KSF Expert Session on Data on the 2th of October 2018

22 October 2018

The opportunities offered by Machine Learning for customer interaction

On 2 October, we held an expert session with the Customer Service Federation (Klantenservicefederatie, KSF) at our office in Amersfoort. The speakers – our colleagues Jeroen Kromme and Sebastiaan de Vries – talked the participants through the opportunities offered by data and the steps you can take.

The presentation began with a picture from “Where’s Wally?”. You can see a picture of a beach full of these figures. In between all those figures you need to find Wally (who always wears the same stripy top). It’s actually the same with data: there’s so much data, how are you going to find and use the data that is relevant for your organisation. The participants of this session are data analysts, innovation managers and customer interaction professionals from different organisations. Jeroen and Sebastiaan (data scientists) guide the participants into the world of data and opportunities for customer interaction.

From insight to application with data science

Data science is about 3 things:

1. getting insight into the data by combining data from multiple sources,
2. making the data comprehensible by dashboarding, and
3. the applicability of the data by asking the question: what can we do with it?

To experiment with this and to innovate, we regularly have “Project Friday”. Which is an initiative to learn and go in search of the boundaries of current knowledge about A.I. and Machine Learning. We write an algorithm and dismantle a coffee machine in order to program it to link faces to preferences. One of our colleagues goes and stands at the machine, the coffee machine recognises them (facial recognition) and then the person is given the coffee of their preference. It is an example of applying A.I. and using data in real time to satisfy the customer’s needs. For an extra strong coffee on a Monday morning ;-).

Maturity model

One of the models we use regularly is the “Maturity model”. It is a model that offers insight into your organisation’s stage of data maturity and consists of 5 stages:

1. Reporting. Data is collected manually and often distributed in Excel format.
2. Dashboarding. Data is in 1 place, fully automated, visualised, one truth.
3. Tactical. Data is proactively used to optimise processes, such as to find the reasons for churn or quality control.
4. Strategic. Data is also used for strategic decisions, e.g. to identify customer needs.
5. Real-time. During a call, the agent offers potential solutions and Next Best Actions.

How the model works

Jeroen uses a practical case study, Vivat, to demonstrate how the model works. Vivat had lots of knowledge about their KPIs and had collected plenty of data, but their data was coming from seven different sources. It was all collected manually and then distributed using lots of Excel files.

Do it yourself

Next, the participants got to work themselves. Not everyone is at the same stage of the model. And every stage has challenges of its own. Everyone wants to get to change 5 but not all of the participants are there yet.

As the participants put it: “There is already a lot you can do thanks to technology, but it also demands a lot of operations and company culture, and therein lies another almighty challenge.” They also believe in the cooperation between a chatbot and a human: “When using real-time data, that cooperation is highly important”. Chatbots are still not far enough developed to be able to replace the agent.

The group was split into four groups. Each group was tasked to work out how you can get from one stage to the next. All the groups got cracking. Here are the groups’ main conclusions:

• Make sure your definitions are consistent so that data systems and other systems can talk to each other
• Transfer the data into your business and make it comprehensible for everyone in the organisation to see what particular data is saying. Data is needed to optimise and improve customer interaction.
• Seek out the correlations: data in isolation is meaningless, but it is meaningful in combination with other data. Adjust the dials and switches to see what happens.
• From reactive to proactive: use data to ensure relevant customer interactions. Also consider environmental factors that may have an impact.
• When you start working with real-time data: keep constantly, on a loop, comparing past and present data. Has it deviated? Why? Why not? By continuing to do that, you can make the system more and more intelligent (Machine Learning). Control is still important, even if the machine is self-teaching. The role of humans is important in this: a great example of the power of cooperation between people and technology.

Valuable session

One of the things that made this session so fun was the interaction from the group. It provoked fascinating discussions and there were good questions. It also turned out to be valuable to gauge how far the various organisations have come and how they are using data to reach the next stage of the Maturity Model.

Are you curious about our “Project Friday” with the coffee machine? See the images here.