How data analytics is helping Bankwest gauge customer feedback
- 04 September, 2014 12:55
Bankwest’s customer and marketing insights team is tapping into an Australian-grown data analytics platform to better track and grasp customer feedback.
Bankwest executive manager of analytical marketing, Simon Edwards, told CMO the decision to seek out a new text mining solution followed the bank’s move to an email-based customer feedback process last year. Previously, customers had been surveyed predominantly over the phone and their responses were classified by staff using up to 40 different overarching feedback categories.
Bankwest receives between 30,000 and 35,000 pieces of feedback annually from its business and retail bank customers. While individual organisation units look at open-ended feedback, head office needs a way to also understand customer trends and feedback across the business.
“With a lot of automated customer feedback platforms, there can be so much information it’s not viable to have real people classifying responses by hand both in terms of volume and cost,” Edwards said.
“But a lot of the key things people are saying are in the way they express it."
In addition, surveys are increasingly adopting more open-ended questions, raising fresh challenges around how to intelligently mine those customer insights.
“Everyone is building capabilities measuring feedback across multiple channels. The next phase is driving insights into what customers are saying with any real granularity,” Edwards said.
Edwards began the search for a solution that could help Bankwest deal with the customer feedback, and looked at both free, open source unstructured data mining options such as R, as well as the capabilities available in its existing marketing and data technology stack, before deciding on an Oracle-based solution from Sydney consumer insights company, Edentify.
The Edentify Web-based tool is based on Oracle’s Endeca Information Discovery and collects unstructured data for businesses from market research, as well as interprets information such as customer feedback and Nielsen insights. It then provides reporting and ongoing tracking capabilities.
At Bankwest, the new solution is collating in-store, call centre and online feedback. Through a three-month process of test and learn using about 2000 rules for classifying customer text data, Edwards said the solution is now as good as having people physically code feedback coming in. Data is anonymised prior to being classified.
Edwards said the bank already had a wealth of knowledge about how to identify the positive and negative things customers reported back, falling into three key elements: Process, product and people’s experiences. Feedback itself is then defined using 30 to 40 categories underneath these umbrella areas. These categories are the same with its new automated platform.
A major selling point for Edentify/Endeca is that it doesn’t require any integration with existing platforms. The platform is a small operational investment in terms of Bankwest’s overall data technology expenditure.
Its Web-based interface is also easier for those who aren’t data scientists to use, Edwards said.
“There is free text mining software available on the Web but you have to be a high-level programmer to use it,” he said. On the other end of the spectrum, existing enterprise technology platforms offer data mining solutions but many of these are very expensive and again, require significant data skills to manipulate.
“This solution has a nice GUI [graphic user interface] that a junior research can use, which was a big attraction for us,” Edwards added.
To highlight the success of the platform to date, Edwards said the bank’s productivity and process improvement teams now have access to a vital customer-oriented data set that reflects how critical processes in terms of end-user experience are faring.
“In a lot of organisations and from a process point of view, there hasn’t been a lot of linkage made between the process re-engineering going on, and the intelligence coming in about the processes,” he said. These teams have traditionally looked at cycle times, error rates and consistency, which is all fine, but the next step is to put the customer view on how good these processes are in a quantitative manner.
“Unless you do the linking for them, those groups are without that [customer] data.”
The Edentify/Endeca solution also offers clear process efficiency improvements for the insights and customer teams, Edward said.
The customer data analytics project was driven by the insights and customer analytics team, which Edwards sits in, and is aligned with Bankwest’s marketing function.
Edentify director, Dan Banyard, said the Endeca solution is about helping organisations make sense of research information and unstructured data.
“We can say [to Bankwest] `When a customer gives you feedback, is it about your products? Is it about staff? Being able to analyse the written words from customers allows you to see why your customer satisfaction scores are either going up or down,” he said.
Yalumba uses Edentify to build brand awareness
Another one of Edentify’s customers is Australian wine producer, Yalumba Wines, which Banyard said is using the data analysis service to get an overall view of how it is doing in the Australian market. Edentify has been doing brand tracking for the winery for 18 months to see if consumers are aware of the winery and, more importantly, buying its wines.
“We have integrated all brand tracking data into the [Endeca] dashboard. This has allowed Yalumba Wines to look at how it is tracking over time against its competitors,” Banyard said.
The next stage for Yalumba Wines is to start looking at other performance data such as Nielsen. This will allow it to look at consumer brand awareness and see what’s happening with price points in stores around Australia.
“They have spreadsheets of information from Nielsen and research graphs which is broken into different areas. They don’t see everything together,” Banyard said.
On a product development front, the next priority for Edentify is to extend its solution to social media channels. Banyard said he hopes to launch an offering by December.
“The reason we chose this [Endeca] platform was the natural language processing. It was important to not only understand information but interpret written words. That lends itself to start looking at forums and social media and summarise the key themes of what people are writing about,” he said.
“Ultimately we want to provide that service to our clients so they can start to analyse [social media] information as well.”
More on data analytics
- Big data analytics: The new black magic of marketing?
- 8 ways to get on top of data analytics
- Quantium data helps Woolworths get closer to customers
- 8 things we learnt about big data
- 3 brands on how theyre dleivering better customer ROI from data