TD Bank Group: Turning Big Data Into Valuable Data [#CZLCHI]


Yuyu Chen

In his presentation at ClickZ Live Chicago, Garry Przyklenk, manager of analytics implementation at TD Bank Group, shared three ways marketers can glean valuable data from a big data reservoir.


There’s no denying that big data is important. But it’s the quality of the data rather than the quantity of data that matters when it comes to making business decisions. So how can marketers ensure they’re working with the best data?

In his presentation, “Valuable Data Not Big Data: How to Truly Boost Your ROI,” Garry Przyklenk, manager of analytics implementation at TD Bank Group, explained that marketers need to find valuable data in order to increase their return on investment (ROI).

“You don’t need a lot of data to make decisions. You just need the right data,” he said.

Before marketers are able to decide which data is valuable, they need to understand the very nature of big data. To explain, Przyklenk used the “four V’s”: volume, velocity, veracity, and variety.

According to him, marketers today are able to have a large volume of data at their disposal, much more than traditional systems can handle. But when they try to figure out the theory behind the numbers, flat data or static data without “velocity” is not really useful.

“When the analyst on our team looks at the flat data, he says ‘I cannot make a decision out of this, I have to see it moving,'” noted Przyklenk. “We need the velocity to see the trends, to see what happens.”

And some data just lacks veracity, he continued, “Something I want to make clear is, especially in the digital marketing space, we are dealing with lots of data that is very, very dirty and inaccurate.”

Also, according to Przyklenk, it’s challenging for marketers to analyze a variety of big data, both structured and unstructured, in the same ecosystem.

Przyklenk explained that there are often “lofty promises and blatant lies” in big data. Big data doesn’t necessarily mean more accurate data, he said, because proliferation of unstructured data actually decreases accuracy. And big data may not improve marketing effectiveness like it is supposed to, because significant work is required to integrate disparate third-party systems. Finally, big data doesn’t guarantee increased revenue and decreased costs, because all technological capabilities come at an opportunity cost that may or may not be quantifiable.

So how are marketers supposed to obtain valuable data from a big data reservoir? Przyklenk shared three ways to do so: a centralized data warehouse, a data management platform (DMP), and customer journey mapping.

Centralized Data Warehouse

In order to deliver value, according to Przyklenk, marketers need to integrate data into one platform. For example, marketers may have different reports based on Foodlight tags, Google Webmasters Tools, or other sources, so they need to consolidate these marketing reports in one place.

In this regard, examples of valuable data, according to Przyklenk, include:

  • Daily granularity keyword data from paid and organic search platforms
  • Paid search cost-per-click data and campaign metadata
  • Email and direct mail campaign metadata and tracking code usage
  • Micro and macro conversion data
  • Transactional details including products, promotions, and discounts


Employing an intermediary DMP is preferred, said Przyklenk. He explained that marketers give their consumers offers based on their behavior data, but offers may not be valid across channels (retail, phone, online, mobile, etc.). In addition, marketers may just have minimal personalization or have no modeling capabilities.

On this front, valuable data marketers can use include:

  • Primary sales channel transactional logs, such as online, mobile, and retail
  • Customer relationship management (CRM) metadata like age range, demographic, and product purchase history
  • Advertising campaign creative and classification metadata

Customer Journey Mapping

According to Przyklenk, many organizations today don’t have a 360-degree view of the customer journey. “Robust customer experience data is available, but siloed in channels,” he noted.

To achieve omni-channel integration and cross-channel optimization, Przyklenk said that marketers can use the following data sources:

  • Raw Web analytics “clickstream” data
  • Phone channel or interactive voice response (IVR) data
  • Retail transactional(purchase and returns)data

In addition, marketers can also use social media interaction and sentiment data if they have, Przyklenk added.