Building a business around cognitive analytics
True to its nature as an agile startup, Goodbyehello was keen to avoid investing in expensive IT infrastructure. It wanted a cloud-based solution that its consultants and analysts could use collaboratively anytime, anywhere – whether they were working in the office or on-site with a client.
Nick Rood comments: “Equally important, we wanted a solution that would produce results we could share directly with our clients. We have years of experience working with traditional statistics packages that produce output that only statisticians can understand. We wanted something much more immediate – because the cleverest analysis in the world is no use if you can’t explain to your client what it means and why you are recommending a particular course of action.”
These requirements led Goodbyehello straight to IBM® Watson Analytics™ – a cloud-based cognitive analytics platform that combines powerful predictive models and algorithms with automatically generated dashboards that can instantly present findings in a wide range of intuitive graphical formats.
To identify the root causes of customer attrition, Goodbyehello leverages both the “Explore” and “Predict” functions of Watson Analytics.
Nick Rood says: “Explore instantly provides suggestions that highlight possibly important patterns in new datasets – for example, it might recommend looking at the relationship between purchases and seasons, or holiday bookings over multiple years. If you think a related topic might be interesting, you can click on it to adjust the variables, and re-analyze the data instantly. It provides a very good quick way to find some interesting areas for further analysis.
“Meanwhile, Predict shows us all of the relationships between variables in the data. There might be dozens at first, but you can easily narrow them down to the most significant ones. The ability to let the data speak for itself, rather than just checking whether it conforms to your initial assumptions, is very valuable. Often it brings up relationships that might not be immediately obvious: for example, we’ve seen that people are more satisfied with their holidays if they have a very clear memory of them.”
The solution’s visualization and dashboarding functions are also important to Goodbyehello, as René Vetter explains.
“Comparing the visualization capabilities of other packages to Watson Analytics is like night and day. In previous jobs, we used to spend hours exporting cross-tabulations from our statistics system, manipulating them in a spreadsheet to create some charts, and then pasting those charts into slides that we could present to clients.
“By contrast, Watson Analytics automatically suggests a set of visualizations that it thinks might be appropriate, and allows you to select and combine them into a dashboard within a few clicks. You can even share the results directly with a client within the platform, complete with interactive filters that they can use to slice and dice the data for themselves.”
In particular, Goodbyehello has found that word clouds are a powerful way to communicate its findings to clients.
“We often ask inactive customers open-ended survey questions about why they stopped engaging with a client, and word clouds are an excellent way to show the most important themes across a group of respondents,” says Nick Rood.
“For example, for one retailer, the decision to stop selling a particular range of products had an impact on customers who were specifically loyal to that brand. And for a tourism company, we found that the most common reason for attrition was a change in family circumstances – for example, their kids had grown up and no longer wanted to go on a big family holiday every summer.”
Driving significant increases in re-engagement
By understanding why individual customers lose interest in clients’ campaigns and product offers, Goodbyehello is in a strong position to design more personalized marketing activities that drive a much larger percentage of lost customers to re-engage.
Goodbyehello recommends a three-stage process, known as the “ping, play, purchase” cycle. The ping stage checks whether the connection still exists – for example, does the person’s email account still exist? Next, the play stage involves designing a creative, entertaining and personalized interaction that will help to reawaken their interest in the client.
Bart Willems gives an example: “Perhaps they have some unused points on their loyalty card, so we ask whether they would like to keep the points or donate them to charity. The key point is that this first contact should not be a commercial offer. We’re inspiring them to think about our client again, we’re not trying to sell to them immediately.”
If the customer reacts positively to these playful approaches, the final step is to send a carefully selected product offer, encouraging them to make a purchase.
“Combining Watson Analytics with our methodology has delivered a 300 percent increase in re-engagement for one of our biggest clients – a major Dutch retailer,” says Nick Rood. “We’re very confident that the 300 percent figure is robust – it is based on more than six months of data. When we have enough data to run a similar analysis on our other clients, we will do so – part of the value of our service is that we can always quantify the benefits it brings.”
The platform’s predictive modeling capabilities are also valuable, as René Vetter explains: “Although it is too late to prevent the customers we’re looking at from becoming inactive, we can create models that allow us to assess the similarities between active and inactive customers, and help our clients predict which of their currently active customers are at risk of becoming inactive.
“For example, one key finding is that people who have purchased a product or used a service more than once are very significantly less likely to churn. So we recommend that our clients should target turning new customers into repeat customers, as the best pre-emptive measure to avoid future attrition.”
He concludes: “Our business model could not exist without IBM Watson Analytics – it gives us the insight we need to design truly personalized campaigns that really speak to inactive customers, and give them the encouragement they need to re-engage. The powerful analysis capabilities and instant visualizations help us show our clients not only how to win back lost customers, but also how to prevent further attrition in the future.”