If a business wants its competitors to read 'em and weep, it needs artificial intelligence, data scientists and a chief information officer willing to lean hard on data collection and management.
A lot of the AI algorithms used by data scientists have been monetized and made available by open source.
"It's becoming more and more cut and paste into terms of putting together your AI model," said Robert Eve, senior director of data management at infrastructure software company TIBCO, in an interview with CIO Dive.
Building blocks of AI models will allow a level of consistency to emerge among algorithms. But the problem is, if a company is more or less equal to its competitor, how can it win?
Advisory firm IDC outlines three factors companies need to effectively compete using data:
A rich data set
Algorithms to make sense of data
High-speed and scalable computing power to run the algorithms on data
Modern companies have the luxury of infinitely scalable computing because of cloud access. "Microsoft, Google or Amazon — just pick one," said Eve. Because everyone has access to the cloud, "everybody has equal footing."
The differentiator is data.
Data comes first
Incumbents, including Unilever and Procter & Gamble, have the potential to outperform digital giants because they possess about 80% of the world's commercialized data.
Digital giants were faster to build business platforms that harnessed personal data, but incumbents have the potential to reach the "pinnacle of personalization," according to IBM.
Proprietary data is something competitors can't match. If a company excels in processing data, applying it thoughtfully, the business is rewarded with the ability to outpace competition.
But it's not just consumer data that has value.
"There's data on everything," said Jennifer Belissent, principal analyst at Forrester, in an email to CIO Dive. "It could be as simple as sourcing weather data to inform decisions on vehicle routing or store merchandising."
Insurance providers, for example, want data on customers committing to a policy and the context "in which they will be insured" because it will influence their premiums pricing, according to Belissent.
Data collection is exploding, with live streaming data flows from internet of things devices, beacon technology and recommendation engines. It also comes from data brokers, aggregators and marketplaces.
More than half of companies sell their data in some form, up from just 14% of companies in 2014, according to Forrester. Operational data is the most commercialized form of data, followed by sales transactions, financial performance metrics and inventory data.
Modifications in data scientists
It's tempting to over-allocate resources to solutions other than the data itself, according to Eve. But companies need to put money and leadership behind data.
More than one-third of "commercializer" companies — those that commercialize data — use machine learning data cataloging, according to Forrester. More than 40% of commercializer companies use big data governance tools.
There has been an evolution in self-service business intelligence tools, including Microsoft Power BI and Tableau, in the last 10 years. All of the products make it possible for traditional business analysts to do reporting and analysis.
This has given way to a new class of data professionals called citizen data scientists because when skills are combined with more advanced tools for data analytics, "mere mortals can get it done," said Eve.
It's not a compromise for companies to turn to citizen data scientists instead of more traditional, highly sought after and costly data scientists.
Vendors are developing better tools for data management so it doesn't require as much "hand craftsmanship," according to Eve. While data scientists might be used to writing in detailed code, the citizen data scientists might have more of a workflow-driven tool.
The tools are "almost as easy as using Excel," said Eve, making "citizen data scientists nearly as skilled as the data scientists."
Modifications in data professionals are leaving room for CIOs to make adjustments to their data strategies. The practices are allowing companies to build a model around acquired data and refine the model as needed until a business outcome is achieved.