Insights: Blogs

Dr Mark K. Smith
Dr Mark K. Smith CEO

‘In the future, the value of this business will not be measured by the communication we are about to do, rather it will be the communication we have already done.’

In the early days of ContactEngine I liked to show off. In many ways showing off is a professional requirement for a CEO in an early stage tech firm. Looking back, many of the things I said were utter nonsense – however, one line that has stood the test of time was this: ‘In the future, the value of this business will not be measured by the communication we are about to do, rather it will be the communication we have already done’. What was in my mind was data.

All science is built on a foundation of a good idea supported by evidence. It’s summarized very elegantly in three words: method, results, conclusion.

Well, unpick that for ContactEngine and the method was ‘omni-channel automated dialogue’, the results were ‘successful automated conversations’ and the conclusion was us ‘making a business process more efficient’ in some way.

What surprises me is just how fascinating this journey became. On the face of it, ContactEngine does a simple thing: we take a process like an appointment for an installation, repair, delivery, etc., and, by cycling across all available channels (text, e-mail, voice, mobile web, IM, video, images, even very occasionally a letter…), we reach out to the customer, listen to what they say (automated with a sprinkling of AI magic), answer them and everyone is happy.

The customer gets what they want, when they want it and the client gets to provide a laser-focused service that we can prove the customer likes. The NPS/VOC surveys we carry out give stunning, industry-leading results for our clients. Why is an improved NPS/VOC score better? Because customers will forgive defects in perfection so long as they know what’s happening and when, and what’s more, we’re gutsy enough to charge for success – per transaction, not last millennium per seat licence silliness.

So why is this fascinating? Well, because I was right: the value is the ‘communication we have already done.’ By asking a narrow question -- ‘Hi Mark, we’re coming to yours on Thursday at 2pm is that okay? -- we get a simple reply -- ‘Sorry, that’s not very convenient now. Could you try next Tuesday please?’ And this is where the magic begins, where the ideas started. So let’s unpick the simple and ask ourselves some basic questions. I know it’s not quite Isaac ’too clever by half’ Newton under his apple tree, musing on gravity but bear with me……

What channel? (Don’t dictate terms, silly – omni-channel ain’t just about buying, it’s about fulfillment. Let your customers decide.)
What time? (Yeah, of course responses vary by hour – and we know which hours because we’ve got the data.)
What day? (You’d be surprised to know which day is best for responses.)
What month? (A two-hour variation in response between mid-winter and mid-summer, anyone?)
What words? (Look, Mark used punctuation – bet he’s 52 and has an IQ of 155.)
What meaning? (Let’s spin that question to our own Machine Learning Algorithm and it will work it out. How so? Because we have hundreds of thousands of pieces of data to learn from.)
What reply? (Hey, maybe we change the words we use based on the language Mark uses. More punctuation for him – he’s well picky.)
What next? (Wait for a reply and answer that too; if there’s no reply – try, try, and try again, and if it’s just a pleasantry, e.g. ‘thanks guys’, then say ‘thank you’ back ‘cause that’s just polite, right?)
What kind of person? (Maybe we know more. For example, tenure, socio-economic group – Grade A, obvs – churn risk, gender, age, network level issues, call outs for warranties, pain in the neck quota, etc.)
How loyal? (Don’t take the p*ss because they’ve been with you forever – look after them ‘cause they like you and they’ll tell others they like you as well.)
How old? (52. Easy, right?)
More AI or more human? (So, it’s a really stressful call – tell you what: why not present a ruddy chat bot? That will win favor. NOOOOO! Use humans when humans are needed and machines when machines are best.)
Escalate? (So someone is not happy, and they text back ‘*&^% off’. It’s best not to say ‘Sorry, we didn’t understand that. Can you say that again?’……. Escalate to a human [see point above].)
De-escalate? (Someone says their stuff is working okay – so get that back to the business real fast to stop that unnecessary truck roll.)

So omni-channel dialogue is really properly fascinating. The more we know, the more we understand and the more we understand, the better we become. And all that from just a few (hundreds of millions of) texts, e-mails, phone calls and the like.

Isaac Newton eat your heart out.