Insight

Are You Getting the Signal?

How to leverage data to predict customer needs

This article is part 2 of a 2 part series. Read the rest of the articles here.

When it comes to Anticipatory Banking or other AI and machine learning approaches to better predict and deliver on your customer needs, the ability to capture signals is critical. For example, if a bank wants to identify customers who need financial guidance, it needs to look for signals like requests for credit limit increases, higher credit card transactions and increased activity in a particular category. If that category is health, the bank might also confirm an increased number of visits to doctors offices, etc. Together, these signals tell the bank their customer is facing a medical crisis and will likely need additional funds to pay for treatment, allowing it to present offers or products that can help the customer meet health-related financial commitments. To capture these signals, however, you need the right foundational infrastructure, as well as technologies to transform data into insights.

Signals versus noise

Small sets of data that can have an impact on the business are called signals, while irrelevant data are considered noise. However, a piece of data can be both; it simply depends on what you are trying to accomplish. For example, a customer looking to buy a house can be identified proactively by her traffic to home buying sites, increased liquidity in her checking account, her location graph showing increased visits in the neighborhood and her searches for mortgage. But for a bank that does not sell mortgage products, this information can be irrelevant. The key is being able to capture the signals amidst all the noise.

Capturing signals starts with data

Spotting signals is becoming more difficult as the sheer volume of data increases. Today’s digital business environment as well as the growth of Internet of Things (IoT) have contributed to the dizzying pace of data generation. IDC predicts that the collective sum of the world’s data will grow from 33 zettabytes this year to a 175 zettabytes by 2025, for a compound annual growth rate of 61 percent.¹ As the amount of data companies own—and have access to—continues to grow, being able to isolate signals becomes even more challenging.

Three types of data that firms must use to generate signals include:

First-party data

Information an organization collects as part of its marketing, sales, service and operations. Examples include account information, transactions, digital interactions, IOT logs, etc.

Second-party data

Information your partners have collected and are willing to share with you. For example, media partners might share ad impressions and audience insights, while loyalty partners will share information about purchases and digital interactions.

Third-party data

Purchased data from data aggregators. Data aggregators collect data from publishers and other data holders and enhance it with their insights. Examples include demographic data, location data, retail data, etc.

All companies have a rich set of first-party data, which provides critical information about a customer’s past interactions with a company. But that alone doesn’t provide enough context to truly understand where customers are on their journey today and where they are going. First-party data are often combined with second- and third-party data to fill in gaps and provide a more complete picture. To identify a prospect for mortgage products, for example, you need to augment first-party data, like core customer information, credit worthiness, increased liquidity, etc., with second and third-party information, such as digital interactions in real estate sites, searches for mortgage rates and activity in mortgage comparison sites. Together, this information will not only tell you if a customer is in need of a mortgage but also at what stage he or she is in the home buying process, allowing you to truly personalize your offer and make it both relevant and timely.

The role of the data lake

To extract business critical information, data need to be organized and processed. But many of the data lakes in today’s banks are siloed, disorganized and difficult to use. In other words, they aren’t optimized to mine the information needed to access signals and their related data. Organizations have to re-examine their data lakes to see if they are robust enough for cross domain/department usage. They have to integrate data, curate it and generate signals so that minimal transformation is required by data scientists, analysts, engineers and other data users. Data lake teams should also look to continuously enhance data lakes with new sources and infuse real-time data—both requirements for using this information to optimize the customer experience. Ultimately, all available data within the organization need to be linked and curated in order to paint a comprehensive picture of the customer.

From data to signals to insights

Being able to recognize and capture signals from vast stores of data is a critical capability for digitally-minded organizations. However, it’s important to go a step further and transform signals into insights that can add tangible value to your customer by promoting products to the most receptive customers, which in turn increases purchase rates and drives revenue.

Learn more about how to compile and clean up your data so you can transform signals into insights in our next piece around anticipatory banking.

Sources:

1. Network World. IDC: Expect 175 zettabytes of data worldwide by 2025.

Read the rest of the series:

{{entry.eyebrow}}

{{CTALinkLabel}}

{{ entry.promoAsset.promoCTALabel }}

Insight

Are You Getting the Signal?

Signals versus noise

Capturing signals starts with data