In the digital age, many brands are awash with data driven by an increasing number of customer interactions from places such as CRM databases, EPOS, customer contact platforms, survey feedback and social media conversations.
However, data and insight is of no value whatsoever unless it is used – and used in the right way. It should be used to improve the customer experience, customer loyalty and ultimately the bottom line. Therein lies the challenge – “I can see there are a few areas we need to improve on, but what should be fixed first and how much should I spend fixing it?”
Getting the basics right
The first consideration, of course, is practicalities. Some things will come to light from data collected that can often be described as ‘quick wins’ and should be done without a second glance. These usually come in the form of clear service failings – things like empty shelves or dirty toilets.
But beyond that, it’s often longer term actions that compete for attention (and budget and resource). In these cases, managers and decision makers need a sense of the resources required to make the necessary change, the likely return on that investment and, perhaps most importantly, the cost to both the business and the customer experience of just sitting back and doing nothing.
Predictive analytics can play a key role in helping to agree priorities, build business cases and set targets. It’s an important part of any business toolkit and sits within a hierarchy of techniques all designed to help aid action planning and decision making.
Key driver analysis is the foundation stone of predictive analytics – it statistically derives the prominent levers of success, typically using regression techniques. Such analysis is fairly common and relatively straightforward to do.
Packages exist to measure the impact, for example, that a friendly member of staff has on the likelihood to recommend a brand and re-purchase. It will also quantify the nature of that impact and how it compares to the impact of other factors or aspects of the customer experience, such as product range.
Once this impact has been determined, it is then plotted against the current brand performance in various areas. In other words, we look at not just areas of the customer journey that matter most to a customer, but areas where a brand is underperforming. Areas that matter most to the customer and where the client is not delivering should then become the urgent areas of focus for development and priority. Predictive analytics provides the headline for the resulting action plan by quantifying the ‘size of the prize’ in terms of the potential return a brand will get if the recommended action is taken.
Show me how
Smart use of predictive analytics involves factoring in other important business variables. For example, brands should take time to look at the volume of customers who are likely to be affected by the change, as well as the type of customers these people are and how much on average they spend (if known).
Predicative simulators can clearly show the likely gain in top level key performance indicators from improving in under-performing areas. This can be helpful in demonstrating why the change is worthwhile, especially to secure stakeholder buy-in. And when predictive analytics can be linked to spend it can be even more powerful and have an even bigger impact on the business.
“If we rebuild the account area and reduce dissatisfaction with it by half, what impact will that have on our KPI?”
For a results focused business, a predictive analytics simulator can be the tool in working through scenarios in finding an optimum combination of improvements to hit targets and deliver a better customer experience or improve overall satisfaction.
Warning! This product may contain nuts
It is important to note, however, that predictive analytics simulators do come with a health warning when used in isolation. What predictive analytics is doing is attempting to predict the future. Be under no illusion that this is, and always will be, an inexact science.
There are ways to limit your exposure to risk. To get the most out of predictive analytics, managers and decision makers should be following these simple steps:
- It’s all in the potential: Think of the uplift shown in the simulator as potential gain rather than a hard forecast. This still helps business planning by giving a clear indication of the size of the prize but pulls back from predicting the future – that would require all other variables, such as competitor activity, to stay static. Which they don’t.
- Stand hand in hand: Work with a partnership or agency who have done it all before. Experience is vital in knowing how to build a robust model, what assumptions to make and how to optimise and interpret the model effectively.
- Get a second opinion. If you are producing your own targets using a self-service tool, run them by your partner before signing off a business case for major change.
- The bigger picture: Use predictive analytics as part of an action planning strategy. Other insight techniques, such as root cause analysis and competitor benchmarking will help to build a bigger and better picture of your customer’s experience within a market context.
- Keep your finger on the pulse: Monitor performance – go back, check customer feedback and see whether the uplift in satisfaction predicted by the simulator has happened. And then use this subsequent insight to fine tune the model for your business.
- Don’t take your eye of the ball: Advanced analytics are great for longer term smart decision making. But don’t forget to get back in the cockpit of live dashboards and keep getting those basics right!