Why Companies Make Slow, Weird Decisions Instead of Fast, Good Ones: Demand Signaling in Complex Systems

We don’t talk enough about demand signaling in complex, open, dynamic, emergent, coherent networks and how it causes slow, weird decision-making.

I’ll bet nobody has ever said that to you before. But, hey, we’ve already established that as a management consultant, I’m a little, uh, different. And I'm sticking to my story. I just don’t think we talk enough about demand signaling in complex, open, dynamic, emergent, coherent networks.

Sorry, I’ve said it twice now. I’d better explain before I accidentally say it again.

Let's start with the networks. Your organization is the kind of network I’m describing. It’s complex, which is self-explanatory. It’s an open system and not a closed one: new people, information, processes, and resources can enter at multiple boundaries and in multiple ways. It’s dynamic, constantly changing its own design and action as time passes. It’s emergent, because the constant redesign is a function of the growth of the network and of the needs, challenges, and opportunities it detects. Finally, it’s coherent – internally coherent, we might say – because it contains its own execution-oriented feedback loops: if you stripped away all outside stimulus, market forces, etc., the system might or might not ultimately cease to exist, but it wouldn’t instantaneously disappear or terminate major functions on the spot.

So, yeah, your organization is a… sheesh, it’s too long to keep typing. Let’s call it a CODEC network.  By the way, other CODEC networks include your local computer network, and your own neurology. Hopefully it’s obvious which of those is more complex.

Speaking of complex, a brief tangent: “Codec” is a real word describing a signal converting system such as the digital audio codec in a laptop – which receives sound (information) from the microphone, codes it into digital storage format (internalization), and then decodes it back to audio for playback through the speakers (execution). That word, “codec,” is a portmanteau of coder/decoder; not an acronym. So, it would be correct (and easy) to say that I’m using the acronym CODEC differently. It would also be correct (and more brain-stretchy) to say the organization is also a people/process system that receives information from the external market and environment, codes it into a format that can be processed and reacted to internally, and then decodes it back to the language of the marketplace and environment for execution. Take your pick. Either way, I would like credit for being the only management consultant you know who writes choose-your-own-adventure elements into his blog posts. 

Anyhoo, end of tangent. Your organization is a CODEC network. (Once more so I don’t forget: it's Complex, Open, Dynamic, Emergent, and Coherent.) So, let's talk about demand signaling in CODEC networks and how it causes weird decisions.

What’s demand signaling? Don’t Google it, it won’t help. Fine, Google it if you want to. You’ll get a bunch of different technical things related to business and supply chain and retail, and something from Gartner that will simultaneously sound brilliant, give you a headache, and feel like job security for a lot of people, forever. They’re good that way. But that’s not the demand signaling I’m talking about.

I’m talking about demand signaling inside the internally coherent network. I'm talking about how the network tells itself when it needs something from itself. Like, Subsystem A of the network needs a resource or action that Subsystem B controls, and it’s good for the whole system if the request is relayed and considered. We don’t talk enough about how that works in organizations, or how it can cause problems.

[[RELATED: Prevent A Reptile Dysfunction: Work WITH, Not Against, the Lizard Brains on Your Staff]]

Sometimes, demand signaling works really well. Like with paychecks. Every pay cycle, the part of the system that understands who the employees are issues a request to the part of the system that holds the dollars to distribute some funds. And while there can certainly be glitches and challenges, payroll systems generally function effectively: Companies routinely pay agreed-upon amounts, at agreed-upon times, to people playing agreed-upon roles, and not to people who stopped playing those roles or haven’t started yet. Payroll demand signals tend to be clean and work well.

Sometimes, demand signaling works sort of well. Think about hiring requisitions. The part of the system that manages a certain area detects that it needs more help to adequately perform its oversight function. The part of the system that allocates limited headcount to the highest-priority needs receives the demand signal, in this case a headcount request, and makes an informed decision about whether to authorize it. There's some noise and posturing, and in most companies the process makes mostly decent but definitely not perfect decisions. Of course, not all demand signals get filled, but that in itself isn't a flaw. (I know it doesn’t seem OK to the overworked manager whose requisition gets denied, but from a systems view this is still fine: “No” is not automatically the wrong answer.)

And sometimes, demand signaling works poorly. To pick yet again on a perennially-picked-on organizational failure, when one part of the system launching the Space Shuttle Challenger in 1986 understood a safety issue with O-ring performance at low temperatures, it sent a demand signal to another part of the system in charge of launch execution, that the launch should be delayed. That signal was not relayed completely nor received fully, so it was never decided upon appropriately, and we all know the tragic consequences. (Real, fully informed decisions by the appropriate signal recipient are crucial: “No” is not automatically the right answer, either.)

The important question I’m pointing at here is a deceptively simple one: how good is your CODEC system/organization at generating, relaying, and attending to its internal demand signals? There’s a lot to unpack in the answer. First, does the system generate demand signals appropriately, too frequently, or too infrequently? Second, does the system generate signals which are clear, somewhat clear, or unclear? From there, we must ask whether the demand signals are relayed efficiently or inefficiently to the appropriate receivers, whether those receivers habitually attend to or ignore them, and the extent to which the receiving subsystems are configured to act upon them or not. And the answers to these questions are qualitative, ever-changing, and overlapping.

That’s a mess, which may be why we don’t talk much about it. So let’s just paint a good scenario and a bad one and contrast them.

[[RELATED: Changing Culture Isn’t Easy, but it’s Not THAT Complicated]]

Here’s the good one: Imagine a system with a generally consistent language and processes for demand signaling across its subsystems, and well-formed pathways for the signals to travel which include intermediaries who can translate between languages as needed. A need detected in one subsystem can quickly be analyzed to determine whether it’s advantageous to the overall system to generate a demand signal. If it is, the signal can be created easily. It will be relayed quickly and effectively, received completely, and dispositioned appropriately by a receiving subsystem willing and able to receive it, and able to act when necessary.

Now the bad one: Imagine a system in which different subsystems speak entirely different languages, working in isolation with no intermediaries translating. The act of making a demand signal is difficult and likely to fail. A need detected in one subsystem will be analyzed not in terms of whether creating a demand signal is advantageous to the overall system, but whether the benefit to the local subsystem of doing so is worth the pain and chance of failure (think: “spending social capital”). If it is, the creation of the signal will be an arduous process designed to produce the most compelling case possible (think: “creating backup slides”). The path for relaying it will need to be custom-built as a campaign of influencing and favor-trading, the message will get garbled in transmission, and it's likely the response by the change-averse receiving subsystem will be inappropriate and confusing (think: “they decided what?! Why?!”).

Some of that type of slow, weird decision making might be tolerable, I guess, except that while good stays good, bad tends to get worse. The decline starts with what happens on the back end of the bad scenario, when the sending subsystem conducts its post-mortem.

Hold on. You know it will perform one, right? One hundred percent definite, there will be a post-mortem on any problematic response to a demand signal. It might happen in a conference room where people gather to try and reconstruct what went wrong. And that would be troubling enough – imagine asking your left hand to guess why your right foot or your brain did something it didn’t expect or like.  More likely, it will happen informally, in hallway conversations and happy hour gripe-sessions. But it will happen. When it does, the conclusions will almost always include some combination of “next time let’s not bother” and “next time let’s make our case even more compelling.”  That's a combination of less signal and more noise – less signal when appropriate demand signals are not sent because doing so feels hopeless, and more noise when demand signals are inflated into giant slide decks designed to convince, rather than information-rich requests for action. The whole system loses out both ways, and the organization takes another go-around of the vicious cycle, as its decisions continue to get slower and weirder.

I’ve spent most of my career on the real root causes of demand signal failure, and how to help organizational systems become more adept at receiving and dispositioning demand signals in real time. And I can tell you that the organizations that do this well consistently outperform those who don’t. But before I can even begin a conversation with leadership about practices to correct the issue [video], they must be able to recognize the problem. Too often, they’re too deeply embroiled in the high-noise, low-signal mess and in the chaos and overwhelming workload it creates. For those working in this mode for too long, what is actually a crucial failure in the function of the CODEC system starts to look like just another day at the office. Leaders can't solve a problem they can't see.

Which is why I say we don’t talk enough about demand signaling in complex, open, dynamic, emergent, coherent systems.

Whoops. I said it again. Let's hope the repetition will help to create some new awareness, so that we can stop the cycle of slow weirdness.

Want more? Visit the Group Harmonics Industry Intelligence Archive for ideas, whitepapers, and case studies about management culture and how it impacts so many facets of the organization.