MECE: McKinsey’s Secret Sauce?

There was once a young Harvard MBA named Barbara Minto. The year was 1963, and McKinsey & Co.‘s Cleveland bureau let it be known they wanted a female MBA to join the office as an associate. Barbara came at the job as a skilled writer, not a number cruncher. By the time she left the Firm in 1973, the entire industry of Management Consulting was changed permanently. What could she have possibly done in ten years to have such an indelible impact?

Today’s Letter is all about the MECE taxonomies that Barbara coined. Mutually Exclusive, Collectively Exhaustive. Are they more than a badly overused McKinsey in-joke? Could they be the single most effective way to organize the way you think about problem solving?

Structured thinking is in short supply in 2020. Recitation of balkanized talking points seems to be a safer alternative to “showing your work” with original thought. Unfortunately, the business world has not been insulated from this shift. Just look at the sheer number of buzzwords required to send the average memo. We’ve become experts at talking around issues instead of talking about them. That’s actually the way we write e-mails:

“Jane, I feel our team should have a discussion around the organization’s approach to broccoli.”

In order to have productive conversations, we need organized thinking. In order to have organized thinking, we need to organize our thoughts. Seems simple – reductive, even – no? But how do you organize your thoughts?

After decades of approaching problem-solving missions with ad-hoc thought organization, McKinsey started to develop “technology” for taxonomies that enabled more consistent decisionmaking. It all came together when Barbara Minto realized that the strongest articulations of arguments resembled pyramids when put down on paper. Read more from the horse’s mouth here.

A strong recommendation at the top
Three supporting arguments underneath it
Three pieces of key data under each supporting argument

Obvious? I agree. At least, it should be. But how many people do you know who actually communicate this clearly?

Barbara realized that when you’re limited to 9 total supporting elements for a recommendation which might imply hundreds of millions of dollars of client spend, those elements need to be 1) non-overlapping and 2) conclusive.

Sound familiar?
1) Mutually Exclusive
2) Collectively Exhaustive

In a MECE taxonomy, the sum of the categories makes up 100% of the whole, with no missing pieces. At the same time, none of the items in a particular category could accurately be placed in another simultaneously. Think: US Males 0-10, 11-20, 21-30, etc.

For consultants, having Jenga blocks removed from the pyramid can have a disastrous effect on client confidence in the recommendations. This phenomenon is familiar to those of us who have ever been pitched a compelling startup opportunity where there is ONE glaring omission or logical inconsistency. Our interest in the company is damaged until that question gets put to bed.

MECE taxonomies are useful for all fields of macro problem solving. As the supporting data gets more micro, keyword-based taxonomies which allow for overlap can become more appropriate, just as larger datasets lend themselves better to non-relational vs SQL databases. While these non-relational thought processes can be useful for establishing hypotheses, they will never be a compelling way to express outcomes human to human.

Our position as stewards during the very beginning of a company’s journey requires that we maintain an objective and well-structured thought taxonomy – in understanding our landscapes, teams, and product development:

1. Organize your Dealflow by More than just Tags. I have the privilege of seeing a large number of investors’ dealflow pipelines. 90% of the time, companies are chucked in an “intake” folder and tagged with keywords about the business. Then, they’re moved from folder to folder as they get going along the path to funding or passing. What if you had an internal MECE taxonomy of all of the keywords you’ll use to describe targets? Every target would have the broad L1 keyword, a segmented L2 keyword, and any number of specific L3 keywords. You can still have “Machine Learning” and “Advanced Manufacturing,” they’re just respectively under “software” and “hardware.” Now, searching by keyword can pull up broad categories and you don’t need to build views using multiple tags; which is annoying enough that you’ll either stop doing it or stop adding new descriptive tags.
2. Overlapping Hierarchies Kill Accountability. Are you staring at an org chart that was designed to assuage egos? I’ll be there are all sorts of non-exclusive zones that require interfacing; not because it makes the business stronger, but because somebody needed to feel important. The Agile software development concept of “one person to choke” can apply everywhere in an organization. Make sure that key managers are properly resourced for success, and ALSO that they are solely responsible for the explosion or implosion of the portions of the business vision they control.
3. Segment Users Cleanly. Every business has users that span a variety of needs and demographics. Segment them with MECE parameters like age, income, or origination source. Building cohorts underpinned by non-MECE units may seem seductive (interests, platform usage, mixed metrics) but ultimately will present a muddy data picture that prevents useful analysis.