Knowledge sharing has constantly adapted.
Our ancestors only had conversations face-to-face, Nintendo famously had a phone-line to help out their gamers in the 1990s and now we are witness to bots automatically answering questions.
James Millard, Vice President, Global Support and Community at D2L, has seen many ways to implement Knowledge Management.
He started in Support in his late teens and for the past several years has been leading teams that drive Knowledge Management strategies.
James, knowing you have a wealth of experience with Knowledge Management, let’s start with an overview of your experiences.
James: About 15 years ago, at Blackberry (formerly RIM), I was first introduced to Knowledge Centered Support (KCS), Knowledge Management (KM), case deflection, portal-based support and all of those things that now have become core to support teams.
At Blackberry, my initial focus was to generate a knowledge base that was support-focused rather than product-focused; focus on giving customers problem solving information, not on how a product works.
Combined with the “KCS” methodology, where Support Agents generate Knowledge-content based on their customer-interactions, this has the potential to lead to the almost “magical” approach of allowing your support organization to create and maintain your knowledge base.
I truly believe that every support interaction has a significant amount of value. You are creating customer intelligence and market intelligence. These are the best sources of what is actually happening at a client level and one of the only true opportunities you have to represent every persona, segment, and use case.
It really is the most honest reflection of required knowledge. If you can capture that, repurpose it, and turn it into a one-to-many resource, you are getting a much bigger return on investment than simply answering a ticket. That’s the basic workflow that I use to build a strategy.
That’s a rather ambitious definition. How do you start to execute a strategy like that?
James: I really believe in the concepts of KCS and KM, but execution is always difficult. There are an endless amount of inputs and outputs. I liken the effort to the Shannon Number in chess; the number of 40-move chess games permutations.
It is larger than the number of atoms in the universe. You have to accept in Chess and in Support that there are an infinite number of things to control and you will never have full oversight.
So, I start with a baseline. The key is to find one that won’t be a failure given your use cases and customer-base. Then you figure out how to measure and gain operations oversight as best you can. And finally figure out and improve over years and years. Knowledge is always about constant improvement, not about meeting a goal by initial roll out.
Craig: I am a big believer in contextual knowledge, Google’s KB has some interesting features where it recognizes you are logged in and can actually annotate an article based on who you are. I also see this trend in CX. Customers love real time help, whether via chat, bots, or in product widgets.
However, contextual knowledge needs that authenticated experience versus public available knowledge. Do you need both?
James: Yes. I have been on a real journey for the last 10 years. At first it was let’s just stand up a knowledge base, literally just a repository of articles. Then the “gold standard” became deflection, which only serves existing customers who need support.
That is likely a narrow segment of who you want looking at your knowledge base. So the next evolution is knowledge sharing across the entire customer journey. And this is where both of those are necessary.
I am going to be investing a lot of time and effort into rebuilding these tools in the coming year. We realized by digging into win/loss reports and competitive analysis, that when a prospect is in the initial phase of the customer journey, they need to understand easily if you are the right solution. No customer is going to read into your technical workflows.
More likely they will skim through and try to understand the basics and what product appears to be the easiest. A big part of what is easy is what happens when I look for help. And, for prospects, it is almost always a non-authenticated part of the journey, and SEO and tagging are much more important. This has to be a priority.
But overall, authenticated experience is really the key. This is where the magic happens for your customers.
Can Machine learning solve contextual knowledge?
James: I have invested the most energy and money in embedded contextual help. We are so far away from automated support. Today, machine learning cannot do the analysis. Meaning that it cannot figure out from scratch how to formulate and deliver the answer to a question.
Machine learning is really good at connecting you to knowledge. There are two parts of this:
- Contextual – I know who you are, I know where you are in the product, I know your product usage history, and I know what most commonly happens with someone who matches your attributes and context. Building this means learning every time somebody who has those attributes does something, then we can get into classical support of answering not only your question, but anticipating, based on workflows, the next couple questions you are likely to have. The technology is developing today to make that happen. But creation of these assets is still a while away for almost all applications.
- The other is Chatbot and Decision trees. These work well when the solution set is relatively simple. But when the complexity is thousands of workflows or use cases, bots which try to deflect through decision trees will likely not work as well.
When it comes to knowledge delivery, there are 3 ingredients:
- Your knowledge assets, as in your written down assets.
- Your knowledge storage, where the knowledge is indexed and stored.
- Your portal, where it can be found by the user.
There are many ways to combine those three, but the idea is bringing all of these into the product and leveraging technology to expose it easily and in context to the user. Asset creation is the missing link for automation. That part will need humans.
If I needed to invest in a knowledge management solution, what would I need to think about?
James: All solutions are not built equally. My greatest pain is using a community platform that was clearly built for an e-commerce community versus a customer community. These are completely different things, but it is often hard to convince the decision makers what the difference is because they both have the word community in them.
So create an ideal client learning journey. In a perfect world, I have designed a learning experience that makes sense and am leading a customer or prospect through the experience I want. The tools need to facilitate that journey.
Also, operational awareness is so important. You need to invest in the data you want. It is easy to get to a point where you have hundreds of thousands of customer touches and they grow exponentially. So you need to have your metrics decided upon to help maintain trust with all stakeholders and decision makers. Support is expensive and hard to understand when you are not in that world, so you need to ensure you can explain the ROI clearly and the tool must facilitate that.
And lastly, you need to have your own private review of “how are we doing?” As I said earlier, knowledge is constantly evolving. It is vital you identify issues and efficiencies everyday, which helps you build your future road map.