BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Narrative Science Dynamically Automates Summaries Of Financial Information

Following
This article is more than 7 years old.

Sorry Tableau, but sometimes visualization is not enough.

Narrative Science brings computer-generated text to financial services in areas such as institutional portfolio commentary — automating the preparation of quarterly performance and attribution summaries.

[Quarterly reporting] “It’s a very manual process and takes a few weeks and a fair number of people for a family of funds,” said Stuart Frankel CEO at Narrative Science. “We have built a product within Quill and now we can do it in a few seconds.” The result is faster speed to market and a reduction in errors and costs, he added.

Nine of the top 25 asset managers use Narrative Science’s Quill, the company says.

“Once we start pointing the technology to a very specific problem we are able to show a great deal of value and address a fairly significant industry activity,” he added. “Quill, when it understands who the audience is and the type of report to write, understands what data and analytics it will need to generate that type of document. In a sense we are taking data and using it as the source material to automatically generate a narrative.”

The company helps businesses which have more data than they can use when they are limited to manual processes, i.e. people.

“We found an unlimited need to help organizations take their data, figure out what is interesting and important, and then communicate that. At a high level, that is what we do. These organizations are sitting on a heck of a lot more data than they have ever had, and it is growing exponentially every day and they don’t have the right tools."

Tableau, SAP, IBM, SAS and many others have tried to deal with lots of data with visualization.

“All of these platforms have a reporting interface that is essentially pictures,” Frankel explained, “graphs and charts and other types of visuals. It turns out that while those are helpful compared to a table of numbers, they are still difficult for an individual user to understand.”

When a report arrives with a chart on top, that visual doesn’t necessarily make a decision easier. Narrative Science has integrated its Quill with some of those analytical tools.

“Visualization is nice but it doesn't necessarily tell you the story, but we do and Quill does. Words aren’t going away — often a screen is read by an employee who has to explain it to others in the organization. We can help an individual employee to communicate to lots of different people.”

The company works closely with Qlik which manages large amount of data to derive business intelligence, often through visualization but increasingly also through Narrative Science which has developed an extension for Qlik.

“The extension itself worked flawlessly with Qlik Sense, and I now use it in nearly every application that I build,” said Paul Van Siclen, who leads Qlik’s financial services group. 
“My favorite application has to be the one focused on Pricing Analytics. I took publicly available data from the Lending Club, which is the world's largest online marketplace connecting borrowers and investors, and imported it into Qlik Sense.”

Qlik Sense looks at KPIs such as the volume of loans funded, the weight average interest rate, and the number of loans per credit grade. Banks want to understand the relationship between risk, price and the customers’ price elasticity, Van Siclen said.

“In a matter of minutes, I was able to analyze 800,000 records and uncover insights such as how the Lending Club lowers rates, by how much, and the related jump in new business volume.”

As he moves a slider bar, the numbers change, the chart changes, and the explanatory text changes.

“People would challenge a number and I can click on it and the narration comes up — I have the answer while I am doing the presentation.”

Lending Club showed its highest growth of high quality loans in Tennessee.

“I can select Tennessee, hover over it, build the chart and Narrative Science provides the text.”

He thinks text can replace the pivot table.

“I can point to a bar and get words that are actionable. Before, I had to find the variance in the pivot table and create that action.”

Modern business intelligence companies have made it easier to set up dashboards and visualizations, said Frankel, but as they get pushed deeper into the business units and beyond specialized business intelligence units with skilled analysts, companies have less and less certainty over the level of understanding users will have.

“We’re making it easier for the end user to understand what is going on in these visualizations.”

Understanding risk through data

“In risk management," said Craig Muraskin, managing director for innovation at Deloitte,“the biggest challenge for risk is understanding and adjudicating what I need to pay attention to — whether I am an internal auditor or a chief risk officer.”

They are trying to understand what could cause harm, and they have to deal with massive amounts of information and data.

“Making sense of all that has required a lot of human intuition. You go in with hypotheses of what to look for, walk the halls, ask people with big data. There are lots of opportunities to understand where anomalies are occurring and what they might mean, but pulling it all together requires  a lot of effort.”

Narrative Science with natural language generation (NLG) can play a big role as someone responsible for risk looks across the structured and unstructured data.

“How do I at the end of the day say here is what is going on? A massive among of time and effort has been spent in gathering data and no human can look at all of it. Narrative Science offers us opportunities to more efficiently sift through large amounts of data and bring out insights more quickly. There are limitations to what humans can do, what they can find. What we find as the opportunity here and is why we are so keen on the technologies, is that we think there is great opportunity to unearth levels of insight not previously possible when we are dealing with enormous volumes of data. Narrative Science lets you get to a much more granular level.”

Most reports are based on some set of processes that have been in place for year, Muraskin added. With NLG users can ask more questions; instead of looking at 7 independent variables a user can look at 150.

“What does that tell us? Maybe it’s overkill, but maybe we get insights. We’re just scratching the surface at the moment. Leveraging NLG allows us to operate at a scale of expertise we have not been able to before.”

Follow me on Twitter or LinkedIn