There isn’t a company out there that doesn’t want to see a full view of customers across all channels . I have done large and small implementations of those in a variety of technologies . Several ISVs – Hadoop companies , NoSQL companies and traditional data warehousing companies – have all built big chunks of their businesses on the idea of Customer 360 .
For structured data and with sources that don’t change a lot , doing a 360 view in a traditional data warehouse is adequate and in fact has significant performance and maintenance advantages . As data becomes more and more multi-structured and prone to frequent change , a NoSQL database starts to become a great choice . When large datasets come into play in these situations – especially on the multi structured side , Hadoop has advantages .
There are trade offs too – everything that works with standard tooling (BI, administration , ETL) etc don’t work in NoSQL and Hadoop the same way as it does in relational systems . The licensing models have a big impact on decisions too . And while interoperability is much better today – it is not yet at a level that relational, NoSQL and Hadoop all work together happily . I am sure it will get there in some time – but not today or tomorrow .
That is just the back end story – the data management side – where decisions are made on the 3V model . The 3V model is splendid , but it is not sufficient once we look beyond the data management side towards the insights side of the equation . There I think of two other Vs – Veracity and Value of the data .
The universal truth about data – especially big data – is that it is usually not clean . For insights to come out of that data – we need more tools to cleanse and govern . And then for getting value we need to rely on BI tools . If you think the job of a data scientist is sexy – you haven’t been in their shoes . Vast majority of real data scientists spend most of their time cleansing and rearranging data before they can model scenarios effectively . It’s grunt work and mostly not fun . There are great tools which can do cleansing and help with some governance – but it’s an uphill battle
On the value side – there are many different types of BI tools depending on what you want to do with the data . For the most part , these tools need a human user to define what needs to be reported on . Various things like massively parallel processing across multiple cores , cheaper memory , more push down of functionality to databases all have helped tools become more efficient . A whole category of visualization vendors have made BI pretty exciting too .
So between all the data management and BI options – we can already have a pretty good view of customer 360 today . So what is next ?
Enter Cognitive computing !
Some of the existing visualization products support the idea of exploring elementary correlations of a given data set , and that is great. The idea is to have a human user then look at them , try different combinations and see what makes sense . Cognitive goes a few steps further and helps formulate questions that the average users hasn’t yet thought of asking . It can find and surface a lot more hidden relationships across data from all channels . For example – by analyzing news and social media , a customer can be flagged as a credit risk even though inhouse CRM and ERP data show them as having good credit .
What about making sense of voice and pictures and video to enable better customer service ? What if a robot can converse with you and give you all the answers you need based on its cognitive abilities ? I am not kidding – take a look at this video of a cognitive robot dancing when it is asked to.
Even if we discount robotics (which we really shouldn’t discount ) , Figuring out a customer’s intent via voice and facial recognition and responding based on data analysis takes customer care to a whole different level . Imagine calling your favorite cell provider with a question “when is my contract ending” and software figuring out from your tone and historical data that you have abysmal coverage in your office and hence you are a customer who will potentially churn quickly. It can also figure out that in your area there are several other customers who have the same issue . The system can then alert the field services team to fix coverage issues , as well as offer you an incentive to stay loyal to your provider . Or the system can decide that you need to talk to a human agent and transfer your call with all this intelligence to the best service rep to take care of you . That is the power of cognitive customer care.
Responding by voice and touch is great , but it is becoming a given in modern UX. What Cognitive adds to make the UX even better is asking questions about the rationale behind the system’s decisions . You can also ask the system what confidence it has on its decision and explain to you other possible solutions . Imagine engaging in a customer chat about your phone heating up. A cognitive system can help you trouble shoot by yourself . The same system can help a service engineer trouble shoot the network and offer different solutions and the odds for each . The picture below is such a cognitive trouble shooting and debugging app !
Another interesting aspect is helping trade off decisions for a customer when there are several options to choose from – like choosing a phone , the right rate plan and maintenance options .
The one last aspect to consider is about reuse of existing customer information systems . There is good news there too – Watson can query those at run time to make decisions . And if there is an API to commit a transaction in a system , Watson can trigger it too (like say update your billing plan) . Not all data needs to be physically persisted in Watson for it to work .
It’s an exciting new world of possibilities that cognitive opened for us . Plenty of customer projects are already in progress too. Let’s just say I am in geek heaven 😉