By: Melissa Wallace, Sales Representative — Risk & EHS Solutions
Risk and safety managers are increasingly being asked to expand their role and weigh in on a growing number of challenges organizations face. These situations are dynamic and require the ability to analyze data and identify key trends. When that data is captured in spreadsheets and legacy systems, however, providing insightful analysis that influences decision-making can become difficult. Overcoming these challenges requires rethinking what makes data actionable and understanding the limitations inherent in using spreadsheets as a reporting tool.
An elusive goal — becoming data driven
As organizations wrestle with making effective decisions in a time of uncertainty, many are exploring or revisiting efforts to establish a data-driven culture. A CIO article defines a data-driven culture as one that “embraces the use of data in decision making. It treats data as a strategic asset of the company by making data widely available and accessible.”
In an interview with the MIT Sloan Management Review, Rahul Pathak, general manager for analytics at Amazon Web Services (AWS), explains why these efforts are so critical. “[B]eing data-driven enables you to start getting all the signals about what’s happening, and then you can put yourself in a position to respond,” Pathak says. “Once you’ve got the ability to understand what’s happening in your business and the ability to act on it, you can start to make good decisions.”
Yet achieving that goal can prove exceedingly difficult. In Creating a Data Driven Culture, Brian LaFaille of Looker Data Sciences, Inc. notes, “While most companies are amassing data at an unprecedented rate, many are struggling to create a true data-driven culture that sticks.” A large part of this struggle may lie with an organization’s choice of tools for working with data.
Why spreadsheets are doomed as analytic tools
Those relying on spreadsheets to power a data-driven process are likely familiar with their shortcomings. Spreadsheets don’t scale well, lack audit trails, obscure analytical logic (and errors), and are awkward collaboration tools. This isn’t surprising given that spreadsheets were never designed to solve those problems.
“Excel is the most commonly misused program in existence,” states the author of Excel Spreadsheets are Often a Giant Waste of Money. Citing business intelligence, reporting, and executive dashboards as popular examples of misuse the author concludes, “A spreadsheet is a desktop application meant to be used by one person. When the application needs input from multiple people or the data needs to be secure and shared by multiple people, the spreadsheet is the wrong tool for the job.”
Since spreadsheets are often used as a vehicle for delivering data (as opposed to just running calculations) many analytic efforts are built upon a problematic foundation.
Assigning homework to others
There are obvious challenges when trying to use spreadsheets to support collaborative efforts: scalability, extensibility, auditing/security, readability, etc. A less apparent problem, however, may contribute to the difficulty organizations encounter when trying to make a data-driven culture stick — lack of relevance.
Each stakeholder has a “What’s in it for me?” perspective when looking at any presentation of data. If extra steps need to be taken to filter out the noise and extract relevant information, viewing a report becomes a homework assignment that, given substantial enough workloads, is likely to be avoided. As a result, the potentially valuable insights hidden in reports will often go unnoticed and unused.
Actionable data: back to basics
Alaa Khamis’ article Turning Data into Actionable Insights suggests starting with data that answers five key questions:
- What has happened?
- What happens now?
- What is the trend of a certain variable?
- What is the relationship between variables?
- How is an item performing with respect to other items or a benchmark item?
Data that answers those fundamental questions leads to more insightful analysis, engaged conversations, and well-defined areas of focus, all of which leads to better decision making. John Spacey extends the concept of actionable data by identifying four key components: timeliness, accuracy/precision, credibility, and relevance. A closer look at each of these components illustrates why spreadsheets are ill suited to deliver actionable data that powers a data-driven culture.
Pathak’s interview emphasizes the importance of responsiveness to becoming more data driven. “The faster you can adapt to the changing environment, the more successful you’ll be at navigating it,” he says. But the age of data also affects the quality of decisions it impacts.
In Measuring Timeliness is Vital to Understanding Data Quality, Experian’s Rachel Wheeler notes, “Timeliness is a key factor when it comes to getting good data. If a company looks at information that's outdated - even if only for a few weeks or months — it can make all the difference.” Unfortunately, spreadsheets are loaded with built-in delays resulting from aggregating and assembling data, to keying (and often rekeying) information, to formatting reports and localizing versions by business unit or department.
These delays damage the timeliness of data, and ultimately limit its value for use in decision making.
A previous Origami Risk blog discussed the well known accuracy issues with spreadsheets:
“Studies show that nearly 90% of all spreadsheets contain errors and up to 50% of spreadsheets used by large companies contain material defects. Lack of an audit trail makes it impossible to identify the point at which errors were introduced, thus limiting any chance of quickly making corrections.”
One key factor contributing to the high error rate associated with spreadsheets is lack of transparency with the business logic employed. It can quickly become difficult for anyone but the original creator to sort out how everything is connected.
The article Meet the Excel Warriors Saving the World from Spreadsheet Disaster profiles David Lyford-Smith, a specialist in ferreting out spreadsheet errors. In his experience, transparency is often an issue. “I would try and look at it and see if I could understand how it works without the person telling me,” he says. “Usually, the answer would be no.”
If someone who spends days at a time analyzing spreadsheets for problems can’t easily decipher the logic powering typical corporate spreadsheets, the odds of a fellow coworker consistently doing it are very low.
An essential part of actionable data is that consumers of the data trust the information presented. One way to build that trust is to minimize the points at which noise and error are introduced to the data. Technologies like portals and mobile apps can push data capture out to the field, resulting in fewer opportunities for misinterpretation or relay errors. Collecting data closer to the source tends to reduce noise and build credibility.
A second approach is to reverse the process and let the problem guide the data. Tom O’Toole, executive director of the Program for Data Analytics at the Northwestern University’s Kellogg School of Management, suggests starting with identification of smaller problems that can be resolved fairly quickly. “Identify a small number of ‘high-leverage’ business problems that are tightly defined, promptly addressable, and will produce evident business value, and then focus on those to show business results,” he says. “The specific business problem drives the team to identify the data needed and analytics to be used.”
Linking improved outcomes to data-driven decisions can have big benefits. “There’s no substitute for business results to build credibility for data analytics and sustain commitment,” O’Toole concludes.
Determining which part of a data set applies to each consumer of data can be difficult when using spreadsheets. If additional calculations must be run to extract unit-level data from an enterprise-level report, for example, it is far less likely to be used. Marketing professionals know firsthand the power of personalization. Campaigns that use personalization techniques to deliver more of a one-to-one individualized approach rather than a one-size-fits-all methodology can deliver a $20 return for every $1 spent according to some studies.
Without personalization, data becomes the “homework assignment” mentioned above. Unfortunately, spreadsheets and “canned” reports from legacy systems too often fail to highlight key information or allow the readers to easily extract the data most relevant to them.
An example of actionable data
As discussed in the Origami Risk webinar Leveraging RMIS Technology: Doing More with Less, the City of Jacksonville produces a Mayor’s Report that is sent out each Monday. The report provides each department with a rolling 30-day dataset including all workers’ compensation, general liability, and safety incidents, as well as weekly and quarterly trends. When COVID-19 incidents began occurring, department heads wanted to see those numbers updated more frequently and so many requested access to the system.
The city’s risk manager, knowing that much of the other data in the system was not appropriate for widespread distribution, followed O’Toole’s quick-win strategy and let the problem drive the data. She built a dashboard that automatically pushes out each department’s COVID-19 incident data twice daily. “They know their numbers, can see the impact of changes, and can see the rate of growth,” she says. As for the “quick” part of quick wins, she commented that it “took me seven minutes to get it out.”
This COVID-19 report contains all four of Spacey’s components for actionable data. Run twice per day, it is timely. Leveraging the incident collection portal each department already uses to directly report their other incidents boosts both accuracy and credibility. And the ability to personalize the report so that each department sees their own data (with no homework assignment required) makes it easy for department heads to use the data to gauge the effectiveness of tactics in real time. Months after rollout the feedback has been nothing but positive and the only change requests have been to add more names to the distribution list.
The City of Jacksonville use case underscores the importance of delivering personalized “quick wins” with data. Had this data been trapped in spreadsheets, the effort required to aggregate and then distribute the report would likely have swamped the risk manager — pulling her away from more important work.
Creating a data-driven culture is a worthwhile goal. Insightful analytics can improve decision making, especially in highly dynamic environments where finding the best options can be challenging. To reach that goal, however, organizations must use the right tools. This means moving beyond the built-in limitations of spreadsheets.
Talk to us to find out how to move your organization beyond spreadsheets, or reach out to the author to discuss more.