Sign up for our newsletter and get the latest big data news and analysis.

Fencemarking vs. Benchmarking: How to Uncover Insights Using Both Methods

Consider a home-room school teacher who is responsible for 30 students and has access to a benchmarking service that compares each student’s semester performance to all students in the district. The teacher can review the results with students and their parents on a regular basis. But the teacher has a second task: to do proactive, selective outreach based on the most noteworthy individual benchmarking results, good or bad. That is, to prioritize. Let’s explore this second task, which builds upon benchmarking but is different, and call it fencemarking.

Benchmarking, a common task in business and government, involves comparing an individual entity against peers in order to understand how that entity is doing, where it could improve, and what’s achievable by similar such entities. The process is driven by data on an entity’s traits, behaviors, and outcomes, as well as changes in these over time. The goals are to take action to improve on shortcomings, acknowledge and possibly share achievements, and set realistic targets.

Here are some business uses of fencemarking: A national corporation’s regional VPs might oversee dozens of physical offices, stores, or locations. A vendor’s account managers may be responsible for a portfolio of dozens of customer relations. A large salesforce may be managed hierarchically by regional managers. Or a vendor that services many individual customer locations may have dozens that belong to a same owner, e.g., a national chain. Such individuals or organizations, let’s call them parents, lack the time and attention to concentrate on every individual under their purview, so they need to focus on the best opportunities uncovered by individual benchmarking.

One fencemarking scenario is the extreme case of a parent who has a stakeholder interest in all the individuals in the dataset, i.e., everybody. For customer benchmarking, that could be the VP of Sales or Customer Success, or even the CEO. For internal benchmarking of chains of locations, a similar high executive is the everybody parent.

The common theme is that stakeholder parents must prioritize, since stakeholders can’t allocate the attention that each benchmarked individual could expend on itself. Now, let’s consider how doing fencemarking well differs from doing benchmarking.

The Fencemarking Task

Fencemarking, like benchmarking, involves selecting and preparing inputs, uncovering insights, evaluating them, and taking judicious action (or inaction). But the inputs for fencemarking consist of all the insights, assembled from a group of individuals that have been subjected to benchmarking.

Note that this is different from benchmarking the various groups. Using the analogy above, a school could benchmark home rooms against each other as a straightforward benchmarking application, using data on averages and top/bottom results across all students.

Here’s the general case: The parent’s fenced area contains N individuals already benchmarked against all peers, such that the parent can access all their insights. What then should the parent focus on? The key fencemarking questions are these:

  1. Where is a significant individual doing especially well or poorly?
  2. Where has a significant individual seen an especially big improvement or decline?
  3. What are the best, average, and worst values on various dimensions among the individuals?

Thus, the answers should address the more-noteworthy insights associated with the more-significant individuals. Noteworthy insights involve someone that is near the top or bottom of a large peer group. For noteworthiness, it’s hard to beat an insight that someone is the absolute best/worst among everybody, but those insights are few. On the other hand, an individual’s significance is application-dependent, and may variously be formulated as a customer’s annual spending, a store’s revenue, a salesperson’s bookings, or a combination of factors. Balancing noteworthiness, significance, and other aspects is part of the art of fencemarking.

Given a parent’s ranked insights, the final fencemarking task is the same: evaluate the insights, decide on what feasible actions to take, if any, and choose one, several, or none for implementation.

For concreteness, let’s consider a counties benchmarking engine (counties.onlyboth.com) which does automated benchmarking on 3,143 U.S. counties and county-equivalents, using over 100 data attributes downloaded mostly from federal websites. Counties lie within states, which have a limited constitutional supervision over counties, so let’s consider the author’s home state Pennsylvania as a parent.

The engine reports 656 insights that feed fencemarking over Pennsylvania’s 67 counties. Here are the top excerpted answers for four fencemarking questions:

  1. Chester County in PA has the most adults of age 25+ with a bachelor’s degree or higher (48.8%) among the 1,560 counties that are not high in natural amenities and have at least 2.7% residents employed in agriculture or other resource-based industries.
  2. Philadelphia County in PA has the 18th-most arrests per thousand residents for Part I violent offenses of all the 3,143 counties.
  3. Indiana County in PA has the 7th-biggest rise over 4 years in the prevalence of obesity (+12.9%) among all the 3,143 counties. That +12.9% represents a rise from 23.8% to 36.7%.
  4. Washington County in PA has the 10th-smallest rise in median rent for a 3-bedroom unit over the last decade (+9.5%) among the 652 counties that contain a high-density urban area and are economically non-specialized. That +9.5% represents a rise from $1,010 to $1,106.

Business and government fencemarking results would resemble these, e.g., a customer fencemarking insight might say “Of the 57 locations that spend at least $100k and use products A and B, location #742 in Pittsburgh is one of only three that doesn’t use product C.” An example action is to bring this to the location’s or owner’s attention; they might not even know that product C is on offer, which may have entered the product line through acquisition.

Fencemarking leverages the outputs of benchmarking and is carried out by parent stakeholders who must prioritize within a sea of insights before selecting the best actions. Fencemarking is uniquely enabled by benchmarking automation, which delivers numerous insights across the many available data attributes.

Fencemarking involves subtly different reasoning from benchmarking. This difference needs to be acknowledged while performing the task, unaided by automation, and in the design of software tools.

raul-valdes-perezContributed by: Raul Valdes-Perez, CEO & Co-Founder, OnlyBoth. Raul co-founded OnlyBoth in March 2014 and is CEO. He co-founded Vivisimo in 2000, a search software company that provided enterprise products and web-based consumer services, serving as its CEO for nine years and as Chairman for twelve until its acquisition in 2012 by IBM. At Vivisimo he was named a 2007 Ernst & Young Entrepreneur of the Year for the North Central Region and a top ten reader favorite for Entrepreneur of the Year by Inc. magazine. Earlier, he was on the Carnegie Mellon computer science faculty. He received seven research grants from the National Science Foundation and published 50 journal articles and book chapters in computer, natural, and social science. He received a computer science PhD from Carnegie Mellon in 1991, where his advisor was the Nobel Laureate Herbert A. Simon, and B.S. and M.S. degrees in information engineering from the University of Illinois at Chicago.

 

Sign up for the free insideBIGDATA newsletter.

Leave a Comment

*

Resource Links: