Navigating COVID-19 Uncertainty with Augmented Analytics
Businesses need fast, unbiased insights that explain the whole story in the data. Here's how augmented analytics can help.
- By Pete Reilly
- November 4, 2020
Businesses and individuals alike have been inundated with "uncertain times" rhetoric in response to COVID-19. It's no coincidence; uncertainty is a hallmark of our current state as businesses juggle increasingly complex decisions.
Typical decisions are compounded by rapidly changing consumer behavior as well as concerns about business and financial performance. Answering questions such as "what's driving business performance?" is essential, and augmented analytics is key.
Augmented analytics leverages enabling technologies such as machine learning (ML), artificial intelligence (AI), and natural language generation (NLG) to assist with data preparation, insight generation, and insight explanation. The combination of these technologies augments how professionals explore and analyze data in analytics and business intelligence platforms.
What does augmented analytics mean in the face of COVID-19?
Automating Analytics Workflows
For data and analytics professionals, more than 40 percent of respondents in a TDWI study felt their employers needed them now more than ever before.
Even amid concerns about job security, analysts recognize their importance in navigating COVID-19 data. However, hiring freezes and furloughs curb analytics resources. The volume of essential work hasn't decreased, and analysts are feeling pressure to deliver insights quickly enough for managers and executives to take action.
When analytics teams are spending hours, days, and weeks on their workflows, decisions can be delayed to the point where growth opportunities are missed. The result is that overburdened analysts spend time querying and merging disparate data sources rather than acting as strategists who can advise leadership.
Augmented analytics provides an incredible opportunity to automate routine analytics workflows. In practice, this means "augmented analytics can analyze your data quickly, investigating metrics across every level and dimension to find drivers."
At a high level, here's what AI in augmented analytics is capable of today:
- AI can compile and prepare relevant data, loading it into a data cube for analysis.
- ML models automate the actual data analysis based on a business question, such as our critical question about what's driving business performance. In this example, brute-force models slice the data cube to evaluate all data combinations. Next, algebra and gradient boosting are used to determine each metric's likelihood of contribution to KPI performance, ranked by level of positive or negative impact.
- AI generates stories in the form of analysis output and narratives, bubbling up insights that users will find interesting, relevant, and significant based on their questions.
This entire process can occur in minutes or even seconds. For analytics professionals, this insights narrative can be the starting point to understanding what's happening in a business.
On another level, augmented analytics' natural language capabilities enable non-technical users to ask ad hoc questions and receive the answers they need, again freeing up time for analysts backlogged with reporting requests.
In this way, business users can become citizen data scientists, complementing the work of data scientists by performing basic analyses themselves. Capable of handling their own queries, citizen data scientists further increase the analytics maturity of the organization as a whole.
Saving time is a competitive advantage, but that would be the case under any circumstance. Automating workflows is especially critical during COVID-19 because of the need for increased business precision.
Augmented analytics enables precision by approaching data differently. Analysts would likely approach a business question by forming their own hypothesis. Analysts would test their assumptions by examining data (after merging, cleansing, and prepping the data, of course).
The analyst's approach is inherently biased. That's not to say the analyst is wrong; there is only so much time allotted to each question, and it's impossible for an analyst to test every data combination.
This is where humans and AI are different.
Augmented analytics approaches data in an unbiased manner. It's simply not limited by the same human constraints. Augmented analytics can provide analysts with an incredible advantage. Analysts can uncover insights that they wouldn't have even known to look for. They can verify their assumptions and immediately understand if those assumptions hold true.
This exhaustive analysis also provides a more thorough explanation of relationships and dependencies. An analyst would consider a number of factors to understand business performance, such as seasonality, competitors, changes, and drivers. Augmented analytics can weave those factors into a complete insights narrative so that the degree to which each factor affects performance is clear.
Understanding exactly which factors contribute to and detract from business performance helps businesses prioritize action items accordingly. Why is this capability critical for COVID-19?
In uncertainty, there are no safe assumptions. Some goods are flying off the shelves as consumers adapt to a new way of living. E-commerce and delivery services are exploding. By the same token, consumers have divested from several industries and switched from their usual brands because of pricing and supply chain challenges in the early stages of the pandemic. Amidst this change, it's unclear which behaviors will stick around in the long run. Trends are not reliable, and predictive models may struggle.
That's why it's critical to understand what's happening now. Businesses need fast, unbiased insights that explain the whole story in the data. With this capability, businesses can be more agile.
An unbiased approach ensures that decisions are truly data-driven and that businesses can zero in on the opportunities with the greatest potential for growth while mitigating risks that introduce the greatest potential for losses.
If nothing else, the rapid digital transformation caused by COVID-19 should encourage companies to evaluate their own technological capabilities.
Augmenting traditional business intelligence with AI is a promising opportunity to help businesses navigate uncertainty. Even further, augmenting analytics itself -- via workflow automation and increased business precision -- can help analysts cement themselves as table stakes strategists.
Pete Reilly leads the charge for acquiring customers and then helping them launch AnswerRocket. Prior to AnswerRocket, Pete founded and led Retality, a firm focused on helping companies conceive, build, and introduce new technologies to the market. Before Retality, Pete was a founding team member and SVP/GM of BlueCube Software, where he led the workforce management business unit before the company was sold to RedPrairie. BlueCube was a spin-out from Radiant Systems, where Pete spent eight years driving the development and market introduction of new products at the company. Pete got his start at Accenture working with global 2000 organizations. Pete has a B.S. in computer science/economics from Union College in Schenectady, NY.