-to-one future was simply too hard to execute on, hence it became easier to default to just trying to “sell customers more stuff.” However, the promise of Gen AI is to change this. Hyper-personalisation rests on doing three things really well:

1. Capturing customer data (quant and qual)
2. Spotting insights and patterns in the data
3. Continuously executing thousands of experiments to test, learn and scale how the personalisation of anything and everything (e.g., content, pricing, colours, offers, processes, images or text) can drive better experiences and build longer, deeper and more profitable relationships

Although that might sound very private sector focussed, it's not. The same principle can be leveraged by public institutions (e.g., health, transport and tax) to deliver more personalised care, more relevant services and earlier or better interventions. Gen AI (or, more accurately all aspects of AI and ML) super-charges all three elements of hyper-personalisation:

1. Capturing customer data: Everything that was previously qual, is now quant. We can now analyse previously unimaginable quantities of unstructured and structured data (e.g., transcripts of every voice call, live video, images, health data, movement, IoT sensor data or weather patterns). ML super-charges voice of the customer research.
2. Spotting insights: Machine learning (at some point powered by quantum computing) will take us way beyond the crude "people who bought nappies also bought baby wipes" example. Machines will spot patterns in the data that humans would never have been able to see (“people who buy broccoli, on Thursday’s, in the rain, default less on loans”, “people with specific DNA markers who exhibit certain stress patterns on a Monday morning are more likely to suffer from a specific health condition,” etc.).
3. The personalisation of anything and everything: As sceptical as I am about the metaverse, it could represent an extreme of hyper-personalisation. Imagine entering a virtual world where you voluntarily provide your data and in return everything you see, hear, touch has been automatically rendered by AI to suit your needs. Every pixel on display, every interaction, every offer, every nudge, every process is personalised to you and your data by AI.

If this sounds like a dystopian version of Minority Report, there's a danger that it will be. Technology always creates both opportunity and threat, and both uses cases will exist simultaneously. It's a hugely exciting time to be exploring the impact of gen AI on CX, but more than ever, it's essential that we explore it in a human-centric way and think through the broad range of dependencies and enablers.