The Silo Reconsidered

The silo is an ambiguous unit of organizational structure that is a favored strawman for management consultants. Admit it: you’ve probably ranted about silos at some point in your working life. The connotations of the word today are nearly universally pejorative. I have only ever heard the term used to refer to inefficiency, creeping bureaucracy, personal fiefdoms and poor communication. When the silo in question is also a locus of technical competency, it is also the target of accusations of narrow-minded scholasticism (“tunnel vision” is the phrase that appears here). This notion of silo is common enough that there are actually leadership courses out there promising to eliminate the silo mentality from your organization.

As you will see, this is not necessarily a good thing. In fact, it can be a terrible thing.

If you buck conventional wisdom and recognize the true nature and value of silos (while acknowledging the pathologies they can manifest) you have an advantage in navigating corporate ecosystems. I will focus this discussion on private enterprises, but much of what I say applies, mutatis mutandis, to nonprofit and government organizations as well.

So what exactly is a silo, besides a compelling Cold War era metaphor? Can it possibly be as terrible an entity as some believe? Is it an inevitable feature of organizations? If yes, can notions of process-re-engineering, six sigma, teams, service-oriented architecture, and the like, erode its influence? Is that a good thing?

I am going to argue for counter-intuitive answers to all these things and assert that silos are generally good things. That some of the accusations against silos are unfounded, and others valid but true only of a high-visibility minority of dysfunctional silos.

Getting at the Concept

Let’s set up a reasonable definition to analyze. Here is the informal definition that I will defend:

A silo is an organizational unit that has established itself in a high-influence core position within the value network of a specific business model, by reliably and efficiently adding clearly-defined and measurable value to any appropriate embedding business process, and offers essential services that satisfy demand from more than one mission critical process.

Here is a picture that illustrates this definition. In this picture, function B is a strong silo across three processes, and function G is a weaker silo across two processes. D is an example of an un-silo, something I’ll get to later.

I will justify this definition as I proceed, but before you jump in to attack it, let me assure you that it leaves room for silos to manifest every dysfunction you care to attribute to them. For now, think of a silo as any organizational unit that has found its sweet spot in the information ecosystem within which it is embedded. The key to the definition is that it is a relative one and hinges on the silo being supply limited in the internal economy of a corporation.

By this definition, a silo can be any sort of organizational unit, from a division down to a single individual: something like a traditional department can be a silo, but also something like a long-running team in a matrix organization that is so successful it is never disbanded. I will ignore here the reductive idea that silos are actually about incompatible computer systems — that may well be a symptom of the existence of a silo (and not always worth fixing with XML), but a silo is fundamentally a human entity.

The definition also determines a complementary notion of the un-silo. An un-silo is an organizational unit that, simply put, is not in a comfortable value-adding position in the organization’s processes. This lack of comfort can arise from two causes: the un-silo has unsecure positions in high-value or critical processes, or it has a secure positions in processes that have not yet demonstrated a reliable value-addition capability. An un-silo can arise in many ways — it can be a nascent growth area, it can be a cohort of new employees bringing on board new skills or it can be a former silo that has been dislodged from its position of comfort due to fundamental changes in the business model. The last phrase is important — an organizational structure that can be dislodged from its position of comfort by anything less fundamental than a change in business model is NOT a silo.

Some examples:

• A technical competency in a large company that solves mathematical optimization problems, in a business model where suboptimal solutions can lead to unacceptable financial performance
• A district sales office that comfortably outperforms its local competitor and delivers above average results compared to other districts with similar profiles.
• An analyst at a large company who develops a track record for accurately valuing small companies for acquisitions.
• A chemistry research lab that develops extraordinary skill in working with a particular class of materials and experimental procedures.
• A product engineering unit that has refined the craft of delivering manufacturing cost-downs over the life cycle of a product.

Silos can be very large and recursively contain silos within themselves. The entire manufacturing organization may be in a silo-like relationship with respect to its ecosystem neighbors (comprising, say, marketing, sales, engineering and senior management divisions).

Based on the definition above, what should we infer about silos? We can expect the following attributes, none of which is surprising. If you are used to thinking about silos in the traditional way, you will read each of these attributes in a negative way. But mull each more carefully and note that none of these is a priori bad.

Attributes of Silos

1. A silo will usually manifest a hard-to-replace knowledge and skills base that is also locally of high value. This follows from its ability to maintain a high-influence core position.
2. A silo will optimize its internal functioning to most efficiently add value to the most critical and high value embedding processes, by minimizing non-core aspects of its functioning. This is the internal analog of a company focusing on core competencies within an industry. It is able to do this because its knowledge base is in high demand. This is the characteristic that is the basis of the “silos optimize locally” accusation. This accusation is naive in its use of the metaphor of optimization. That is a whole other post however.
3. It will develop an internal language — this is perhaps the signature feature. This follows from the fact that a silo’s services are in high demand, which permits it to accept or prioritize requests that it can satisfy with the least amount of lateral learning, and with customer processes that are the most willing to adapt to its demands in order to obtain services. This is actually efficient. The internal economics dictate that the partner currently adding less value in a value-chain link — say A, F, H, C and D, must do more of the work to enable the transaction. Taking on translation tasks (either to natural language or other silo languages) is not a good use of a silo’s time under high demand conditions. Other parts of the organization must take on these tasks.
4. If two silos are adjacent to each other in a value chain, as B and G are in the picture above, the current weaker partner must manage the communication. In times of tight budgets for instance, a research division must talk the language of the business unit. In times of rapid growth, the business unit must learn the language of the research lab, since it has a surplus cash situation that it must quickly convert to investments. Creating a universal language for all times and silos is a hopeless task.
5. As long as the underlying business model maintains an ecosystem that generates high, critical demand, silos persist robustly. Small or large changes to individual embedding processes do not affect them because they respond to an internal supply-limited market of many processes. No single-process improvement champion can make a compelling argument for internal change to a silo without understanding the complete demand profile for the silo from all it’s client processes.
6. Even large changes to many internal processes are unlikely to affect silos because they provide a critical black-box capability to their embedding processes. The criticality makes them immune to elimination during redesigns. Their black-box nature, a function of their manifesting deep expertise, makes them immune to non-expert suggestions for internal change.
7. Even in hard times, silos are peculiarly immune to resource cutbacks, layoffs and other unpleasant economies. This is because the way a silo’s internal expertise is organized and deployed is so mysterious to non-initiates, and so critical to financial health, that it can protect its resources far longer than peripheral or low-demand units of the organization. A sign of a true silo is that it survives intact during a reorganization. Recall from point 3 that this mysterious language is not necessarily a pathology.
8. Silos are resistant to replacement through new hiring since they represent skills that can only be learned through non-trivial apprenticeship periods, during which mission-criticality can foster a non-democratic culture (again, not necessarily a bad thing).
9. Silos that represent commoditized functions across many industries (such as payroll) are susceptible to outsourcing but in a broader sense, still represent positional influence since their functions must be conserved. This stability in the face of outsourcing is what supports trade, professional and academic associations around commoditized silos.
10. Silos are the atomic units that survive structurally intact during mergers, acquisitions and outsourcing deals. A unique silo (say the chip design engineering team of a semiconductor startup that is absorbed by a larger company) will retain both its structure and the individuals filling roles within it. A commoditized silo — payroll say — will persist structurally even if individuals lose their jobs. The disciplinary bodies associated with the silo will help displaced workers find positions faster than displaced workers in non-silo roles.

Can we summarize these attributes of silos within a simple explanation? Yes we can. Silos are the manifestation of the competitive advantage of a business model and constitute it’s DNA. Their inertia along many unimportant vectors is what makes for extraordinary responsiveness and efficiency along a few critical vectors. Their strange internal languages are highly optimized to express and work with the knowledge they operationalize, and the resultant communication problems are a visible cost that must be paid by less valuable parts of the value chain.

To this unsurprising list of attributes, each of which can be validated through countless examples, let me add one falsifiable prediction about silos that meet my definition: the only way to externally influence a silo to change is to change the entire business model. This prediction, I hope, makes this definition the basis for a theory of silos. Process re-engineering efforts should not have significant and non-cosmetic impact on true silos.

Why silo inertia can be a good thing

Once we learn to appreciate the value of (healthy) silos, we will understand that this inertia to everything smaller than business model change is actually a good thing. The only good reason to force change externally onto a silo is a looming disruption threat to the entire business model. Why is this the case? The answer lies in the criticality of a silo. The value of a mission-critical component of a mission critical process is the value of the entire business. This may seem counterintuitive, since we are taught in operations research and management that processes are mechanisms of “progressive value addition.” This is true from a normal functioning viewpoint. But when considering changes to a critical process component, we must estimate its shutdown cost due to failed change efforts.

Here is where the DNA metaphor comes in handy: just as the the selfish gene rather than species survival drives biological evolution, it is the selfish silo rather than corporate profitability/survival that drives economic evolution. No companies survive today from the trade guilds of the medieval world. The age-old function of accounting survives though, and is part of the DNA of modern corporations.

So we’ve figured out a few things about silos. Before we get to exploring the pathologies of dysfunctional silos, let me list the signs of a properly functioning one — the ones that do their jobs so efficiently and quietly that we tend to forget they are the reason employees get paychecks.

Characteristics of a Healthy Silo

A healthy silo:

1. Uses its high-demand position to prioritize requests that maximize value to the company and the vigor of it’s overall business model.
2. Consciously and fairly interprets its unfathomable internal needs for the benefit of larger resource allocation discussions and makes reasonable demands.
3. Cultivates a community of practice around itself by creating processes of knowledge capture, apprenticeship, fairness in dealing with requests, and measuring its own performance in ways understandable outside itself.
4. Recognizes its “shutdown criticality” in core processes and consciously adopts sufficiently cautious and resistant attitudes towards well-intentioned but potentially disruptive attempts to change its ways, usually emanating from efforts to optimize an individual client process. This is a critical flaw in all process-focused improvement paradigms. No process is the sole user of any silo function. This resistance is framed in firm but reasonable, transparent and honest ways and in association with a limited willingness to listen (the limit being determined by the amount of distraction it can handle without disrupting core activities).
5. Consciously monitors the overall business model and attempts to adapt in concert with it, and initiates a repositioning dialogue early with all stakeholder processes and senior management when it anticipates a change that is coming faster than it can adapt.
6. Anticipates and manages it’s own dismantlement in a graceful way by providing resources for its members to retrain and reposition themselves, and gradually scaling back with the declining demand profile for its services.
7. Leverages its position of high comfort and being in an operational “sweet spot” to contribute unselfishly and disinterestedly (though not un-interestedly) to broader discourses, where other parties usually have hunger-driven agendas to further.
8. Actively mentors and provides a model of efficient operations and healthy exercise of influence to other emerging areas that are not yet established in comfort zones (proto silos). In a related role, it provides an environment of healthy skepticism that untested organizational forms and value-adding activities must survive.
9. Delivers its products in a form that adds the most value to a reasoned mix of the overall company’s goals, customer process goals and its own goals for its continued survival and health.
10. Retains control over value it delivers only to the extent that stewardship of the value requires its expertise, otherwise allowing value to migrate to the optimum locus in the organization. Does not attempt to use control over valuable assets as an unreasonable bargaining chip

Characteristics of a Dysfunctional Silo

1. Uses its high demand position to cancerously increase its own resources with no regard to the value to the company or the continued health of the business model.
2. Uses the incomprehensibility of its internal structure and the criticality of its operations to effectively blackmail the rest of the corporation into providing more resources than it needs, and diverts the extra resources towards increasing its technical depth far beyond the point of diminishing returns, and to better integrate itself into its extra-organizational disciplinary roots.
3. Cultivates a community of “black magic” around itself by indoctrinating apprentices rather than teaching them to think critically, trades in favors and politics when dealing with requests, and is opaque in measuring its own performance in real terms. Conducts itself internally with undemocratic high-priest driven decision-making (hence the common “fiefdom” pathology).
4. Uses its “shutdown criticality” and operational opacity to maintain an unyielding inertia in the face of legitimate and manageable requests for change.
5. Consciously monitors the overall business model and attempts to resist change that would disrupt its functioning, so long as it can do so within its own planning horizon (which might simply be the time required for key silo priests to find more lucrative positions).
6. Anticipates and manges its own dismantlement in a selfish way by buying as much time for its own survival as it can, even at the expense of making the company unprofitable.
7. Leverages its position of high comfort to bully, dominate, trivialize and otherwise render ineffective the efforts of less comfortable organizational units to gain traction. Engages actively in ideologically-framed adversarial discourses that bolster it’s own model of value addition while undermining alternate models. Delegitimizes external data by fostering an NIH (not-invented-here) culture and prevents internal data gathering by blocking efforts to pilot ideas.
8. Delivers its products in a form which lock-in value-chain partners to its services and maximizes its local value accumulation.
9. Retains control over all value it delivers in a form useful for repositioning bargaining (in an extreme case, control over its own fate when a company goes under, though I have no obvious examples of this).
10. Becomes judge jury and executioner by controlling all individuals that can evaluate its operations.

Even an unhealthy silo is not always the product of manifest ill-intention. Often extreme demand, expected effects of technical focus, and other attributes that make for a silo’s success, can lead it into unhealthy patterns. These can be corrected within limits. When pathology is the result of ill-intention though, the corporation is being held victim by a blackmailer pure and simple, and must deal with it the way one deals with any blackmailer — with no sympathy at all, and using any method necessary to extricate oneself from the blackmailed position.
And don’t forget, healthy silos must be encouraged and rewarded for their stewardship of the vital air-supply of any company, it’s cash flow.

So how do you promote healthy silos, and identify and correct or destroy dysfunctional ones? How do you prevent change processes from running amok and disrupting a business model by attacking healthy and unhealthy silos uncritically? Those are topics for future blogs. In the meantime, It’d be great to hear about examples of healthy and unhealthy silos from you.

Endnote: Is this essay just an exercise in semantics? Can’t we just rename healthy silos something else, ignore the abstract notion of silo, and be done with it? Here is the justification. The term silo is useful for its metaphoric structure. It is one of many “tall cylinder” metaphoric frames. Per a reasonable theory of a colleague of mine, “silo” probably comes to us from grain silos via missile silos, whose crews were isolated from the world for long periods of time. Ivory towers, “academic smokestacks” (a term due to Bjarne Stroustroup) and frogs in wells are other examples of the “tall cylinder” metaphor. The organizational structures that map to this metaphoric frame form a general class that includes both functional and dysfunctional examples. To drop the term silo for the healthy instances would be to lose the value of the metaphor. To let it remain at its current reductive definition would be a loss to our conceptual vocabulary. Enough said.