And why you can't afford to drag your feet.
The value of data-driven decision-making has become so firmly established in business culture, one might think widespread data literacy has naturally flowed forth. That is, if you build the data revolution, they will come.
There’s no doubt that, generally speaking, data literacy has improved as of late. As Aaron Kalb, chief data and analytics officer at Alation, told ZDNet, many people were bored and intimidated by statistics and data as recently as 2019. “But after a year of scrutinizing margins of error in election polling, watching exponential COVID[-19] case curves and learning about ‘R-naught,’ those topics certainly seem important and impactful, and more accessible, too,” he said.
Expect that trend to amplify. By the end of the decade, you’ll be as unlikely to see “data proficiency” on resumes as you are to see “Office proficiency” now, Sudheesh Nair, CEO of ThoughtSpot, said in the same ZDNet 2021 predictions roundup.
Despite the positive trend, the quest to broadly instill data literacy at the organizational level is not yet won, according to Caitlin Davey, manager of learning experience at data bootcamp provider General Assembly.
Companies such as Microsoft, Intuit and Bloomberg have used the enterprise corporate data-training program she helped build at General Assembly.
“There’s still a lot of need, because it’s a complex problem to solve to reallybe a data-driven organization,” said Davey. “We’re still seeing a lot of need across a lot of different sectors.”
Paul Barth similarly sees evidence that true, cross-functional, organization-wide data literacy is still hardly a given. Barth is managing director of enterprise data management at Qlik, an analytics company that offers data-literacy training for all its employees and also makes its curriculum available as part of the Data Literacy Project, a partnership with Accenture, Cognizant, Data to the People and others to democratize foundational data education.
“I was talking with [employees at] one of the largest food producers in the country the other day, and they said, ‘You know, we’re a multi-billion-dollar company, but we don’t really have analytics and data built into our core business processes,” Barth told Built In.
That said, evidence of the push toward increased literacy is all around. Last year, Tableau debuted a free training program, and data-science bootcamp Metis, a subsidiary of the educational services provider Kaplan, just announced a program aimed at both technical and non-technical roles. There have also been calls for a fundamental rethinking of how we teach pre-college math, to foreground statistics and data analysis and nip the data-literacy challenge in the bud.
Until that happens, companies will have to help pick up the slack.
But how? How exactly should businesses go about instilling a data culture and ensuring company-wide data literacy? We peeked inside the curriculum of General Assembly and Qlik — both of which have long been spreading the data word — to get a sense of what works.
At Qlik, the training isn’t hyper-technical. Salespeople should learn how to analyze statistical correlations in sales data and translate a set of SQL records into a time series, but they aren’t expected to build predictive AI models. Don’t discourage anyone from advanced studies, but start simple.
For data newcomers, Qlik’s curriculum focuses on concepts like aggregations, distributions and basic data visualizations. People starting with a more advanced understanding go over topics such as Bayesian inference, selection bias and cross-validation. Wherever a person slots, expect more concepts and less code.
“If you look at DataCamp, edX and some other offerings, they often are super technical. A lot of data literacy starts with an introduction to Python — and you will see almost none of that in here,” Barth said of the Qlik roadmap.
Companies should resist overeducation, particularly at the top. General Assembly’s entry-level curriculum consists of two courses built specifically for upper-level professionals: Data for Leaders and AI for Leaders. They teach leaders how to look for opportunities to use data or AI to solve a business problem, but they position the student more as a consumer of data than a creator.
Managers and individual contributors need to be able to wrangle and analyze data more than the C-suite, “who is consuming reports or dashboards but not producing them,” Davey said. (That exempts a chief technology officer or chief data officer, of course.)
In education, we all come with different baselines. Data literacy is no different. After some time educating staff, Qlik realized that trainees tended to fall within four levels of expertise and receptiveness. From there, Qlik built four “data personas.”
At the bottom is the “data avoider” — the data-skeptical, trust-my-gut type — and at the top is the “data guru” — someone experienced in data analytics and maybe even well versed in data science. In between is the “data newcomer” — receptive, but green — and the “data apprentice” — someone “eager to further their skills in data science, algorithms and statistical analysis,” according to Qlik.
Most people are either newcomers or apprentices, Barth said. To see where someone slots, Qlik has employees take a short, 10-question survey. Questions include:
To avoid baptism by fire, the training is then spread across eight weeks — two weeks each on reading data, working with data, analyzing data and communicating with data. Each course contains a variety of modules, tailored to expertise level, which range from an explainer on why analytics is important all the way up to — for the gurus — multiple linear regression analysis. The modules, like the assessment, are available via the Data Literacy Project.
General Assembly also breaks down training depending on whether someone is a novice, an expert or somewhere in between. Its “in-between,” however, allows for deeper technical dives, including work with Excel, SQL and Tableau and an option for a one-week Python accelerator.
At least for a time, the pandemic led to a massive open online course (MOOC) revival, with online courses serving as professional counterpart to sourdough- and Peloton-styled self-improvement. But that upskilling didn’t necessarily extend to organizational ramp-ups in data literacy.
“A lot of companies were hoping to delay training so that they could conduct it in-person, because of the secondary benefit of creating new connections and getting new ideas together in a room,” Davey said.
But remote-ness shouldn’t pose any significant barriers to a successful data-literacy initiative. The ability to create cross-location cohorts, draw on collaborative documents and shift into breakout rooms all kept the process streamlined, according to Davey. In other words, don’t use remote work as an excuse.
More importantly, the proof of value is in the numbers. Researchers at the MIT Sloan School of Management found that data-driven firms have between five- and six-percent higher output and productivity than expected. And research commissioned by Qlik and conducted in partnership with the Wharton School found data literacy boosted enterprise value, gross margin and returns.
Throughout our conversation, Barth stressed one point above most: Don’t throw a bunch of training modules at the problem and expect it to be solved. That is, for data literacy to mean anything, a company needs to find tangible opportunities to make the education actionable.
“The first thing we recommend is something we call value engineering, which is identifying the business need and the business opportunity for introducing data-driven processes and analytics,” Barth said.
He pointed to how the pandemic shined a harsh light on suppliers that didn’t have integrated data sources and therefore couldn’t effectively model demand. Part of that is technical, of course — building a data pipeline that catalogs and indexes quality data so that it’s easy to understand and use across the board.
But it’s just as important to keep data-literacy education directly linked to practical, real-life business projects, whether that’s churn prevention in sales, customer lifetime value calculation in marketing or whatever fits your company best.
You can do it in short order — “in a few weeks, really,” Barth said. Identify some business cases, “then build out a plan to say, ‘How are we going to transform the way we run this business and introduce data analytics, and what skills need to go along with that?’”
RelatedA Day in the Life of an Analytics Translator