How to Design an Analytics Stack that Humans Actually Use

By Julie Lemieux

Published on 2021年8月9日

How to Design an Analytics Stack that Humans Actually Use

Julie Lemieux is the former Head of User Experience at Databricks and Founding Member of Chief, a private network for senior women leaders. In her current role as VP of UX, Design & Research at Sigma Computing, she deploys human-centric design to support data democratization and analysis.

Less than 40 percent of Fortune 1000 companies are managing data as an asset and only 24 percent of executives consider their organization to be data-driven. While an overwhelming majority of these executives (92.2 percent) cite culture – a mix of people, process, organization, and change management – as the primary barrier to forging a data-driven culture, it is worth examining data democratization efforts within your organization and the business user’s experience throughout the data analytics stack. A data engagement experience audit is likely to reveal significant gaps in the journey that need to be filled before you can reasonably expect business users to engage with data with any sort of regularity or confidence, let alone, leverage data to inform their daily decisions.

My career has been dedicated to improving the user experience (UX) through research and design. Over the course of the last two decades, I have developed a human-centric approach to product design that employs empathy and compassion for users as guiding principles. Through my work, I have discovered the most common points along the business user’s data engagement journey that can determine the success or failure of a data democratization initiative.

5 Tips for a Business-Friendly Data Analytics Stack

Democratization isn’t “done” just by giving employees access to the data. One must humanize the data analytics stack by making it easy to navigate and by providing sufficient context for everyone who is expected to be “data-driven.” The below are just a few examples of how to make your data analytics stack approachable and functional for business teams:

1. Help People Find Data On Their Own Terms. Most people don’t think in data terms. They think in business terms. Data sets need to be thoroughly annotated or tagged with keywords that business users commonly use.

Data catalogs, like Alation, are often a business user’s point of entry to data. Catalogs close the distance between data and business users because they allow data practitioners to create department or user-centric collections of data sets, tables, and other content, and tag them with business vernacular keywords and context so business users understand what they can use their specific line of inquiry.

The alternative investment management firm, Blackstone, exemplifies a business that saw the importance of humanizing their analytics stack. “We saw that making data searchable via human language through a data catalog was essential to supporting a data-driven culture,” recalls Blackstone’s Chief Data Architect and Liquid Markets Chief Technology Officer, Tom Pologruto.

This insight led Pologruto to launch a data democratization initiative, which he quickly realized would require the company to move much of its data analytics stack to the cloud. Today, Blackstone’s data analytics stack includes Fivetran, Snowflake, Sigma, Jupyter Notebooks, Alation, and various other tools.

2. Help People Get Into An Analysis Flow State – Fast. Companies with 2,000+ employees have an average of 175 applications. Business users are suffering from serious SaaS fatigue. It isn’t uncommon for a business user to see something on a dashboard that intrigues them and submit a request to the BI team for that data. It is eventually shared with them in a CSV file that needs to be opened in either Excel or Google Sheets for analysis and visualization. This all-too-common workflow forces people to switch tools, learn multi-step processes or worse, develop their own software hacks or workarounds to do their work which doesn’t allow them to enter a flow state; forcing them to deal with software problems rather than allow them to focus on business problems.

The integration between Alation and Sigma at Blackstone is an example of how to solve this workflow issue. Once people have found the data they are looking for, they need to be able to immediately begin their analysis. Pologruto solved this problem by giving people the option to analyze data in the tool of their choice: Jupyter Notebooks for data scientists and more technical teams, and Sigma for business teams.

3. Let Humans Be Humans Part 1: What-If Scenarios. Many years ago when working at SAP/Business Objects, we asked our customers what they wanted to be able to do with their BI solution (we did not constrain them to reality). Across the board, customers wanted to answer “what if” questions and know what the future might hold. While we did not build a tool that looked like an interface from Minority Report, I did walk away with one major insight: People need to be able to change variables and model various versions of the future to take full advantage to find insights and make decisions.

Any “modern” data analytics stack must allow people to work in familiar ways. Business users love pivot tables and lookup functions. You’ll never convince an Operations or Finance manager to give them up. Deploying an analytics solution that provides these necessary features, as well as simplifies what-if analysis and scenario modeling, will significantly lower the barrier to entry for business users when it comes to analyzing data because it mirrors how their brains tend to think about data analysis and it allows them to rapidly enter a state of data exploration flow.

4. Let Humans Be Humans, Part 2: Add More Data. It is a rare occasion that all of the data a business user needs arrives in a single, perfect table. More often than not, a business user will have additional data they want to add more data to their analysis than what is contained in any single data source. People need to be able to add related data to their analysis so they can consider additional variables, which often leads to more impactful insights. Adding more data also lets people tell the data story their way – the very definition of being data driven.

In an ideal world, an analytics tool would notify people when data sources have shared or complementary data that can be used together, as well as when other users within their organization frequently use two data sources together, have the same or similar data sources, etc.

5. Let People Tell Their Data Story In Their Own Way. Business users are told that they must be data-driven and they must justify decisions with data. Yet, they have few means for contextualizing data or data storytelling that are as easy to use and customize as their old standby: PowerPoint.

Business users are bound by the dashboards others have built for them or the charts and graphs they are able to create in Excel or Google Sheets from a data extract. Screenshots are taken and the occasional table or graph from a spreadsheet is embedded into a slide presentation where the critical context is added via text boxes. This is hardly an ideal workflow and the data on which this story is based is out of date the moment the screenshot is taken or the data is extracted from the cloud data warehouse.

The way a story is told will vary from person to person, especially when those users are on different teams, and from project to project. The stories that a product manager needs to tell are quite different from the stories that a financial analyst or marketer will need to tell. Flexible data storytelling completes the cycle for business users. If they can’t communicate their findings in a meaningful way, then what is their motivation for engaging with data at all?

Above all, your data analytics stack needs to be as human, friendly, and flexible as other tools commonly used by business teams if you want your data democratization initiative to have any hope for success.

Humans are explorers at heart. Making data analytics approachable and usable will spark curiosity and people will take pleasure in exploring data, rather than be intimidated by it. Remove the barriers that have stood between them and data for far too long and you will be delighted to see how many people are ready to exercise that curiosity and join the data conversation.

See how a data catalog empowers more people to join the data conversation. Sign up for a weekly demo today.

    Contents
  • 5 Tips for a Business-Friendly Data Analytics Stack
Tagged with