Building COVID-19 dashboards with Tableau at Bank of America

Profile picture for user ddpreez By Derek du Preez November 3, 2020
Summary:
Bank of America needed new insights into customers requesting mortgage holidays during the height of the COVID-19 crisis. It used Tableau to present the data.

Image of Bank of America logo
(Image sourced via Bank of America )

When COVID-19 hit the United States in full force earlier this year, one of the primary concerns for citizens during the national lockdown was - and continues to be - financial uncertainty. Responsible companies across the US soon realised that customers needed some leniency when it came to financial and contractual obligations, as people navigated the initial stages of the crisis. 

Bank of America, with its 66 million customers, soon found itself in the position, like many other banks, of needing to offer loan payment holidays and deferrals for homeowners on their mortgage. However, the speed at which the crisis escalated meant that the Bank needed to implement new processes rapidly to deal with customer enquiries and also make use of trusted sources of data to visualise and map how COVID-19 was impacting its customer base. 

These data visualisations needed to be made available to executive teams quickly so that they could have a real-time and easy to understand way of understanding the escalating situation. That's no small task for an organisation of Bank of America's size, with multiple teams making use of multiple sources of data. 

Brian Barnes, Senior VP of Mortgage Analytics and Reporting at Bank of America, was speaking recently at Tableau's virtual conference, where he explained how the bank used the Salesforce-owned BI tool to build COVID-19 dashboards for the leadership and how he managed the changing processes in the early days of lockdown. 

Barnes said that the scale of the challenge quickly presented itself, where his team were tasked with managing the data relating to mortgage payment holidays, with very short timeframes. He said: 

Going into the events that happened this year with COVID-19, it was a Thursday, March 17th, when Bank of America announced that they were making a programme available to customers who wanted to get relief on their mortgage loan payments. By Monday we were getting 5,000 requests a day from customers who were wanting to take advantage of this particular programme. 

So my boss called me on that day and said we are getting operational processes being stood up over night, we've got data coming at us from all different directions and we need an end-to-end dashboard that really provides insights and executive views into the volume of requests we are seeing. He also said, by the way, he needed the first dashboard in a week. 

One week isn’t a whole lot of time

As noted above, sourcing data in an organisation such as Bank of America and building out something usable in such a short period of time is no easy task. Barnes said that the first week was "crazy" and he and his team were working non-stop, with very long days. 

His initial priority, Barnes said, was getting the correct data together and making it workable. He explained: 

My first priority, frankly, was the data. Everything starts with the data. I had to find what the right source of the data was, there are a lot of copies flying around and you don't want to grab something that's got transforms in it. Then I needed a place to work with it, I needed a sandbox, so I can take that data for use in the dashboards. By day three I was able to get through that. Immediately after I started building out our base table. Our base table is what's going to be used for not just the Tableau reporting, but down the road will be what everybody uses for their ad-hoc research as well. They want to insure they tie out into the Tableau dashboards, so that they themselves become a source of truth. 

Whilst working with and using data is a familiar process for Barnes, he added that using Tableau for visualisations was new to Bank of America. The team had never designed a dashboard in Tableau before. In order to make sure that he had the right approach, Barnes decided to start with a simple use case in order to ensure the tool's effectiveness. He explained: 

I wanted to really visualise that data so that the insights are coming out of the page to the customer. So I spent a lot of time in the beginning trying to figure out how to get the design right. To make it a little bit easier on myself, I chose a dashboard to go first where the data was already solid. Something easy, so to speak. 

So I grabbed the publicly available COVID-19 case data, I overlaid it with our customer data, who was asking for assistance, and I could build a really nice dashboard, and work with the visual analytics concepts within that dashboard. A side benefit is that it helped me establish the process of building the first dashboards, getting the data to them, getting them published so that users could access them. That was our first week, just getting that accomplished. 

Having trust in your data

The dashboards that Barnes and his team have built for Bank of America with Tableau in recent months have helped the operational teams managing the customer requests, he said. Barnes explained that these operational teams now have clear insight into how many people are in their queue, what plans need to be set up for them, making sure that everything is being carried out correctly and no customer is left behind. 

However, Barnes was aware that for the project to be a success, the people using them need to have trust in the data and what they are being shown. In addition to this, given Tableau is new to the Bank, consideration needs to be applied to the users that perhaps are more familiar with working with other tools. Barnes said: 

Not only was it a big part of the first dashboards that we built, but it's a constant theme. Trust in your reporting is given at the beginning. If you break that trust, people will look at your reports differently. More questions will come up. It's critical throughout the process that we are really thinking about how the quality of data is managed prior to being visualised. If there's a problem, I want to catch it before it gets into the reports. 

That helps build confidence in the reporting. If you think about what people are used to, they're getting Excel files to do their jobs. It would be easy to drop a grid into Tableau, we don't want to do that. We thought about blending the old with the new. We could put a grid in Tableau, but we surround it with additional visuals that provide extra insights that are not easy to extract from a grid. By marrying the old plus the new, it eases the transition for people that are working with BI for the very first time. 

Coupled with this approach, Barnes also explained that Bank of America - particularly given that it's a financial services organisation - has a very structured and strict approach to data management. This helps in the long run, as there is authority with the data being used. He added:

Every year we've seen growth in how [the bank] manages and controls the data to ensure that it's protected and used correctly. Obviously there are efforts that are ever increasing around enterprise data management and governance. We have authorised data sources, where you should be able to trace everything you're doing back to an authorised data source. If you're not that needs to be corrected. So we have oversight on all those things. 

We are a bank, we are heavily regulated, so not only do we have to ensure that these things are in place, we have to have it all documented with evidence. So we are constantly going through all of our processes and the data that we are using to ensure that we can trace it back to the authorised data source. But coming from an authorised data source, there are additional benefits to that. They are putting in the data quality controls there, so that when you use it you know it's trusted. 

Then your customers are getting those metrics and can say ‘yes, this is coming from a trusted source'. It's a big bank, lots of different groups need this information, so they need to be in sync with each other. So having that structure in place helps ensure people are getting the same numbers, the same reporting across the enterprise.