Gareth Humphreys, CEO at SPG
Hot on the heels of the chatbot craze comes the latest promises of Customer Experience (CX) nirvana, OpenAI’s ChatGPT, AI-powered Bing and Google’s Bard, which lie in wait, ready to replace us humans with unnervingly human-like machines. Most people have, by now heard of these computing phenom, and many of us have played around with it, asking a variety of fantastical questions from “will AI take over the world” to “write a poem about cats” and everything in between. But before you trim your team in favour of the world’s favourite artificial intelligence, how close are we really to machine led CX?
CX transformation initiatives often go like this: an already beleaguered C-Suite, dealing with say inflation, or a cost-of-living crisis, or the aftermath of a global pandemic are asked to make further “efficiencies” and the exciting solution that’s presented is simple – just drive more automation. At this stage, it’s the end-state that’s being imagined and not the cost, effort or pain of getting there. Cost-cutting usually focuses on the resources that are the most expensive and the most difficult to maintain, which is usually your people. This often leads to the conclusion that by digitising the customer experience by launching an app you can get rid of the people and save a ton of money? Great, when can we start?
If only it were that simple – ethics aside. Many organisations have attempted to make efficiencies and improve service by driving people online and letting “artificial intelligence” (AKA chatbots) do the talking. So many strategies are focused on getting customers off the phones and onto a portal, a chatbot or a mobile app. The trouble is, these channels usually don’t solve your customer’s issue if the problem means asking questions which deviate off the linear, well-defined, and easily communicable path.
That’s not to say ChatGPT and other automation tools aren’t the genuine article. Let’s be clear – this is one of the most impressive advancements we’ve seen launch in recent years and has the potential to be a game changing. Yes, upon mass testing there are some clear mistakes, and idiosyncrasies but, for software that’s in its infancy, the expectation shouldn’t be perfection – especially for a machine that’s designed to learn.
So, where does it fit within a CX strategy? That depends on the robustness of your strategy in the first place. Is it specific and unique to your business, your customers, your systems, and your staff capabilities?
The reality is that most digital CX strategies focus primarily on the “path to purchase”; the linear A to B process of moving a customer down the funnel to click “buy”. It is only by understanding the end-to-end customer journey (and the personas of who will embark on those journeys) that an executable digital strategy can be realised.
Furthermore, it’s essential to create consistent, meaningful and measurable digital CX metrics and benchmarks. Although there may be industry or sector baselines, it’s rare to find two organizations with the same targets. It’s also difficult to present a consolidated view of performance given the many disparate data-points obtained by computation or human interaction/feedback. These factors, set against the backdrop of an increasingly knowledgeable customer base that has access to the same or better information than your team, can make it even more challenging to deliver impactful digital CX.
Let’s not forget the pressure from above to unlock big savings fast. The internal teams who’ve seen these digital-first initiatives launch with a bang only to fizzle out when leadership realises it’s not a quick-fix will be nodding along. What remains is the legacy technology that pervades every corner of the organisation, holding back the momentum of change.
Back to our digital friends – the chatbots and ChatGPT in particular, which is only as good as the data it’s been trained on. The impact? Well, the chatbot is largely meaningless with respect to line-of-business processes unless, it has been trained on these prior to launch, which is only possible if there is a mechanism to upload and refine said training data. Currently, the process for doing so is complex and expensive for businesses to do so themselves (those cloud-based processing cycles don’t come for free unfortunately). In addition to the cost and complexity constraints, the enduring concept of “Garbage In, Garbage Out” (or GIGO for short) means that limited or inaccurate data will result in incomplete or incoherent responses to the customer. This can be especially challenging for companies with customers who speak languages other than English that cannot or have not yet trained their chatbot for specific regions.
Customer obsession should be where it starts. An overhaul of your CX cannot begin without a deep dive into your current customer journey, whether that’s one journey, ten or 100. Getting a realistic picture first is key to making the changes needed to deliver efficiencies, and to understand whether its commercially viable to change a journey or not. With this knowledge, you set realistic expectations, preferences, and the building blocks of what makes good CX strategy rather than assume that ‘digital-first’ is the only answer.
TL;DR: your customers still want to talk to you on the phone.
Gareth Humphreys is co-founder and CEO of SPG, a group of companies providing digital transformation, tech resourcing and specialist software solutions to its clients with complex challenges based in the North East of England and Leeds.
