It’s tough to find that one use case that serves as the catalyst for change, and which avoids the trap of overloading your teams with cool, shiny new capabilities. And that’s why it’s time to get cracking with the ‘brilliant basics’. Just because they’re labelled basics doesn’t mean they aren’t impactful. They just don’t sound sexy, and so they often get overlooked. The call centre is one of them.
The call centre is one area that gets a lot of bad press, often getting labelled as an archaic cost-centre. In a modern experience design, the service blueprint of an organisation should be centred around intelligent data, but this doesn’t automatically point to deflecting everyone away from human touchpoints. All of the data usually exists to create a Call Centre Intelligence model focused on uncovering the patterns of behaviour which lead to higher or lower propensities to call, as well as other key outcomes (e.g., conversions, higher LTV, and lower churn). The big challenge is unifying the data sources at scale and transforming the data through models which can identify and drive change… AND trying to avoid sending data to any external vendor’s cloud storage… AND achieving all this in weeks, not months.
Being too prescriptive and jumping to a digital self-service strategy for large cohorts of your audience is a risky path when not informed by a comprehensive journey analytics model. Journey analysis is central to Call Centre Reflection (yes, reflection not deflection), taking all the events that lead to a service engagement, organising them into sequences, and identifying the paths which lead to contact. Once this is set up, action strategies can be co-created to address the pain-points and inefficiencies in the service blueprint.
Journey analysis is central to Call Centre Reflection (yes, reflection not deflection), taking all the events that lead to a service engagement, organising them into sequences, and identifying the paths which lead to contact.
More importantly, when executed well, this model reveals non-linear paths that lead to multiple outcomes (e.g. contribution to future re-purchase, return, and higher lifetime value). Taking a holistic behavioural approach to analysing, predicting, and acting on call drivers ensures that future investments can be focused on the right mix of experiences for your operating model. This may include assistance during critical decisions, such as: “Should I have live chat?” “If so, why, when and to whom?” Then, and very importantly, providing the central data and decisioning capability to test these types of experiences, utilising your own unified intelligence hub. These can include:
- High call propensity from a single page or web journey, indicating a poor UX and/or poor messaging or signposting to a help article
- Increase first call resolution (FCR) by including enriched behavioural context in agent portals
- Attribute value to service touchpoints in order to predict and recommend the true next-best action
- Build and deploy a deep-learning model that is underpinned by cross-channel behavioural context to drive AI service tools including: chatbots, contextual search suggestions, authenticated context push to personalised IVR, and dynamic pre-emptive bill messaging/explainers (although that still recognises that human engagement may be necessary via active-takeover)
- Incorporating emerging service touchpoints into the analysis, such as voice apps (Alexa, Google Assistant), Facebook Messenger, and WhatsApp
There are many capabilities and use cases for a Call Centre Reflection model but some of the more common starting points would be:
- Identification: To leverage a comprehensive identity graph and stitch touchpoints together into sequences
- Contextualisation: To apply a behavioural schema that combines intent with call/voice/transcript logs translated via text/utterance modelling into usable, multi-modal interaction flows
- Sequence: To deploy sequential journey modelling, leveraging insight detection which uncovers common pain points, as well as hidden positive paths
- Recommendation: To execute productionised predictive models to output propensity-driven next-best action suggestions based on 1:1 and look-alike behavioural modelling
- Simulation: To enable productionised multi-variate testing across touchpoints to take action on the insights
I feel fortunate to have a network of kindred spirits, with whom thoughts and stories are shared over drinks. One particular friend of mine who works for a large pureplay shared her pain and what she wants right now from service insights. She’ll remain anonymous because what happens in the pub stays in the pub.
“We have really loyal customers, but we fail to proactively identify and solve their issues. In fact, we seem to put up roadblocks that drive up the volume of contacts via call, email and live-chat. I want to generate a hypothesis based on user complaints and use them as the primary ideation technique for the optimisation program.”
Anonymous, large global pureplay
I believe the New Normal for customer experience is to optimise in context, not in isolation. No ifs, no buts, and certainly no central brain sitting externally to the big data environment. The biggest constraints for optimising cross-channel touchpoints at a 1:1 level seem to be:
1. I don’t want to send PII (Personally Identifiable Information) to my experience activation platforms
2. My call centre system has no connector to analytics or big data systems
3. The modelling will take so long and I need to prove value right now
4. Every platform charges a variable rate for data storage, modelling and then API/Query calls
And (thank goodness) the emergence of voice and face biometrics to remove friction in identification and security when connecting via IVR, app or messenger is calling for on-premise solutions to bolster the intelligence leveraged in big data environments and drive these experiences. If we authenticate biometrically, by nature we can expect truly personalised experiences.