Ever since the publication of this story in The Economist a few years back, "Data is the new oil" has become a common refrain. The analogy is apt, as much like raw crude oil, data on its own has limited use. As a consumer, virtually every action of ours, from hailing an Uber to logging into our bank accounts generates and leaves a footprint of data.
Data privacy regulations and increased consumer awareness of these (and very rightly so) have started creating standard definitions on types of data. These are based on how that data is collected, and on the contextual relationship related to the units on which data are being collected. For example, zero-party data is defined as data proactively shared by a consumer with a brand, while third-party data is essentially obtained by the brands from data providers and aggregators who may not have a direct interaction with the audience on whom the data is collected.
This blog highlights keys to success using third-party data in the financial services sector. In recent years, there has been an explosion of third-party data categories and availability, especially with cloud enabled ‘data exchanges’. Given the plethora of choices, brands need to have a solid action plan for each of the top five pivots for a successful data strategy:
- Well determined key performance indicators (KPIs) and appropriate use cases
- An objective understanding of third-party data challenges
- Streamlining the data sourcing function
- Integrating with analytics – fuel that powers the data
- Data refresh – don’t let your data get stale
Ask the right questions
Creation of a strategic framework of KPIs that you are trying to impact with your data strategy is an essential first step. For example, the third-party data categories that you’d want to look at to understand when (or which) consumers are coming to the end of their current auto insurance policies will be different from those that will tell you if the exact same set of consumers are also ripe for a life insurance cross-sell offer (along with their auto policy renewal offers). While there are many advanced data science techniques that are available today, the old-fashioned business and contextual prioritization of data categories enabling you to ask right questions determine downstream success of your data strategy
Objectively evaluate the data ‘quality’
Once the KPIs are ascertained, you need to consider data from different sources, as there are many third-party data vendors and providers. Evaluating data must encompass both qualitative and quantitative aspects. The stability and viability of the data, as well as the data provider is an important consideration, as is the additional scale the third-party data will provide to the brand. Checking accuracy of the appended data (very often you can benchmark against your own first-party data) is critical. For example, a medicare insurance provider conducting targeted marketing needs an accurate assessment of the age-based eligibility for its target universe. One of the most important considerations should also include match rate (put simply, what proportion of your customer universe can be matched against the third-party data provider’s universe). Your data strategy should incorporate a comparative weighting schema for all these important metrics.
Streamline data sourcing
Data sourcing is a very specialized function. Often it needs to become a full-time commitment for a team depending on the size of your brand and complexity of your data strategy. Working with independent providers and then trying to bring all the data back together at the customer level is not only a time consuming task, but also a fairly technical undertaking from the coding and data matching perspective (Merkle, for example, has several level of matches available for the highest fidelity using terrestrial and digital connected recognition schema). Furthermore, the nuances around data privacy are getting quite complicated as different US states may have slightly different rules of interpreting data privacy, as well as regulatory and compliance aspects depending on the industry (e.g. FCRA, GLB and the SOX rules commonly used in financial services industry data). Alternatively you can choose to work reputable data providers and agencies that will do most of the legwork for your brand, including logistics negotiations with the data providers, help you ingest and aggregate data from different sources (including niche cloud based data marketplaces) and ensure that all data transactions and usage are privacy and compliance safe.
Integrating analytics with your data strategy leads to dozens of third-party data applications, of which I will mention only a few:
1. Enhancement of your own first-party data using third-party data at scale.
The primary driver of this is the requirement of the brand to know the holistic profiles of your customers beyond what the first-party data tells you.
2. Expansion and scaling your target audience through modeling and targeted media activation remain one of the most powerful applications of third-party data.
3. Brands may not typically associate this with third party data, but it can be helpful for research. While there is no substitute for primary research, third-party data-based analytics on syndicated data sources (like consumer and shoppers’ panels) yield rich insight. For example, you could discover that your most valuable target audience segment shows certain types of propensities when it comes to media engagement and consumption, and that sets you ahead as you formulate marketing plans.
Don’t let your data get stale
Data refresh strategy is a very critical requirement, though, and is an area that brands often spend the least energy and focus on. If there is one thing that is constantly changing in the information age, it is the variety and velocity of data available. Brands should continue to look at the applicability of their current data for the KPIs and use cases they had procured or sourced the data for. Static data models and predictive statistical models should be re-evaluated at almost a regular frequency (at least once a month/quarter) using technology and latest analytic methodologies. Given the size of the optimization audience, even a few basis points ‘lift’ on the current performance of a predictive model, will make this effort worthwhile, and a well thought out data refresh strategy will ensure that the performance and return from your data and analytics models will continue to remain positive.
Examples of successful applications of third-party data are numerous. Within the Merkle dossiers alone, we likely have dozens of such examples, but I wanted to highlight two that shows the power from ‘intersection’ of data dimensions
1. The first one is around application of geo-location data in auto insurance. As with any purchase, auto insurance is highly correlated with ‘the right need at the right time’. Using geo-location data, Merkle was able target consumers who were visiting establishments that showed a high causal relationship to auto insurance purchase (for example, an automobile dealership), and show double digit lifts in test/control setup for auto insurance purchase rates.
2. The second example is the integration of first-party and third-party data to identify ‘engagement media and channel’ propensities for customers who would be ‘high value’ checking and deposit customers, as applied for a large regional bank. Activating this audience through the Merkury platform showed not only higher engagement (and conversion rates), but also significantly higher ‘quality metrics’ (like deposit amounts and direct deposit incidences) for the test populations.