Data analytics and customer intimacy in Financial services
Unlocking big data to disrupt an industry
The wealth management industry stands at a moment of reckoning as a generation shift occurs with Millennials taking the reins as the client group with the largest number of “relationships in motion”. As a generation that has witnessed the effects of 2007–2008 financial crisis, Millennials have demonstrated a preference in direct-to-retail financial services (like Robinhood, MooMoo and SoFi) driven largely due to better digital offerings. A UK study suggested that over 85% of millennials are comfortable with robot financial advisors, a trend which will only increase with future generations. As financial firms rush to attract future generations with features like gamification, social sharing, alerts and mobile accessibility, these firms are facing rising resistance from regulators.
Within this changing market, disruptive fintech firms are leveraging the power of big data to perform real-time analytics and predict customer behavior. With 90% of historical data having been generated within the past 2 years, technology firms offering financial services are able to implement sophisticated strategies to create personalized customer experiences, and have gained popularity among a new generation of investors. For example, a key feature separating Robinhood from traditional brokerage houses is the emphasis of data analysis on strategizing user experience. The Robinhood data lake is built on influences from user-focused platforms such as Netflix and Uber, as also traditional financial institutions like FICO and National Bank of Canada (Satia, 2019). Analyzing Robinhood user’s purchase patterns, from previously open-access data, proves to be a thoughtful analysis of the impact of platform design, social network effects and data analysis on modern financial applications.
The 2021 market manipulation of GameStop stock by users of discussion forum platform — Reddit resulted in losses for both hedge funds and individual users. Major backlash was faced by the equity trading app Robinhood for barring customers from executing trades in the stock during price surges and resulted in huge losses to retail investors on the app. The event has caused a debate among securities regulators about the impact of apps which have encouraged millions of inexperienced users to enter the stock markets.
50% of Robinhood users are novices without any trading experience prior to joining the platform. As of May 2020, Robinhood boasted a higher number of users than traditional e-brokerage firms like Charles Schwab and E-trade. A primary reason of the increased use of the app by first-time investors are features such as offering a ‘free stock’ to new users or allowing current users to refer friends to the app to gain free stock. The free stock is disclosed to users by scratching off an image of what looks like a golden ticket on the app.
Gamification features such as an explosion of confetti on your screen on purchasing new stock, and steady notifications prompting users to trade on the site, create higher activity among users than traditional apps. A New York times article suggests that Robinhood users are far more active with users trading 9 times the share volume compared to E-trade customers and 40 times the shares traded by Charles Schwab users (Technology, 2020). It is also found that users are far more likely to display ‘herding’ behavior and concentrate 35% of buying activity in 10 stocks as compared to 24% by an average retail investor (Brad M. Barber, 2021).
The stocks with the highest trades on the platform are those of the ‘Top Mover’ list which are prominently displayed on the home page of the platform and lists 20 stocks, and changes daily. Many users are not strictly influenced by the app itself, but also tend to focus on a particular subset of stocks while sharing information on forums like Reddit’s WallStreetBets (Zweig, 2021). Increased investment activity on stocks which gained media coverage, and stocks with consistently high trading volumes were other notable behaviors of Robinhood’s social and media-influenced users.
Diving into the Data Lake
With firms utilizing cookies to personalize user interactions, new products are becoming more customer focused than ever before. Big data is a field of data science that aims to derive insights from large and complex data sets.
The 5 V’s of data help understand the business strategy behind big data analytics:
- Volume
In 2020, the world’s total amount of data is around 40,000 exabytes, up from 130 exabytes in 2005. Today firms are witnessing extremely vast pools of customer data, which are invaluable to formulating business strategy.
2. Velocity
With faster internet speed’s the flow of data is witnessing a continuous and massive upswing. Businesses are focused on reacting to data in real-time to enable quicker reaction to customer cues and drive positive business outcomes.
3. Variety
Data is received in 3 basic formats:
a. Structured — The easiest to process and draw insights from. Structured data conforms to defined lengths and input formats and are well-organized for analyses by either humans or machines.
b. Unstructured — Data which does not conform to a defined input format and is chaotic.
c. Semi-structured — Data which falls mid-way between structured and unstructured. They require to be manually processed before machines can analyze the data. This makes is more time-consuming to derive insights from semi-structured data.
4. Veracity
While gathering large amounts of data, it is difficult to assure the quality of the data. During data collection, there is a probability that accumulated data will comprise of a myriad of sources and types. It becomes an important step for firms to incorporate a data validity exercise to ensure accuracy of data insights.
5. Value
As data processing begins, the value of collected data can be measured. Utilizing a custom data processing system, firms can process large volumes of structured data to identity trends and add value to the decision-making process. With cloud computing gaining popularity, the volume and scope of big data analytics will see compounding growth. The ability to utilize remote data servers to store data, coupled with the world’s largest technology companies developing new tools to analyze copious amounts of data, will allow firms to provide clients with better financial products and services to suit changing user patterns.
The ability of emerging fintech’s to apply insights of big data analytics to improve customer interactions, is proving to be formidable competition to traditional players in asset management.
The effects of utilizing big data analytics in fintech operations are visible in areas including:
- Risk assessment
With most millennials used to managing their professional and personal lives through smart phones, using digital form-based data entry allows businesses to improve customer experience. Millennial users are averse to filling lengthy paperwork such as risk-profiling questionnaires, that are required by financial regulators. Using a digital approach to data gathering and risk profiling allows firms to improve both, data accuracy and customer engagement. Using digital data gathering tools, firms can store, and process data is a structured manner and enable quicker business decision-making.
2. Behavior reinforcement
Using customer data to encourage habits has become popular with fintech apps incorporating aspects of gamification based on psychological concepts. Using positive re-enforcement such as positive alerts and notifications of completing tasks, create a sense of accomplishment and increases the probability of customers performing suggested tasks. While gamification can be a force of good in financial apps that are focused on savings, the tendency of such features to elicit addictive behaviors among users are a big cause of concern.
3. Security Improvements
Identifying fraudulent activity within datasets becomes easier with well-structured and sophisticated data analysis systems. Through analyzing datasets, fintech firms can create reliable fraud detection systems and identify irregular transactions. A result of quick identification of lapses in data security and malicious activity is the ability to quickly notify clients about such lapses through digital modes of communication. With a unified and automated customer data system, it is possible to identify affected client accounts and offer customized communications.
4. Customization and improved customer service
Using big data, a firm can create detailed user personas and using clustering, offer tailored services to customer’s based on their perceived requirements. Predictive modelling allows for data to get processed in order to derive insights based on demographic patterns including age, relationship status, location, spending behavior, among others to offer custom solutions and products. Through digitalization, a customer’s activity on a platform can be mapped to identify errors and offer solutions quickly.’
The fundamentals of creating a robust solution for customers across different user groups required a deep understanding of the business and the data. Using appropriate servers and joining data across stores with the right queries, would allow a business to deliver future-resistant, and extendible frameworks to deliver insights and formulate business strategy.
Framework of implementing a Data Lake
While creating a data lake, it is important for data accessibility to be secure and uniform. Allowing storage of different types of data formats, creates complications while finding insights across separate silos of data. It is important to allow such silos to be accessible to deliver high quality insights. To create a seamless interaction between data types a layered framework allows for an optimum data lake solution.
1. Ingesting
In this step, data enters the system in either batch or stream form from internal applications or external feeds. External data entering the data lake is streamed through a databus and is archived in storage layer, while internal data is passed to the storage layer directly.
2. Storing
Storage should be cost effective and reliable with good performance. With large players in cloud storage and computing, it becomes easy for financial technology firms to obtain a high functioning distributed file storage system. Within storage systems, custom datasets can be created to store data, for particular frequent use-cases.
3. Processing
Frequently, incoming data requires to be processed due to size of the raw files, or other factors. In such cases, data is processed using distributed data processing tools and can be easily managed through serverless cluster management products available via cloud service providers, e.g., Microsoft Azure, AWS or GCP. The metastore can be partitioned to process new data in clusters. Certain raw data can be compressed in this stage as well.
4. Querying
After data is processed it becomes available to teams within the firm to query and derive insights. During this stage data stored in the storage layer should also be accessible for creating visual representation of datasets and identifying trends.
5. Validating
Thresholds to ensure the data entering the data lake follows certain required definitions and thresholds, allow for admins to validate incoming data, and avoid data lake failure cases. Thresholds could include constraints on values, category uniqueness, expected number of rows, among others.
6. Inferencing
Using a plethora of available data visualization or business intelligence tools, teams within the firm can utilize the available data and draw out hidden patterns in user activity data. Using this data to create training and testing models allows business teams to implement changes on the platform to improving the end-customers experience.
Case Study : Robinhood’s data platform architecture
Robinhood uses a range of tools and vendors to ensure data incoming into the application is processed to create valuable interactions. The data lake architecture at Robinhood allows over 4 PB of data in the lake, processing over 10 TB of data per day. This data is analyzed by data platform teams within the firm to improve product features on the app and increase customer retention.
A growing number of studies have indicated that Robinhood users demonstrate herding mentality in holding stocks with higher trading activity, or investments made in obscure stocks including cannabis stocks, risky technology firms, etc. It begs the question, with the data architecture at the hands of Robinhood’s platform team, would not adding communication and alerts related to cautious investing be a value-add to the over 6 million novice investors on the platform.
Currently, data analytics tools provide users with access to features including: Personalized user dashboards, featured news articles, targeted alerts and notifications on stock fluctuations, automated communication regarding tax forms, among others. All these features create a seamless user experience and increase customer retention but lack an added value-add to their customer base of investor education.
Robinhood’s mission statement proudly mentions, “democratize finance for all…[and] make investing friendly, approachable, and understandable.” While the application has used technology to provide accessibility and promotes investing, access to information and disclosures does not equate to good outcomes for their user base. We can surmise that how data is made available affects investors and influences decisions in ways which are capable of both helping and hurting users.
CONCLUSION
“(Gamification) provides a way for individuals to rewire their brains and bodies and achieve better investment behavior against the imprints generated by financial events and the experience of their formative years.”
- Paolo Sironi, fintech thought leader
When a large portion of incoming millennial wealth is being ushered into asset managers hands, it is important to understand the motivations of these users’ investment decisions. Scared after witnessing financial crisis and bankruptcies or large institutional financial entities, these incoming investors are cautious about financial services, while remain largely confident in robo-advisors. It is an interesting era for fintech firms, where the use of data can create brand loyalty among this generation of apprehensive investors. Utilizing prompts and tools in a gamified context creates an illusion of free-will, which gains trust and improves user stickiness on such financial applications.
My purpose in this article is to shed light on the practices of emerging fintech firms to utilize data and drive customer retention. Using a case analysis of Robinhood, an application that has gained popularity and criticism in equal measure, we are able to observe the highly layered data lake which drives customer analytics and insights. Although Robinhood is a popular platform, the features are focused on novice investors and remain quite simplistic.
Through a focused case analysis on Robinhood, I aim to demonstrate how data layers used in analyzing customer activity can help the firm provide over 13 million users with access to an easy-to-use platform which simplifies equity and ETF investments. However, in creating a simplified user experience with low friction, Robinhood has shaken the industry and attracts critics about the sheer simplicity of use and gamified features.
While we are witnessing growing popularity of using data analytics tools, automation, and AI within financial services, it is important for users to understand the underlying nature of the business they participate in. At the end of the day, does gamification mean encouraging financial risk-taking, the results of which are seen in real-world outcomes and affects lives of the many users of such fintech applications? If the answer is yes, it would be important for users to take a closer look at their own investing patterns and course correct before the game wins.
Works Cited
Anna Zakrzewski, Joe Carrubba, Neil Pardasani, Hans Montgomery, Craig Prager, Blaine Slack. (2020). At the Crossroads Wealth Management in North America. BCG.
Bird, D. (n.d.). Game on: the potential for gamification in asset management. Willis Towers Watson.
Blog. (2020, May 27). Retrieved from Bornfight: https://www.bornfight.com/blog/understanding-the-5-vs-of-big-data-volume-velocity-variety-veracity-value/
Blog. (2021, January 11). Retrieved from Growth Gorilla: https://www.growthgorilla.co.uk/blog/why-is-data-analytics-in-fintech-so-powerful
Brad M. Barber, X. H. (2021). Attention-Induced Trading and Returns: Evidence from Robinhood Users. SSRN Electronic Journal, 60.
Breaking News. (2021, March 31). Retrieved from Star Advertiser: https://www.staradvertiser.com/2021/03/31/breaking-news/robinhood-cans-the-confetti-unveils-new-celebratory-designs/
DigitalSuccess. (2020, December 31). Retrieved from IndusNet: https://www.indusnet.co.in/gamification-in-fintech/
Gaming. (2020, January 30). Retrieved from Headstuff: https://www.headstuff.org/entertainment/gaming/gamification-of-fintech/
Satia, S. (2019, September 9). Engineering. Retrieved from Robinhood: https://robinhood.engineering/data-lake-at-robinhood-3e9cdf963368
Technology. (2020, July 8). Retrieved from NYTimes.com: https://www.nytimes.com/2020/07/08/technology/robinhood-risky-trading.html
UBS Chief Investment Office Americas. (2017). Millennials — the global guardians of capital. UBS.
Zweig, J. (2021, January 31). The Real Force Driving the GameStop Revolution. Wall Street Journal.
