By Neil Movold | May 7, 2018
Like many journeymen, I have had the fortunate opportunity to build up a long career of experiences across many industries and around the world. The best part is that most of those experiences had involved working at the bleeding edge of technological evolution. From being named one of the most influential individuals in Bermuda’s history when I opened a whole nation to the wonders of the Internet, to starting a company focused on building a real cognitive system based on the concept of team cognition.
I regularly see things as they can be and ask why not, while at the same time having the unique ability to connect the dots between business and technology that can create step-wise, actionable value where most people struggle to do so. Some may see that as being either a good thing or a bad thing.
My journey so far has led me to be living in New Zealand, with my Kiwi wife and two kids and a passion for using Cognitive Computing technologies to augment and amplify human abilities so that we can think and perform our jobs better and faster. In particular, I love focusing on projects that have socially beneficial outcomes.
I picked the main title “Thoughts from the wild” for two reasons. First, we are in the early days of understanding and achieving meaningful, scaled outcomes from the evolving world of AI and Cognitive Computing technologies – hence the “wild” part.
Second, I wanted to author an article (and hopefully a series of follow-up articles) that pushed aside the overabundance of hype and marketing speak, to share what I have learned and experienced first-hand, mainly around how New Zealand and Australia are embracing the cognitive era. Given the connected world we live in, I am sure there will be similarities to other economies, industries and organisations, so please do not stop reading at this point.
In this article, I will be discussing organisations moving from seeing people as transaction makers to real relationships with people that generate transactions.
Struggling to exit the Transaction (Consumer) Economy
Let’s start from a macro-perspective. New Zealand is a young, evolving, small market economy that along with Australia, have robust and entrenched transactional economies and businesses. The economists will tell you that things are going well in each country, with all things considered. However, as we all know, the world is changing rapidly.
Transaction-based economies, systems and processes have their foundations based on one-and-done transactional events, where metrics such as average sales price are critical to success. Business systems and technologies applied to monitor and manage these transactions and associated metrics are based primarily on structured data and are deterministic in nature.
There are three overarching issues with transaction-based economies which are fueled mainly by the use of structured data, namely:
- only 20% to 30% of all corporate data (aka corporate memory) created is ever used, creating huge risk over time as stakeholder disconnects across silos undermine the data’s value; most if not all in some cases of this data is of a structured nature;
- structured data makes up no more than roughly 20% of the ever-increasing amounts of corporate memory (aka corporate data); and
- structured data holds little to no context associated with an event other than of an implied nature.
These issues are growing with increasing impact in our ever-changing world. They leave the corporate memory with blind spots when making decisions, increasing the gap between business value, cost and risk over time. Along with these issues, the value of knowledge assets within the corporate memory traditionally diminish in value soon after their use (e.g. project related documents and emails once the project has finished) and are often a source of increased risk to the organisation in the future.
For most organisations, the following diagram represents the typical evolution of their transaction-based business continuum which attempts to bridge the gap between business and technology through mainly the analysis of structured data:
Organisations of all sizes can be at any point along the continuum, and that point depends on any number of influencing factors, including the level of data culture throughout the organisation, board engagement, skills and resourcing. For most, the concept of cognitive analytics is so new and foreign to them that it sits outside their mandate, mainly because of the lack of clarity around what it is.
With the issues listed earlier around the long-evolving love affair with structured data, businesses can only see things on a superficial level as they continue along the continuum trying to glean value from the collected transaction-based events taking place internally and externally to their business. The unintended and avoidable costs associated with this ongoing approach can have significant downstream consequences.
Ironically, as I was authoring this article, a recent report published by New Zealand’s Productivity Commission, showed that in New Zealand, "the hours worked per capita are the highest in the OECD, but the value produced from the New Zealand labour force is the lowest. Our wages are lower than other OECD countries, again because our productivity is lower."
This situation did not happen overnight and likely involves many influences and variables still to be understood, possibly even yet to be identified. Although I do not have conclusive evidence, I believe that this research supports the notion that the tools of economic measurement need to refocus broader than current structured data models. Achieving this with a new set of lenses and approaches will harness all the data within the corporate memory.
Technology is a key driver and enabler of productivity, and most of the leadership ranks understand that. However, from my experiences to date, I can only summarise my thinking to believe that most decision makers are stuck in an evidence-based paradox, while at the same time knowing they cannot ignore the overwhelming influences of AI and cognitive technologies. This paradox has led to widespread tinkering with these new technologies, often struggling or failing to understand the inevitable conflicts that will strike against their organisation’s transaction-based business models and mindsets.
I was at a breakfast event recently, and in response to one of the speakers, everyone who put their hands up to acknowledge engaging with cognitive technologies (primarily chatbots) have not gotten past "playing with technology", even after 12 months working with the technologies in most cases. Moreover, it was not due to a lack of skilled technical resources.
In some cases, decision-makers made a mistake in believing the marketing hype that an all-in-one solution that "uses AI" will do “the job”, creating new value opportunities for their business. I am not saying that these solutions might not do what they say they can do, but from my experience, the often-over-hyped offerings that we have today just don't come with the batteries included that the buyers think they do.
So, where is the disconnect?
Moving forward in the Relationship Economy
Our world is made up of an ever-changing web of human relationships, yet several generations of learned behaviour propping up the transaction economy, have many of us relate to people like a transaction and mistake what we have for a relationship. For many organisations, operational processes and activities focus mainly on keeping these transactions moving, with growth strategies focusing on creating more transactions more efficiently.
They do this without the awareness of the distinctions between a transaction and a relationship. Without clarity and understanding about the nature of our connections, all sorts of unproductive behaviour and outcomes can happen – leaving a trail of people behind due to unmet expectations, lack of poor communication and confusion.
Humans and our relationships with others are in a constant state of becoming, yet in an economic environment, we often relate to one another as fixed in time, where the focus of attention is often on the average selling price over lifetime customer value. Hence, there is a natural outcome of little to no tangible commitment to each participant of the transaction’s long-term success.
Today’s successful businesses are embracing the Relationship Economy. They know that rapid scalability enabled by modern technology has made relationships and their implications global. They understand that relationships exist both internally and externally to their organisations and both need to be built with trust and loyalty cemented to enable meaningful and long-term ROI.
Going beyond traditional branding efforts, these organisations understand that long and profitable relationships have their foundations based on highly relevant and value-driven conversations.
This universal human experience is the disconnect that I believe is missing in most organisations. It is all summed up in one word – conversations.
Conversations are at the core of economic and social transformation
Conversations have played a critical part in human evolution. They enable the exchange of knowledge and understanding between two or more individuals, along with a timeline in a way that utilises thought, experiences and the senses.
I see conversations as being the convergence of three key enablers – relationships, a timeline and human cognition. These enablers have been at the heart of human conversations since the beginning of documented history. What has changed over time are the mediums through which we engage in conversations and capture their essence. The cavemen used drawings on cave walls as their medium for capturing their conversations around a variety of activities or events important to them.
In today’s world, we often hear that conversations (i.e. customer engagement) build the relationships that drive customer value and experience. These conversations are increasingly taking on more digital forms, such as emails, documents, social media, voice or video.
However, where are these conversations?
We know that around 20% of all corporate data is of a structured nature, representing such things as average sales price, sales volumes or inventory counts - all bits of transactional data that tools such as business intelligence or analytics have had entire industries built around them.
The remaining 80% is made up of dark data – usually unstructured data that organisations capture and store, often as part of their regular business processes, that most organisations simply fail to use and find expensive to store and secure, but in some cases do so for compliance reasons. Unfortunately, it is mostly in this expansive dark data that conversations are captured in their various digital forms, such as emails, documents, and audio calls. The obvious realisation then is that these conversations hold invaluable insights that traditional approaches to managing corporate memory cannot deal with and surface.
Let’s stop and think about this for a moment. Take a business-to-business transaction. There is a clear starting point (e.g. initial customer inquiry about a product) and an ending point (e.g. customer purchases a product on mutually agreed terms). With a more complex transaction, the timeline may go into days and weeks. It may involve multiple individuals and entail many conversations between these individuals as they work their way towards a worthwhile outcome. There may be emails, documents, audio or video calls that capture the conversations that enabled the successful transaction to take place.
In the old days, conversations were often made in person, where each could use their cognitive skills to understand the other person’s needs and wants. Those involved were constantly learning from each interaction and providing (hopefully) greater value in the future from these experiences. Most of the conversation, in all its complexity, was captured in their heads, ready to be harnessed in the next conversation as the participants move along a timeline to some, hopefully, worthwhile outcome.
In today’s world, conversations are increasingly made digitally, with face-to-face interactions becoming more of a rare thing. While I would agree that interacting through digital means using ongoing advances in technology can improve productivity and drive economic growth, there is a growing elephant in the room that most are just not seeing or believing to be a real problem that can be solved.
Let me put this into perspective with a recent experience I had. Although not in the business world, I hope you connect the dots as I did when reflecting on the experience as it relates to the theme of this article. Several weeks back, my son was in the hospital to have his spine fused as a result of being born with spina bifida and enduring 13 years of life as an active para-athlete. His orthopaedic surgeon (Haemish) has known him since birth and since that time my wife and I have had many conversations with him as our son is a frequent flyer within the hospital – ironically called Starship Hospital.
I have seen Haemish record his thoughts on an electronic recorder, including remarks about my son’s athletic pursuits, type notes into the hospital systems and on paper for his team to deal with and pull up numerous x-ray, MRI and CT scan results on screens to review. In the case of this recent surgery, Haemish also consulted with overseas experts to determine how best to approach the surgery, even though Haemish is a world-renowned expert in his field.
Why would he want to do that?
There was a traditional approach in place, and several resident doctors (usually fresh out of med school) that follow Haemish around went straight for the traditional approach when asked what they would do. Their response was predictable though as they were trained to respond with the most common approach to such situations, but that was not the approach that Haemish decided on in the end.
You see, only Haemish was involved in all those conversations since my son was born and the conversations leading up to the surgery. From all those conversations, he knew what the best outcome would be to improve my son’s quality of life as an active para-athlete. There were no additional costs involved, no additional resources required and no greater risk of achieving a successful outcome. There was just a learned alternative way of handling the situation with the tools he had available to him.
Haemish could take everything he had heard and learned over the years from the conversations he had to, in my opinion, out-think the traditional approach. However, as I mentioned earlier, many pieces of these conversations have been digitally captured over time and if someone were to go back through all those captured pieces, they may too have come up with the same conclusion and approach that Haemish did.
Sure, we could digress into a discussion around the inefficiencies of modern health systems, and I know them well, but that is not seeing the elephant in the room and best left to another place and time.
Conversations and their involved context
From the perspective of your organisation, try and answer this question – where in your corporate memory are the conversations that enable your business to exist?
If you are struggling with how you should be thinking about your answer, it is critical to understand that for every conversation, there is an involved context. Read that again to ensure that it sinks in. A conversation is meaningless to anyone without an accompanying context.
In the face-to-face conversation, each has a greater chance of understanding the context that would enable a successful transaction to take place. That is because we can more readily pick up on subtleties such as hearing changes in the other person’s verbal tone, hearing the use of certain terms or phrases, noticing changes in physical posture or observing something of value in the surrounding environment. Each will remember and re-use elements of the conversation’s context in future interactions and again learn from the outcome.
So now think about this question - how can we understand the context that formed the circumstances which enabled a successful or unsuccessful transaction to take place?
Context is not a thing that can be identified easily or concisely, like an average sales price or profit margin. It can be a complex beast that requires a different set of lenses to understand, especially in our ever-changing world. Context can build over time, often based on human data that is made up of ideas, is diverse, dynamic and lives everywhere in the corporate memory.
As you might now be thinking, context plays a part in many ways along an organisation’s journey, regardless at what stage of maturity the organisation is at. More pronounced in recent years, corporate amnesia is an area of growing concern to most organisations. Corporate amnesia is seen to result from the assumed loss of accumulated corporate knowledge due to employee departures, changing workforce demands and a fragmented IT landscape. I purposefully used the word “assume” in that sentence. Remember the issue I stated earlier in this article that only 20-30% of all corporate memory is ever used?
To this point, it is of interest to know that corporate amnesia is of particular interest to asset-intensive industries, such as the energy sector. Drawing on my recent experiences in that industry, it has become clearer to me that the problem of corporate amnesia is not how best to capture the knowledge that is departing.
The problem is more in finding the knowledge that has been already captured and left behind, understanding the context within which the knowledge applies and reusing that knowledge where valuable in the future.
There will certainly be emails, documents, reports, photos and possibly audio and video content too. Think about this. Much of a departing employee’s knowledge was created through conversations with others, and much of those conversations were documented in one or more digital means.
Traditional approaches fail to process this valuable data. However, with ongoing advances in cognitive technologies, organisations now have a growing ability to leverage all of their corporate memory. This ability is especially valuable when there are situations where the overall context is complicated, such as when there are different parties and numerous individuals involved.
As humans, we excel at reasoning, deep thinking and solving complex problems, but our ability to read, analyse and process huge volumes of data is quite poor. Plus, humans are generally time constrained, subject to personal limits and limited by unconscious biases that influence the decision we make.
Enter the growing world of cognitive technologies to the rescue!
Using cognitive technologies to provide conversational understanding
Consider the world of call centres or telehealth services. A world where humans historically call a phone number to talk to someone knowledgeable that can hopefully help address their need or problem. These are commonly recorded calls, and the audio files are usually attached to a call record within the organisation’s CRM systems. For each call, the call taker would be required to fill in fields within the call record and capture as much of what they thought was valuable.
Now without going into all the details, it would be a good bet that the accuracy of the information captured by the call taker is varied at best across calls during their shift or period of their employment. After all, they are human. What happens when the follow-on conversation moves to emails and possibly handed off to a different person to handle than the original call taker?
Have you ever played that game where you stand in a circle with a group of people and one person whispers a short story to the person beside them and so on until you get back to the original person and the story they hear is not what they said in the first place? A pretty simple scenario but think about this in a business environment at scale. The original context of the conversation is inevitably muddled.
What are cognitive technologies?
Bringing everything I have said so far together, cognitive technologies are now offering cost-effective ways to help with these previously unaddressable scenarios or problems. Cognitive technologies are powerful for augmenting human intelligence with capabilities that are related to the ways humans think and demonstrate their cognitive abilities in everyday life. Cognitive technologies can be used to create a customer knowledge layer that enriches data collected over years of interactions, both internally and externally. Every interaction and outcome can educate the platform, helping it become even more effective over time.
The best part about the growing world of cognitive technologies is that in their basic form they are lightweight and quick to deploy through APIs. Every major vendor (e.g. IBM, Microsoft, Amazon, Google, Salesforce) and growing number of startups offer a suite of APIs to access a variety of cognitive services, such as speech-to-text, text-to-speech, natural language processing, image and video processing to name a few. The pricing model across most of the vendors is a pay-as-you-use model, with a free plan for usage under a certain level of API calls. This is ideal for smaller organisations that cannot afford to pay a monthly minimum if their usage will be inconsistent.
I must make it clear though that with these technologies, the batteries are not included! You need to have the capability to build a software orchestration layer to call out to the applicable APIs when needed for the task at hand. While vendors are providing access to their cognitive services through APIs, most also offer full software solutions that use their cognitive API services. At the moment, these solutions are mostly targeting the large organisations at a price point that only they can afford.
I can assure you that taking advantage of the currently available cognitive services APIs is both cost-effective and value transforming over a relatively short period. Plus, these services will only get better as the technology evolves.
Let me put that into perspective. I recently wrote less than 150 lines of code for a proof-of-concept that allowed me to analyse video files, translating speech-to-text, pulling out static frame images with facial recognition and showing me tone and sentiment analysis across the transcription. To produce that kind of output, each API surely had thousands if not millions of lines of code that I did not have to write. How valuable is that?
Engage cognitive technologies now!
With the benefit of accessible, cost-effective cognitive services APIs, organisations can cost-effectively engage now to enhance customer engagement through personalised understanding. They can also scale and elevate their human capital expertise, infuse products and services (internally and externally) with intelligence, enable automated, intelligent business processes and power disruptive data discovery and exploration (after all, we don’t know what we don’t know).
My point is that from my experiences so far, there is no need to wait until the end of the traditional business continuum to explore AI and cognitive technologies. I often get the impression, directly or indirectly, that many decision makers see engaging in a project involving cognitive technologies as being at the same scale as implementing a new ERP system, with the required level of resourcing for implementation, development, change management etc. This could not be further from the truth in most cases, albeit it does depend on the size of the organisation.
Start with a small proof-of-concept that addresses a narrowly scoped problem. This can be done with little to no interference with existing workflow and processes. For example, use a tone and sentiment analysis API to monitor incoming and outgoing emails within a contact centre to identify situations that may be heading to the point of conflict and an unhappy customer. I developed a proposition just like this for an insurance company, and the value of this little bit of tech heavily overshadowed the costs involved.
The diagram below shows a revamped business continuum that can embrace cognitive technologies at any point or multiple points along the continuum, at a scope and scale that fits with current business and technology capability. Cognitive technologies can be used to enrich existing processes with simple API calls and assist humans and processes with more complicated and robust intelligent systems.
Bringing it all together - the value of lived experienced voices
I want to finish with another real-world example that pulls together all the things that I have said in this article. It is a fascinating and socially impactful project that I am currently leading in the telehealth space, including mental health services. The national organisation I am working with has thousands of hours of conversation audio stored away. Although captured, these audio files are at an arm’s length link away from their associated call record within the CRM system. It is a time-consuming manual process to go back and review the call audio files, and that is primarily only done for quality assurance purposes and randomly at that.
The lived experienced voices contained in these audio conversation files hold insights unattainable through the current systems that capture the nature of the conversations in a naturally deterministic and biased fashion. Each conversation audio file contains the interactions between the call taker and caller and is fully observed.
Research shows that conversation quality directly influences client outcomes. That could not be more critical than in mental health services where the single most critical KPI measured is whether a life is saved or not through the use of the service. The problem definition for the project focused on this key question – how can conversation quality be cost-effectively improved?
Overall, the goal is to provide help for understanding, monitoring, and improving call takers’ capability and confidence in dealing with a broader set of presenting cases. Specific outcomes include identifying and providing new opportunities for training in areas of adaptability, dealing with ambiguity, employing creativity, conversation progress and changing perspectives, along with improved quality monitoring and answer/suggestion support tools (both for the call taker during an active call and for the caller in a self-help manner).
We are using a variety of AI and cognitive technologies, including speech-to-text, deep learning, machine learning, natural language understanding and processing, acoustical processing and graph databases. As mentioned earlier, we pay for the APIs only when we need them, which in healthcare is a massive win in cost efficiencies.
For example, we are using these technologies to go back in history to process all the existing conversation audio files to understand the situational context of each caller better and identify potential patterns in speech and behaviour across callers and geographical boundaries that could help predict where scarce resources could be applied more efficiently. Another example is to provide real-time support during an active call to notify the call taker of potential changes in speech and behaviour during the conversation along with options to take that could aid in improving conversation quality and a better outcome for the caller.
In both of these examples, we are not changing the existing workflow and processes that have been established over years of effort. Instead, we are finding key points in those processes where augmented intelligence will add the most value and infusing small pieces of software (think plugin) that fit seamlessly into the user experience, with minimal distraction but a noticeable improvement in quality overall. This is much different than thinking about replacing an entire system with a new one and going through all the lengthy bits of change management, training etc.
Hopefully, you can see by now how valuable it is to focus on the conversations that enable your organisation to exist and how the use of cognitive technologies can both augment and enhance the value your organisation provides.
I have by no means finished my journey, and every day I experience new things and learn more about how our world can benefit from the growing advancements in AI and cognitive technologies. I hope to write about these new experiences in future articles.
I believe that economies and businesses need to out-think and outlearn the competition, by embedding learning in every process and empowering people with all forms of digital intelligence. This learning and digital intelligence should focus on the conversations that enable the business to exist, recognising that conversational data spans across all facets of corporate memory in both structured and unstructured forms.
With this kind of reshaped data culture and focus, intelligence harnessed from conversational data through cognitive technologies can help lower costs, lower risk and increase value. I want to emphasise that lowering risk is often left out of the equation as it is seen as difficult to quantify. Do not make that mistake. Since the technology is lightweight and quick to deploy, cognitive technologies can positively impact organisations within weeks.
Reach out to Neil Movold on LinkedIn.