How much does it cost for that train to move a ton of freight?

Imagine you’re the daughter of a parent who asks the title question as they watch a freight train pass overhead on a bridge. These were the type of questions typically asked of candidates seeking jobs in the management consulting industry.

How much will it cost to move Mount Fuji 100 miles to the north? Illustrative Consulting Firm B-school Recruiting Question from the 1990's

The interview candidate would be expected to estimate the volume of a cone (1/3 height x the area of the base), the approximate height of Mount Fuji, the approximate volume contained in an earth mover, the cost to operate an earth mover for a day, etc. The candidate was expected to come up with a methodical approach to arriving at the actual cost it would take to move a volume of earth from one location to another 100 miles to the north.

Who cares how much it costs to move Mount Fuji?

No one really, but the concern is whether or not the interviewee could build an initial estimate of the cost under the pressure of the interview. It was to test the analytical capabilities of the candidate. Could they step back from an impossible task and break it down into bite-sized pieces that could then be added back together into a credible whole. They later learn that certain variables in the cost equation turn out to have a much greater impact on the final answer, and to focus early analysis on those variables.

R&D researchers need to learn to estimate all the variables end-to-end and focus on those variables that most affect the end game+ (either in commercial viability or in the direction of the research). We’re looking for a map of all the important variables that stretches from beginning to end. We prioritize those variables according to an estimate of their influence on the end. We want a program that takes the results from each of the analyses and updates the map continuously.

It’s important for World Class R&D researchers to work comfortably with estimates. In innovative pursuits where we end up can be quite far a-field from where we first intended, we tentatively map out a research program knowing it’s only a sketch, and fill in the sketch as new data arrives. This approach is good in getting everyone to think through all the variables that influence the results, not to inadvertently skip key variables, not to over-work any step in the map before its time, and to prioritize research toward those variables that most influence results.

This approach means researchers need to be able to make estimates, to round off answers that are estimates anyway and to forge ahead through the entire chain of thinking to estimated results. This allows researchers to gain a tolerance for ambiguity+, to recognize that the most important causal elements are often localized to few variables; that many variables can be off by an order of magnitude and not significantly affect the final answer. We make sure (empirically) we’re working on the right thing before we spend too much time heads-down on the exacting work of research.

This is the concept of the skinny path. Find me a skinny path to the customer with this R&D concept you have. Let’s find out all the variables that affect the customer purchase sooner rather than later, and which are the most important factors influencing the buying decision. When you think you have the answer that’s only the start. You can’t spend enough time thinking about this question. It’s another way of saying that causality must be a part of the wharf and weft of our daily activities. It guides daily work. Instead of moving from beginning to end using Inductive Probabilistic Reasoning+, we instead work to gradually bring the entire narrative into focus. It’s not the individual links that are the most important: it’s the chaining together of the links.

System dynamics as a modeling tool captured the interactions between variables. The underlying premise was that estimates of the interactions (if A increases 10%, then B decreases 40%) far outweighed estimates for any of the variables (e.g., A = 100, B = 10). Focus on the interactions. Benefits promised by consultants selling this tool are typically overwrought, claiming the interocular+ as their own, but the concept is sound. Don’t focus on the variables, the price of gasoline, but on the interaction between the variables: how the price of gasoline affects the attractiveness of competing modes of transportation for our customers.


Use this approach merely to ensure you have considered all the variables and their relative weighting. It is not a modeling exercise. We improve the contribution of the human mind; we do not build a quantitative model. The problem with researchers is they often think through the variables, but give little thought to the relative weighting of those variables on the final results.

It is also important for creative endeavors that neither the beginning nor the end be fixed. We need room for discovery and re-prioritization. We need a program that allows us to revisit prior results in light of later findings. Often the purchasing decision for innovative products is broken down into gotta-have, nice-to-have and extraneous. These three categories are fluid based on how we position the product in the mind of the consumer and / or payor, and so should be our approach to evidence gathering.

Note also that this approach only influences one facet, the rational facet, of our thinking process. If people are involved in the process, as in most creative endeavors, then qualitative factors can have a much greater weighting in the final results. What are the variables that excite the passions of our researchers and get them to want to build an even deeper understanding of the underlying science and commercial needs. Even management consulting firms have moved away from the rational style of interview. The human dimension of consulting, the need for empathy and personal identification, are much more important in effecting major change in client organizations.

In the end, the daughter in the opening scenario came back with the usual response of the researcher:

Who cares?

It’s like a foreign language to PhD students in molecular biology. If you’re trained to dissect scientific questions down to their molecular fundamentals, then you have little interest in how they roll up into the larger questions. It’s just assumed that someone else is worrying about that question.