How organizations can best succeed at data 'hunting'
Data commercialization has become increasingly common, and the supply of data products and services will continue to grow. According to Forrester Research data, more than 75 percent of firms have launched initiatives to improve their use of external data or plan to in the next year.
In a new series of research, Forrester’s Principal Analyst Jennifer Belissent argues that it’s not enough to gather data – but that firms need to actively hunt for it. As data is the lifeblood of business insights, companies will use all sorts of data to help them create better differentiation – from news, weather, social, demographics, census data and any other socioeconomic indicators. But as the use of data spreads across organizations, small data teams can’t keep up.
Information Management spoke with Belissent about what he means to be a data hunter, and what the true “big game” is in this effort.
Information Management: In your new series, you argue that organizations need to do more to actively hunt for data, not just gather it as they have in the past. What are some examples of how organizations can hunt for data?
Jennifer Belissent: Organizations must be proactive in their quest for external data. The best way to do that is to assign the role of data hunter or data scout. That person must be able to translate business requirements into data specifications, and sniff out data that meets these needs.
Whether you want to inform a competitive strategy through descriptive analytics, enrich a predictive model, or train a machine learning model, data requests must start with the specific purpose of the data, the time frame, and the frequency needed. Some data teams use a data request form (sample below) to capture these requirements. But data hunters must also keep close eye on the dynamic data market.
Data providers have proliferated in recent years with new originators building data sets from unstructured data, new data brokers and aggregators bringing data to market, and new data marketplaces allowing buyers and sellers to find each other.
One way of keeping up with new sources and getting ideas for where to look for something unique is to participate in industry forums or peer groups. To identify potential sources, the chief analytics officer (CAO) of Enova and his team participate in topic-specific roundtables and industry groups, which sometimes include data providers.
Examples include the Merchant Risk Council, an industry association focused on eCommerce and payment fraud, and Communications Fraud Control Association, an industry association focused on fraud prevention in the telecom industry. The International Institute for Analytics' Analytics Leadership Council brings together client-side analytics leaders in financial services to discuss and share challenges and best practices of their analytics organizations.
IM: How does an organization best know what data to hunt for – what is the real “big game” in data?
Belissent: The real “big game” can be elusive. And, sometimes you don’t know what it is until you’ve had a chance to get your hands dirty with it. That’s why it’s important to leverage try-and-buy programs. But there are a couple of signs that a piece of information could be big.
The first is novelty. Decision-makers are always looking for an inside scoop, a little local knowledge or some piece of information that could shed light on a problem or inform a decision to be made. For example, a higher than average number of cars in a retailer’s parking lot suggests strong sales. Job announcements quickly filled suggests an employer is a good place to work. Many job announcements suggest growth.
The second is exclusivity or close to it. If no one else has a piece of information, it is definitely a differentiator. Being the first to discover correlation between data points can deliver competitive differentiation, or a solid head start. The first to aggregate insurance policies on new cars as a proxy for automotive manufacturers’ sales figures was able to get real-time insights into sales performance rather than wait for the monthly figures to be published.
But it’s not always one size fits all. What is “big game” for one company might not fit the needs of another. Look for the required granularity of the data or the quality of the data. Predictions of propensity to spend and risk of catastrophic failure require different types of data. Data hunters might be looking to enrich an existing data set with new attributes for a given person, place or thing to complete a profile. Or extending a data set with new people, places or things.
IM: For those organizations that are struggling with this, what do most do wrong?
Belissent: It’s less a question of doing it wrong than not doing it at all. Companies that do not coordinate data acquisition end up with multiple licenses to the same or similar data. Many have uncovered egregious sales practices among some data vendors with multiple contracts sold to a single firm: sales, marketing, customer service, and finance have all purchased customer data from the same source.
In the words of the CDO of S&P, a more coordinated approach enables "negotiating from a position of strength." At S&P, a new coordinated process reduced full-time equivalents (FTEs) and data spend, resulting in millions of dollars in cost savings.
IM: How do those laggard organizations best recognize their struggles?
Belissent: As requests for external data increase, data teams feel the pressure. Responding to ad hoc requests don’t scale, and can result in wasted time and budget.
At one large US insurance firm, the two-person market data services team began to feel overwhelmed with the data requests. When requests begin to come from multiple business units or functions (sales, marketing, finance and operations) and spend starts to increase, it’s time to put a more coordinated data acquisition process in place, and assign specialists to help identify new sources.
IM: For those organizations that are leaders, what distinguishes them?
Belissent: Leaders typically have: a formal data acquisition strategy, a coordinated approach across the organization and specialized data hunters dedicated to identifying new and differentiating data sources.
They have built a data sourcing competency with their organizations that includes:
- Data discovery: Finding the right data requires identifying requirements across teams. A cross-organizational lens eliminates potential duplication and creates an economy of scale when negotiating with data providers. A data hunter scouts out new data sources.
- Data acquisition: Governance must be address upfront to ensure new sources meet quality and compliance requirements. A coordinated approach to due diligence spreads that work load across data users. That coordinated approach also enables enterprise licenses rather than one-off purchases.
- Data review: Data sourcing doesn’t end at acquisition. Teams must capture where data is used, how it is manipulated or transformed, and where it delivers best results. In the words of one CDO in the financial services industry, "We don't do anything that we don't describe." And, that ensures transparency and accountability, a requirement for compliance with strict data regulations.
Take a look at the “path to coordinated data acquisition” in the Forrester Infographic. https://www.forrester.com/report/Forrester+Infographic+Think+Youre+Doing+Data+Sourcing+Right+Think+Again/-/E-RES151895
IM: What are top tips to all organizations on how they can get started hunting down the data that they most need?
Belissent: To get started data leaders must:
- Inventory external data sources, including terms and conditions, and current use. Your first order of business is to know what external data exists within your enterprise. Where data acquisition policies have been liberal or nonexistent, external data is usually lurking in corners around the company. One CDO lamented that "since our data is siloed, we don't even know which data assets are where."
- Establish a cross-functional steering committee of data owners, and a formal data sourcing process. Whether or not the actual data use remains in the business or in a centralized team, the "owners" of the data are often those who benefit from the insights — that is, the business teams. Form a cross-functional working group with representatives of these teams to ensure coordination of data needs and acquisitions.
- Quickly post an opening for a data hunter. Expect to see more job listings for data hunters. These data sourcing specialists are the hot new job in the insights world. With growing interest in artificial intelligence and the need to train models, demand will grow, and salaries will follow. Getting a jump on this will be worth it. Some business units resist this centralization, but most recognize the flaws in a fragmented approach. You may be able to replicate S&P's reduced FTEs and data spend, along with its significant cost savings.
- Include lessons about sourcing in data literacy programs. There's a growing interest in raising data literacy within companies. Data leaders launch literacy programs that include a wide range of learning paths and address many data and analytics topics. Data sourcing must be one of them.
Here is a data request form that can help: