What a High Rise View can Teach Us about?Data

Here's an idea that's been on my mind for a few weeks! Below you'll find a 90 second and 6 minute version. Many thanks to Julia Han, Nic Carlson and Jen Cox for the incredible help.

90 Second Skim: There are two main data types: “high” and “low”. Seeing a city from above is like assessing “high” data: financial statements, operational figures, etc. You have a sense of scale and can see patterns and relationships between different components. Seeing the city from the street is like experiencing “low” data: wide range of sensory inputs informing dynamics with suppliers, products, and customers. Close exposure to different components in "low" data provides a better understanding of forces contributing to the "high" data patterns and relationships. The unique perspective and experience from the ground also reveals unique opportunities more likely missed from above. These opportunities you discover from the ground can then be scaled to other business functions and regions through high-level decision makers typically surrounded by “high” data. The implications of viewing data in this framework are to (1) treat employees working most closely with suppliers, products and customers as a valuable source of “low” data and ideas to unlock value and (2) to focus “high” data analysis on its unique strengths while providing close connection to “low” data sources to fill understanding gaps.

6 Minute Full Idea: Gazing out the window from the 45th floor, it dawned on me that peering down at the flow of commuters was a bit like scanning a financial statement. Yes, maybe it’s telling that after only a few weeks in consulting it’s hard to enjoy a view without some business analogy creeping in, but in that moment, I realized a new appreciation for the unique perspective of “high” data.

For one, I had a sense of the scale: it was a cloudless summer day in Chicago, so when facing north, Wisconsin was just past the horizon. Similarly, when assessing a business from a screen, you can neatly view reported revenue, asset value, total employee count, real estate footprint and other measures of size.

Secondly, I could see, from a high-level, patterns and relationships between different components. With the idyllic weather, rooftop pool-sides around the city seemed to be packed but not many people were swimming. I could see a delivery truck drive pass an alley entrance and continue around the block just to back into the other side. I could see an increasing number of white sails pop up on the lake as the day progressed. Similarly, high-level quantified data makes it easy to spot patterns and relationships. For a hypothetical paper distributor, you might discover that absenteeism in the Scranton branch is up 30% from last year and that revenue in New York has grown at twice the rate as the rest of the country.

This unique perspective highlighted certain opportunities that might otherwise easily be missed. Looking down at the all the blank space on top of buses, trains, and delivery trucks I thought of the potential for targeted advertising towards high rise workers/condo dwellers. While the economics of this venture are shaky, the idea would probably not have arrived walking on the street. Similarly, there exist opportunities that are highlighted more clearly in spreadsheets than plant floors. Benchmarking comes to mind: the Akron warehouse has a lower cost base and turnover than Albany, so they could probably pass on a few ideas.

An important caveat: The advantages of "high" data are not a given because you can't see out of a dirty window. Many businesses lack a clear view into parts or all of their operations. Their windows are stained with fragmented "high" data, sourced from an array of legacy systems that don't work together particularly well. Groups that can't convert "high data" into a clear view of the business (or don't have "high" data to begin with) need to coordinate decisions makers and IT teams to clear the windows.

You can probably tell it was a rare slow day, so I decided to enjoy lunch outside. The elevator plunged down and I passed through revolving doors onto the busy sidewalk. The sweltering heat combined with the noise of people talking, distant sirens, and passing traffic stood in stark contrast with the insulated, very air-conditioned office. The cars and people that had previously felt a bit like data points seemed so close, so tangible. This “low” data also provided an appreciable unique perspective.

First, the detail on the street level could help explain some of the patterns and relationships seen from above. Exploring the alleyway that the delivery truck circumnavigated might reveal a massive pothole or perhaps extra eating space blocking one end. Similarly, “low” data can reveal the factors driving patterns seen in a dashboard or spreadsheet’s “high” data. In the analysis context, “low” data will rarely be discovered in a standard financial report. It might be an open-ended survey, anonymous interview or even visit with the Scranton branch where you discover employees are disgruntled with the new, arrogant manager. It’s very hard to understand the messy, emotional, human dynamics that permeate our thinking and decisions with “high” data.

In addition to explaining issues highlighted by “high” data, “low” data can organically highlight problems and opportunities easily missed in spreadsheets. To see and hear a store or factory up close can plant ideas for improvement unseen from a distance. Additionally, these ideas, having originated from the front lines of execution, will more likely be grounded in the realm of “do-able”. Those working on the front lines — closest to suppliers, products, and customers — naturally amass “low” data. Without an opportunity to act on salient opportunities or problems, potential improvements might sink away as unspoken frustrations.

Advancements in artificial intelligence and machine learning continue to make it possible to scale data that could previously only be understood in-person. Cameras and sensors can detect quantity and species of insects in fields, recommending exact locations and doses of treatment. Algorithms aggregate an enormous range of digital and in-person transaction and search data to recommend most relevant offers. These advancements will still work in tandem with “low” data as their scopes compliment the broader perspectives of frontline workers across the business.

Given the unique and complimentary value of “high” data and “low” data, we should make sure we’re properly using both. In the system of a large business, it’s possible to overlook the nodes at the end of the management web. With a new appreciation for “low” data, it’s clear that the final nodes (writing code, stocking shelves, checking-in guests, cutting vegetables, building models in my case) are people with access to unique sensory data that can generate major opportunities and solve problems for the business if given the chance. Others have pointed to the importance of using the perspective and ability of “final nodes”, like Paul Akers in his book “2 Second Lean” with the concept of unused employee genius as a major form of waste. Tactically, if ever helping lead a firm I would want strong flows of information from those closest to the business to upper management. I might have a weekly executive office hour session exclusively for roles closest to suppliers, products, and customers. Maybe frontline employees would be given a day off each year (staggered) to think deeply about their frustrations with and opportunities for the firm- not unlike 3M’s “15% Approach”. You never know what ideas might arise while the end nodes are hustling across the plant, balancing a plate of entrees, or gazing out the window.