Using alternative data for business insights
Lessons from industry leaders
Using alternative data for business insights

Published by: Bloomberg Professional Services

Exploring different kinds of data can help organizations find their competitive edge and use it to make smarter business decisions. But as consumer expectations and competitor strategies are changing faster than ever, there’s a growing knowledge gap between early adopters of alternative data and their late-coming peers. Unless they close that gap, slower moving enterprises will eventually see their existing data strategies grow obsolete.

At the recent Bloomberg Enterprise Tech & Data Summit, stakeholders from several “early adopter” organizations shared their learnings with moderator Carl Reed, Global Head of Bloomberg’s Data License Product, and an audience of 150 data and technology professionals.

The top takeaways from their experiences can help newcomers to alternative data craft better strategies from the start.

1. Earning ‘additive’ insights

Alternative data should enhance and augment – not replace – the analyses companies are doing with traditional data. By using supply-chain data, for example, to supplement regression analyses done with existing methodologies and tactics, companies can gain deeper or faster insights into how distribution metrics affect revenue performance (and iterate their future strategies from there).

As they work with alternative data at a tactical level, companies need to focus on first understanding what hypotheses they want the information to inform or support. From that baseline, they can use alternative data to assess that understanding.

“Sometimes we’re using [alternative data] to ask, ‘Can we confirm that thesis that we had? Or is the data disproving of it?”

said Nadine Terman, CEO/CIO of Solstein Capital, at the Bloomberg event. and added:

“The data doesn’t automatically say anything in itself. It’s part of that process you’re building to make sure that there are pieces along the way validating or invalidating what you thought.”

For AI algorithms, the benefit goes beyond the alternative data sets alone, as it takes human judgment (and sound planning and backtesting) for companies to gain useful insights. By mapping and modeling past outcomes with look-back context into the strategy, companies can explore what signals from alternative data could have been integrated to inform their decisions for better.

Enterprise Data insights delivered monthly to your inbox

2. Building ecosystems for understanding

The usefulness of a dataset depends on both what decisions or assumptions the data is meant to inform, and how robust or proven the data set is.

Since biases can exist in any data set (or algorithm created from it), it’s important for data science and data engineering teams to assess scope, volume, and quality before incorporating a new one into the overall intelligence ecosystem.

“I tend to look at all data sets as pieces of the puzzle,”

said Jon Neitzell, Chief Data Officer of Fundamental Equities at Goldman Sachs.

“With datasets that are more established, or have been around for longer time periods, where they fit in the puzzle is technically better known. As the new datasets come online, the big challenge is trying to figure out where exactly in my mosaic does this fit – what’s the coverage, what are the demographics, what customer segment is it informing on,”

said Neitzell.

3. Structure is key to success

Successful use of alternative data requires strong structure and governance around what information is available to which stakeholders across the business, and how that data is both stored and used.

“If you don’t have your data governance and management organization in place for your existing structured data – even extending out to spreadsheets and other semi-structured data – then working with historical and alternative data is just going to make that problem worse,”

said David Saul, Senior Vice President and Chief Scientist at State Street Corporation.

As they design alternative data strategies, every company also has to decide how much data they want to “hold” in house (by buying and storing lists) versus how much they want to remain “outsourced” (or accessible on-demand from their solutions providers).

The balance of in-house to outsourced data often affects cost structure, ROI, and usage rates. What that mix looks like will be unique to every company, and stakeholders need to design data ecosystems that align with the business decisions they make regularly.

And every decision should be made with compliance in mind; enterprises increasingly need to be “look back ready” as regulators grow more and more interested in companies’ use of data.

4. Remember the limitations (and the opportunities)

For now, most alternative data sets remain less structured and proven than traditional data sets. Collection methods for data from internet-of-things sensors and other technologies are still being developed, and private information is only available from certain sources. Enterprises must be wise to work with trusted partners and iterate their lineup of data sources over time.

Ultimately, it’s a brave new world for companies in terms of what information is available to them. And as regulations around data use evolve, there remains a need for collaboration: community-building and standard-building among companies will help enterprises learn from each other, legitimize the industry and embed alternative data assessments into their processes for smarter, more informed strategies.

As Jon Neitzell pointed out on our Bloomberg panel,

good data drives great questions – helping leaders see gaps in their thinking, challenge existing assumptions, or spot unexpected business opportunities.