From First to Third, and Alternative Too: A Guide to Data Types
One way to create a competitive edge in your business is to use the latest and greatest analytic techniques. Stream processing, deep learning, and graph analytics all fall into that category. But another way to drive competitive differentiation is by incorporating different types of data – ranging from first-party data to alternative data — into your analytic recipes.
First-party data is data that originates come from your own file systems and databases, which hopefully are brimming with years’ worth of data describing your products, customers, partners, and business. In many ways, this internally sourced information from CRM, ERP, sales, marketing, and point of sales (POS) applications is the ultimate data differentiator, because nobody else on the planet but you has access to it.
Your company’s Web and mobile properties are another good source of internal data that hopefully only you can access (although some of it may be accessible to others if you outsource it). Knowing how your customers navigate your company’s digital properties can provide insight into their wants, needs, and expectations, specifically as it relates to how they interact with your company’s main products and services.
These internal systems often constitute the core data feeds for journey analytics projects that track the individual experiences of users in an attempt to improve customer experiences. They’re also fed into Customer 360 initiatives, which are increasingly being standardized by big tech players like Adobe, Oracle, Microsoft, Salesforce.com, and SAP.
Many data analytics projects start with first-party data, which is the most valuable data in your organization. But to really get the analytics party going, you might want to branch out and incorporate other data sources into your data mining or predicative projects.
Second- and Third-Party Data
Second-party data is basically first-party data collected by somebody else that you have been given access to, often as part of a mutually beneficial arrangement between two or more companies.
Second-party data can be quite valuable, particularly if it gives you data that you wouldn’t normally get. For example, if you’re a clothing manufacturer, you might partner with a retailer to gain insight into what styles of clothing are selling and what styles are building up on the shelves.
What makes second-party data valuable – is exclusivity – is what separates it from third-party data, which is data that’s been gathered and aggregated by large data brokers and other outfits There are hundreds of third-party data aggregators selling data that’s applicable for a wide range of industries, although some of the biggest and most well-known have names like Acxiom and Dun & Bradstreet.
The U.S. Government has a mandate to share many of its datasets, and so it makes many datasets from the Federal Reserve to the Bureau of Labor Statistics and the U.S. Census available to download free of charge. These would be considered third-party data sets, since they’re widely accessible.
Third-party data can also refer to data that originates from credit reporting agencies, loan information providers, and even telecommunications providers, which package and sell its customers’ geolocation data in an aggregated and anonymized form. There are also online marketplaces where analysts can access third-party data, including Lotame, which also offers first-party and second-party data services.
Some caution is required when using second-party and third-party data. Last May, FINRA issued a warning about the risks posed by using data aggregators. “Many data aggregators may operate under limited regulatory oversight and are not subject to the same regulation that registered financial institutions are subject to, particularly in areas of data privacy and security,” it warned.
The use of so-called alternative data is an emerging trend in the financial services industry, which is often ahead of the game when it comes to innovative uses of analytics and business intelligence. Investors often look to incorporate alternative data feeds into their quantitative analyses, where it’s used as a source of “alpha” to drive investment decisions.
Alternative data augments the standard types of data that analysts or “quants” can be expected to use as part of their normal data crunching routines. By adding into the mix sources of alternative data – such as Web traffic data, geolocation data from mobile devices, credit card transaction data, social media posts, satellite imagery, and IoT sensor data – an investor can potentially gain a competitive edge through better insight into consumer activity.
One knock on alternative data is that it can be difficult to handle. Most alternative data types are unstructured, to varying degrees, and the quality can be inconsistent. In that regard, alternative data shares many similarities with traditional “big data” sources that are mined for insights in massive databases or fed into machine learning algorithms to generate predictions.
Bloomberg recently added support for alternative data in one of its information products in an attempt to streamline access. With a few clicks, a Bloomberg Enterprise Access Point customer can download alternative data that describes metals inventory, the sentiment of equity bloggers, the drug approval pipeline, consumer footfall, parking lot activity, construction permits, geopolitical risk, and mobile app utilization.
Gerard Francis, who’s the global head of enterprise data at Bloomberg, says demand for alternative data is growing among clients.
“Our clients are looking to integrate alternative data into their investment process. As Bloomberg is a leading source of reference and pricing data, the integration of alternative data into the same Bloomberg Enterprise Access Point and Hypermedia API is compelling,” Francis says in a press release. “We make it easier for our clients to contract, ingest, test and normalize their alternative data. Firms can come to one destination for data that can meaningfully inform investment strategy.”
Expanding Data Universe
Bloomberg sources alternative data for its Enterprise Access Point offering from 20 providers, including Thasos, Apptopia, TipRanks, PredictWallStreet, RS Metrics, Orbital Insight, OWL Analytics, Predata, Evaluate, 280First and Symphony Pharma. The New York company has plans to add more sources in the near future.
Some of these outfits provide specific types of data, like RS Metrics, which specializes in tracking the world’s “shadow” metal inventory using large-scale analysis of satellite and aerial imagery. OWL Analytics, however, specializes in helping to boost the quality of unified data across diverse data management systems, from Apache Hadoop to Apache Kafka to IBM Db2. “If your data is wrong, your business is wrong,” the company says.
Other companies are navigating the alternative data wave too. Silicon Valley’s Import.io has built a business based on making it easier for analysts to incorporate Web data into their big data projects. Web data, the company says, is the biggest form of alternative data, and also the richest in terms of its diversity and potential for application.
“Over time what was once considered alternative, non-traditional data becomes widely adopted by all companies, while new sources of alternative data are constantly emerging,” the company states on its website. “It is important that you grasp the opportunity and begin to form an alternative data strategy today or risk being left behind.”
In many ways, we’re still in the early days of the big data analytics movement. The metadata “exhaust” generated by the proliferation of Web 2.0 and mobile properties near the turn of the millennium spurred the need for bigger and better data collection, storage, and processing mechanisms. That in turn opened the door for even more elaborate data management and analytics techniques on an even bigger scale, which has created all sorts of interesting new data avenues to explore.
Whether the fuel for your analytics engines comes in a third-party or alternative-data form, the potential benefit to be reaped from the mass digitization of humanity is truly immense, and in itself stands as a remarkable achievement of our time.