Last week of September saw a series of discussions held in a recently refurbished County Hall overlooking the Thames. CX Company hosted a series of discussions under my facilitation. The topic was intelligent assistance and in particular how it is impacting self service adoption.
Probably the most revealing point made by delegates was how new the whole topic was to everyone. AI, as a disruptive force, has only burst through into mainstream consciousness over the last year.
For instance, the increasingly intense competition between metabot vendors such as Amazon and Google to become ‘preferred vendor’ within consumer homes had its birth as recently as 2014 when Alexa first popped up.
As a result, many delegates were in listen and learn mode looking for advice on where to start and the real world benefits in this journey: outside the increasingly hyped headlines that accompany a new generation of technology.
Collect and act on NPS-powered customer feedback in real time to deliver amazing customer experiences at every brand touchpoint. By closing the customer feedback loop with NPS, you will grow revenue, retain more customers, and evolve your business in the process. Try it free.
As a sign of how fast organisations are moving, there has been some breakthrough progress.
In spite of knowing nothing about the topic at the start of 2017, one operations director shared a new service his organisation had just launched.
Traditionally, getting a quote for vehicle damage is a drawn out process. Normally involving a lapsed week of visits to a body shop, booking an assessor appointment and then waiting for them to inspect the damage and calculate the cost.
One strand of AI has completely reinvented this customer journey. Image recognition is now able to instantly price the damage based on an uploaded customer picture. The fact that the organisation in question was offering this without human mediation is testament to the accuracy they found in the image recognition algorithms.
How long before other insurers are compelled to catch up?
Do We Wait Or Get Involved?
Another topic that came up was when is it best to join in? Given the undoubted truth that each year there is going to be significant functional improvements, is it smarter to invest now or wait and therefore leapfrog competition?
This had become a real quandary for one delegate as their senior management had got into a deep debate about strategic timing.
We agreed that in deciding to wait they were ignoring one vital factor in their assessment: the learning that is gained from having a go.
For instance, by January 2017 Aviva had already built a skill to enable its customers to have a chat with Alexa about the meaning of their insurance policy and provide basic education on insurance jargon.
What they will have learnt to date from the analytics of enquiries and user behaviour already provides them with competitive advantage. By the way they are now 15,000 other skills. So clearly, first mover advantage is being recognised.
In a related discussion we debated the relative merits of polling customers before designing anything versus the more agile approach of ‘minimum viable product’. This approach relies on rapid learning from customer reaction and more specifically from analysing customer questions.
Both approaches have their value. But in this fast moving market, speed of execution is certainly at a premium and it is perfectly possible to design and launch an intelligent assistant within an eight week period.
Accessible For Everyone
Moving on, we explored digital adoption and the importance of recognising customer choice.
Even UK government with its long term intent to bring all services under a digital first remit has to recognise that in serving the whole population, they are catering to the broadest of preferences.
People still want to talk to people, so a core design principle in all forms of intelligent assistance is to make the path to live assistance transparent and easy to access.
In fact, understanding these escalation paths is vital. How much traffic still flows into the contact centre as a result of online inadequacy?
When we want to chat or talk is that because another person can really add value? Or is it the result of poor UX, frustrating journeys or some other irritant? It’s an expensive approach if all an advisor provides is a conduit for finding the right answer.
So part of the solution is to make journeys as simple as possible, incentivise customers to adopt self service and provide proactive messaging to ensure customers feel fully informed throughout the journey so the need to check is managed.
But is just for the younger generations?
Apparently not. An energy company recently shared how self service adoption had been consistent across every generation of customer using a conversational interface. Possibly given how similar it feels to a live conversation.
Making Conversations Work
Of course those conversations are only going to work if they deliver what the customer wants. Yet as we know, customers are seldom precise in how they phrase their intent.
Normally an intelligent assistance has to contend with a conversation that begins more like a guessing game based on whatever keywords have initially popped into a customer’s mind. It is seldom a complete, accurate description of their needs.
So, expert conversational design is part of the secret sauce. This involves encouraging the customer to reveal more about their intent by asking clarifying questions and offering suggestions to narrow the scope of possible answers.
Compare this with the experience of using a web site search service which typically returns a long tail of possible answers for the customer to wade through. Of course they won’t and instead look for a faster, easier option.
The underlying technology for recognising context and intent is one of the reasons for the current interest in AI. Both natural language understanding and machine learning act as conversational building blocks that create a dialogue that can be trained for greater effectiveness over time.
The 2017 version is smart enough to draw customers into conversation. The three year horizon as they rapidly mature looks even brighter.
AI driven self service needs both conversational design and technology to deliver convincing user experiences.
Knowing Where To Start
One way to develop a shortlist of self service candidates is to identify your most common customer journeys then prioritise those that can be automated.
Bill Price’s value irritant model is a great way to surface this list. As a former VP of service at Amazon, Bill instinctively thinks in terms of reducing or eliminating unwelcome effort. The model can be easily found with an online search.
Another way to think about the opportunities is to ring fence journeys that are inherently complex, emotional or involve relationship nurturing. In theory, all remaining customer requests are candidates for self service.
However if you are new to self service, then it is more sensible to leave the advanced use cases for later phases in your programme.
Starting out with a low risk, rapid turnaround example increases your chances of a successful outcome and a receptive internal audience willing to invest in more ambitious use cases later on.
Therefore your prime use case is transforming the uptake and outcomes of your existing FAQs and self help on your web site. It’s only an eight week sprint to launch.
I’ve joined forces with CX Company around rapid ROI for intelligence assistance. It’s become a mission to set organisations on the right path before the hype sets in and failure becomes the consequence of over ambitious plans.
We have whitepapers, videos, webinars and articles such as this. All this adds up to a conversation we are holding with the UK contact centre industry on making a success of your first intelligent assistant. We believe the best way is collaboratively. In conversation with sharing of lessons learnt.
If you want to add you own ideas to this discussion or ask for advice, the debate is happening here. We are listening. Please join in.