Negotiating the fine line between hyper-personalised and hyper-personal23 Jan 2018
Set to take the stage at Big Data World London this March, Kirk Borne, Principal Data Scientist and Executive Advisor at Booz Allen Hamilton, speaks to The Stack on using data in an experimental way to learn about your business and customers
Hyper-personalisation and Journey Science are recent phenomena that apply primarily in the context of customer marketing and experience. While relatively new business strategies for customer insight, their potential impact is already spanning much wider than the realm of marketing.
Data and analytics can be used to track all different types of user journey, from hospital patients looking for a more personalised treatment to an employee searching for precise recommendations on training and development.
For years we have been used to seeing recommender engines, with retailers recommending all sorts of products whenever you are online shopping. As an extension to this, people are now expecting something a little more personalised than just a generic recommendation.
For instance, hyper-personalisation would not just include suggestions for items that you may like or have seen before but would show results that are context-driven and rely on alternate factors such as time of day, the day of the week, IP address or location.
As the availability of data increases, businesses are more likely to be motivated to experiment with these hyper-personalised strategies. It is critical that this approach is not just a one-off activity, but that it follows an entire user journey. Understanding that journey is an important part of ensuring that the personalisation is unique and specific to the user.Opportunities for business
Hyper-personalisation has enormous potential to benefit businesses, not only from the point of view of building awareness and attracting more customers but also retaining those customers. When users are exposed to this type of personal treatment they are increasingly likely to come back to use the service more – they become loyal customers, or even advocates (the highest level of achievement).
As an internal HR tool, Journey Science can also offer insights and opportunities to support businesses with employee retention. This topic is a very big deal at the moment, certainly in the tech world where it is difficult to keep really talented people with the vast amount of offerings available. This is even more exaggerated in the data science and analytics fields.
Journey Science is one of the ways we can understand more about our employees, their interest in certain projects, team dynamics and preferences for training and professional development.
Online discussion boards are a good place to start with businesses able to conduct sentiment analysis to gauge job satisfaction and work with the analysis results to provide solutions.
Looking at logistics and manufacturing, Journey Science can further be applied to non-human physical objects. Here we are no longer thinking about a person, but about a product or a machine.
Businesses can track the performance of their assets, whether a wind farm, a submarine or an aeroplane, using a computer-simulated model or a ‘digital twin’. Machines and equipment have lots of moving components, they require maintenance at different times, and parts fail. Simulating all of these different scenarios in a digital twin model can help predict what maintenance will be needed and when something might fail.
Conversely, if there is a failure in the real-life system the live stream of sensor data can be fed into the computer-simulated version. Filtering this real data through the simulation, engineers can see how the simulation reacts, giving them a much more informed idea of how to fix the real system.
Where to draw the line?
When considering the human user journey, there is a fine line that has to be negotiated between ‘personalised’ and ‘personal’. The aim is to be hyper-personalised, and not hyper-personal. Trying to figure out the difference is tricky.
Take for example Target, which looks at its loyalty card data to tailor its customer experience. A few years ago there was a case where one particular person was buying products that indicated pregnancy so the retailer started sending advertising and coupons to the family home for related products.
The loyalty card owner, it turned out, was a father whose teenage daughter was using his loyalty card to purchase maternity products at the store and had not yet told her father that she was pregnant.
There was plenty of discussion at the time around whether this had crossed the line of being too personal.
Businesses can promote exactly what the data suggests as a recommended product or service, but we have to do better at understanding what information customers would rather not disclose, particularly with the forthcoming GDPR.
If it was easy to find the line, we wouldn’t cross it, but the problem is that it is confusing and sometimes you cross the line before realising it is too late.
These are all situations that we need to be wary of. A company not only risks losing a customer in these cases but as soon as it reaches the news it can cause deep reputational damage.Finding the best approach
Many businesses will assume they know where to draw the line and act accordingly. They will probably get that wrong because it is a complex problem to do with human emotion and behaviour.
Others may be extremely safe and not even come close to the line, sticking to very simple recommendations and simple customer interactions. The risk here is that businesses can miss out on the benefits that all of these advances can bring. In the modern digitally-disrupted world playing it safe also means the end of your business.
The final option is to carry out more experimentation and A/B testing to probe the boundaries of whether your activity is too personal or not.
This brings to mind a scene in Jurassic Park where a dinosaur is probing an electric fence. When the power is turned off it realises it can break through the barrier and go further than it had ever gone before.
In this way, companies can run trials with different recommendations and different experiences for users. If there are positive responses and increased customer loyalty, it is a move in the right direction. However, signs of negative feedback suggest that the activity is too close to the line that shouldn’t be crossed.
To develop a culture of experimentation also helps to avoid reliance on HiPPO (Highest Paid Person’s Opinion) decision-making. Too often decisions are made by the HiPPO, but this does not mean that it is the right decision – now we have the data to back up any decision process.
Businesses need to be more proactive with hyper-personalisation strategies, carrying out tests to find out what users respond to and what they find most engaging. This is the sweet spot – not way back in the safe zone where there is no evidence of advancements, nor without thought of consequence of the risks and dangers.