Pattern Recognition for Management


Simon Tegg 

Middle managers have it pretty rough. Pop culture knows them as Dilbert’s pointy-haired boss or the insufferable David Brent from The Office, but their job is essentially to solve unsolvable problems.

Before Loomio I was occasionally thrust into positions of managerial authority over my fellow slackers. A queasy feeling would come over me when I realised that I would constantly have to infringe subordinate’s autonomy by trading off with business objectives. I would quickly downshift back to my previous minion role. In Venkat Rao’s terms I was a Loser -that is, someone who loses the economic status game by making the strategic, rational decision to reduce stress levels and slack off rather than ‘climb the ladder’.

Since I started at Loomio things are much more complex. Everyone is the Boss. It just depends on their domain-specific experience and track-record, for how much of the Boss they are at any given time. Again I’ve found my self with some authority, or at least strong opinions on how particular things should be done, but with a personal investment in Loomio’s success I can’t shy away from messy problems. I’ve also had my fair share of management failures and learnings from my year in a fluid, startup, co-op, social enterprise.

To help think about the problem of management I appropriated the Competing Values Framework and remixed it, with inspiration and heavy borrowing from Fred Kofman’s and Venkat Rao’s recent posts about the impossibility of management. The result was the following 2×2, which I’ve found useful for deciding which management pattern to apply to particular work contexts.

Before we dive into the patterns, to navigate the 2×2 you need a grasp of three key variables of work and business: motivation, uncertainty, and coupling.


Management is still coming out of the shadow of Taylorism. For the Taylorist, productivity is a simple matter of providing the right incentives for performing tasks in the most efficient manner. Incentives can be either positive, e.g a bonus, or negative, e.g. a reprimand or the threat of being fired.

Kofman splits incentives into individual-level outcomes or team-based outcomes. When managers incentivise individual-level outcomes (track individual sales targets, lines of code written etc), it’s easy to spot the shirkers and low-performers and coach them or move them on, while at the same time attracting top performers by awarding bonuses. Unfortunately, collaboration and overall performance suffer as individuals focus on their own targets at the expense of helping others.

With team-based performance management, the team can easily share tasks and risks more efficiently by buffering each other. The downside is that individual staff have the incentive to skive off without being noticed by the boss, and arguably star individual performers aren’t attracted to a work environment where the team will take the credit. Kofman has a talk about this tradeoff here:

This is a useful insight, but if you’ve been paying any attention at all, you’ll know that there’s much more to motivation than rewards and punishments.

Dan Pink has popularised intrinsic motivation with his book Drive. In Pink’s model (get this), we can perform tasks for their innate joy, and are happier and produce higher quality work when three conditions are satisfied:

  • Autonomy – the task performer is free to choose the task.

  • Mastery – the task is neither too hard nor too easy.

  • Purpose/Relatedness – the task connects its performer to some greater mission and provides a supportive social environment.

Most workplaces are hierarchies where the boss chooses what you’ll be working on, you rise to your level of incompetence (the Peter Principle), and you’re striving to make the shareholders richer, so satisfying all three conditions is pretty rare.

To be fair though there’s more of a continuum between extrinsic and intrinsic in the underlying research. Its possible to leverage quasi-intrinsic motivation by indoctrinating your employees with a feel-good corporate mission, team-building exercises, and granting degrees of autonomy etc. But even clever tech companies like Google struggle to match tasks to the appropriate skill level.


The second variable is the degree and quality of uncertainty that a venture currently accepts. According to Rao there are three uncertainties relevant here:

  • When – are you trying to minimise time uncertainty (i.e. a hard deadline),

  • Who – role uncertainty (who will be doing what),

  • What – output uncertainty (you need a very specific thing)

In a mass-production factory it’s possible to reduce all three uncertainties to levels where next year’s production forecasts are meaningful. Management can be a colour-by-numbers activity.

But for a startup or creative service venture, like the ‘Fast, Good and Cheap’ adage it’s impossible to have all three simultaneously. Accordingly, Rao defines talent management as: “dynamically managing the returns on human resource investments by exploiting the coupled uncertainties involved in who, what and when decisions.”

In creative industries talent management becomes an artform. The skilled manager must make constant tradeoffs among business uncertainties and between these uncertainties and workers’ motivation and wellbeing.


The third variable, coupling, is a constraint. Different types of work tend to require different degrees of coupling between workers, and workers will generally have their own preferences.

Software development often requires highly coupled ‘pair programming’, code reviews, and QA to ensure large projects are maintainable long-term, while a sales rep can operate independently of the mothership for long periods.

Simply put, highly coupled (functional) teams will help satisfy social needs but often reduce autonomy, while loosely coupled work can operate with minimal managerial overhead and satisfy autonomy, possibly trading off with social needs.

The Patterns

Management patterns for worker run businesses

To make our 2×2 we’ll split off the two factors that we have the least control over, coupling and outcome uncertainty and see how the other parts of the puzzle intersect.

Coordinate: Explicit Outcomes + Tight Coupling

The coordinate pattern should be familiar – it’s textbook project management. You require explicit outputs, and to achieve these you have a team of specialists, each working on sub-projects with dependencies.

For our team here at Loomio, this pattern crops up whenever we interface with a government bureaucracy. Bureaucracies are notorious for requiring explicit outcomes (“can we get this stack of forms filled out in triplicate, please”), often due to well-intentioned regulations to ensure accountability. These tend to need specialist knowledge like legal, accounting etc, and different parts of these projects will have to be arranged in series, rather than in parallel (“Once legal have signed off, we can review, then send it off to accounts payable”).

You’ll probably need someone with an overview of the entire process (the project manager) to keep things ticking along, who, in a traditional company, will perform this role on an ongoing basis. In Loomio flatland, these roles are temporary (I like to think of this as “temporary hierarchical zones” in a sea of collaboration).

Inevitably, there will be time and cost overruns. Individual specialists and high dependencies means the system is fragile to any slip. Due to serial dependencies, a slip in one specialist’s workflow ripples through the project as lengthening delays as everyone falls out of sync.

Hierarchies use the coordinate pattern by default. ‘Who’, and ‘What’ uncertainty are minimised at the expense of ‘When’ uncertainty blowing out. While the coordinate pattern is often necessary, getting stuck in it tends to burn out workers. The work tends to be routine, high pressure, low autonomy, and extrinsically motivated by the threat of missed targets (which typically happens).

Collaborate: Flexible outcomes + Tight Coupling

Moving across to the collaborate pattern, we now have a tightly coupled cross-functional team working towards flexible outputs. Devs know this as Agile Software Development. Here the desired outcomes are penciled in and the team is free to implement as it sees fit, using some highly structured processes, within which the details are emergent.

For teams that adopt sprints and report some measure of value, the ‘When’ uncertainty is reduced, at the expense of ‘Who’ uncertainty, while ‘What’ uncertainty slips as the implementation details emerge on the fly. We know that at least some value will be delivered on a continuous basis, but with all that collaboration there are no legible metrics on who did what, reducing opportunities for managerial control.

The lean startup crowd love continuous value delivery. When the product evolves rapidly the likelihood of finding market fit goes up. Further, working in a tight, collaborative team, with a degree of autonomy is going to strike a decent balance between social needs, task-skill match and autonomy.

Compete: Explicit outcomes + Loose Coupling

The compete pattern usually applies to the work of the sales and marketing teams. They get some explicit outcomes alright: “make it rain”. But this doesn’t need to be tightly coupled with everyone else. Sales seems to be one area where the commissions and performance bonuses may work, and seems to suit the goal-focused type (what Dan Pink calls Type-X).

Personally, I can only go about 2-3 weeks on 100% Compete mode before burning out. Whereas I can probably sustain a 7030 Compete/Create split for 2-3 months.

Create: Flexible Outcomes + Loose Coupling

Finally, we have loose coupling and flexible outcomes. For a traditional organisation the create pattern carries far too much risk. You just let workers just go off and create stuff? On company dime? ‘What’ and ‘When’ uncertainty shoot up, you only know ‘Who’ is not available to do all that other urgent work.

Yet only the create pattern offers optimal conditions for the innovation that your organisation desperately craves. You’ve got autonomy, the worker can self-select an optimally challenging task, management just has to supply some social support and a higher purpose worth achieving.

Companies like Atlassian, Google, and 3M have leveraged the create pattern with their famous 20% time. Though it seems in Google’s case this often more to do with talented bullheaded employees than official policy. Apparently, Gmail, Adsense, Autocomplete and Google News originally started in 20% time, so that worked out okay.

The balanced portfolio of management patterns

Try thinking of these patterns as a balanced portfolio. For us, Collaborate provides baseload returns through continuous value delivery and reasonable levels of worker happiness. Coordinate and Compete are tactical moves that depend on the terrain, and Create is high-risk play with high potential returns.

For example, Jon shifted from his responsibilities as coordinator of the development team and went into create mode. We got a major boost in Jon happiness, a comprehensive set of mobile mockups, an entire new UI for Loomio 1.0.

At the personal level, you may want to consciously adopt a split and consider a deloading phase. I’m using weightlifting terms here, but the same concept applies to mental and emotional work. Weightlifters recognise how spending too much time in the same pattern can lead to burnout or stagnation. My personal preference is a 4060 create/collaborate split. I’ll shift into this split after heavy episodes of coordinate and compete to rejuvenate.

I’ve found that many organisational problems (whether that’s a project failure or unhappy workers) stem from choosing the wrong pattern for a given scenario. A critical ingredient to this is knowing your position on the map and knowing when it’s time to move. The pattern needs to be chosen intentionally, rather than by default, and reconsidered frequently.

Thanks to Richard Bartlett, Alanna Krause, and Matthew Bartlett for their thoughts, editing and graphic design help.