How business model innovation can be goosed: increase the likelihood of serendipity

I’m following up on the theme of innovation arising from periodically or systematically rearranging the workplace seating layout (see Embrace randomness). Here’s a snippet from Rachel Feintzeig in a NY Times piece that touched on workplace intermingling at a New York ad agency:

Rachel Feintzeig, The New Science of Who Sits Where at Work

MODCo Media, a New York advertising agency, has tested three different seating arrangements over the past few years. For about six months, the company intermingled its accountants and media buyers, hoping they would begin to absorb each others’ skills through “osmosis” and “overhearing phone calls.”

The experiment ended up saving MODCo “a couple hundred thousand dollars a year,” says CEO Erik Dochtermann, but it turned out badly for the accountants. The media buyers began to understand the financial side of the business so well that MODCo no longer needed a full accounting department. Now, the media buyers “do the accountancy on the fly” and the company’s chief financial officer checks their work, says Mr. Dochtermann.

From my perspective this is a great example of spontaneous desiloing of the business instigated by the chance interactions of the MODCo buyers when placed in proximity to the accountants. The transition of the necessary foundational knowledge about accounting took place without a plan, simply because the buyers were observing experts, and then began to copy them.

But Feintzeig and the MODCo CEO are using the wrong metric to measure the true impact of this business model innovation, which shouldn’t be money, but time. There is no doubt that the new approach — where the buyers record the accounting of their activities — means that the financials are getting closer to real-time. In the older model there was a lag capturing and communicating the financial data, then the effort expended by the accountants and various delays there, and of course some checking by senior finance people.

The likelihood is that the CEO would have been willing to spend serious bank to shorten that cycle, to buy new software to optimize it, etc. But, instead, the business model innovation came serendipitously, by changing the mix of people’s desks. And specifically people that were formerly housed in separate buildings or floors whose work was actually closely linked.

I’ve written recently about the premise of organizing the company’s operations around hiring and retaining the high performers, which Reid Hoffman at Netflix calls increasing the talent density (see What top performers do, and how to do it). This is a parallel idea, which is cross pollinating ideas and knowledge by rotating or randomizing seating, which increases coincidensity: increasing the likelihood of serendipidy.

And the apparent downside — the reduction in accounting staff — is only a downside to the accountants, not to anyone else. The shifting social contract in today’s accelerating economy means that lifetime employment is an antique premise, at best an empty promise. And note as well that the accountants may be just as likely to be picking up the skills of the buyers as the buyers did theirs.

Embrace randomness

In a post yesterday, I mentioned findings that show that when teamwork is mandated, random teams perform better than self-selected teams (see Voluntarily formed teams perform better than alternatives). It turns out that voluntarily-formed teams in the absence of a mandate turn out the best performance overall.

I started to think about randomness in the workplace. Are there other findings out there where randomness works better than deliberate decision making?

A group of Italian researchers (Alessandro Pluchino, Andrea Rapisarda and Cesare Garofalo) wanted to test the veracity of the Peter Principle, the maxim articulated by Laurence Peter that said,

Employees tend to rise to their level of incompetence.

In a hierarchic organization, then, some proportion of people have been promoted past their level of competence, and therefore are confronted by decisions beyond their judgment. This also means, mathematically, that while incompetence will be spread across the organization, it will tend to cluster farther up the pyramid.

The researcher modeled a six level, 160 person company, varying the degree to which the competency at a higher level in an organization was met by a subordinate being promoted into that position. In the ‘common sense’ transition the newly promoted remained competent in the new role with a slight decrease, while in the Peter Principle transition the individual’s competence at the lower level is unrelated to the needs of the new, higher one, and for the simulation the researchers randomly assigned a competence measure to the individual assuming that new role.

The results are intriguing, and show that promoting the ‘best’ people — based on their competence in their existing job — only works as a good strategy when the common sense premise holds. Said differently, promoting the best person only works when the competencies required in the new role are tightly related to the skills in the former. In other cases, the Peter Principle means that promoting the best leads to the growth of incompetence. And in that latter situation, promoting the worst player works better.

The most important finding is that adopting a random promotion strategy works best when it is unknown which of the two hypotheses are operational. In other words, if you can’t characterize the closure between the competencies in the existing job and the new position. For this amazing finding the researchers received the Ig-Noble Prize in Management. They went on to speculate and study the possibilities of random selection of members of parliamentary governments, and for US citizens that idea sounds awfully attractive considering the past few months.

Looking at it from another way, the world is changing so quickly, and the future is increasingly uncertain. How can we know — with certainty — what sort of character, skills, and background are most likely to make someone successful in a new job? It may not be the profile of the last person to hold that position, since the company’s and individual’s network of connections is constantly changing, and a person’s network is a major source of professional success (see What top performers do, and how to do it).

Leaving aside the Italians’ research, which can be dismissed by anyone who argues that it is based on simulation and not the real world, I still think that there is merit in permitting randomness in the workplace. Perhaps upsetting the long-established practices of hiring and promotion by randomness is too much of a stretch, but the effectiveness of voluntarily formed — and self-managed — teams has been demonstrated empirically, not through simulation. So, one practice might be to replace the conventional model to team formation — hire a team leader, allow team leaders and others to pick team members from a pool of candidates — which certainly skirts the edges of the Peter Principle, and leads to the known issues of mandated, self-selected teams. Best is to allow teams to form voluntarily.

Here randomness might play a role, again. I recalled that Hubspot, a Boston-based inbound marketing firm, randomly reassigns the desks of workers every quarter. This leads to new patterns of interaction, new connections, and new teaming relationships, I bet. They do support trading, so the pathological stick-in-the-muds can avoid moving, but overall the workforce is redistributed — and old patterns are disrupted — every few months. So, this increases the Brownian motion in the business, so that when people get around to teaming on a new project or initiative they will have met and worked near a wide and diverse set of people. Their choices are greater, and then the results are more likely to be wiser.

My bet is that businesses would improve if everywhere there isn’t a strong reason to do otherwise randomness were used to (dis)organize things. Seating? Randomize. Need a representative from each group for some initiative? Select randomly. Need notes taken at a meeting? Draw straws. Three smart and eager candidates for a new product marketing role? Embrace randomness, and roll the dice.