FeedbackLoop-v2.jpg

We live in an accelerating world.

Information is being created, disseminated, processed, and regurgitated at unprecedented rates. But our brains aren’t adapting to this breakneck pace; as the amount of things increases, our attention spans have been decreasing to protect ourselves from the incessant barrage of information.

And the world is changing to compensate for our lack of focus. Facebook and Twitter deliver endless feeds of bite-sized content, bombarding brains with dopamine — Burberry’s releases fresh designs one day after showcasing them — and the twin artforms of emoji and memes distill our thoughts into images. The best organizations understand our fascination with the new: Amazon, Google, Facebook — these companies feed our need for instant gratification. They do this by shortening the feedback loops that drive their businesses so they can bring us more of what we want. And it shouldn’t be surprising that the fastest companies are tech companies; software development is iterative and adaptive by nature.

But here’s the rub: every company is in tech now. It’s inevitable. Code has infiltrated every industry, and modern companies are now faced with a Darwinian dilemma: adapt or die. This isn’t a new metaphor, of course; Thomas Huxley was already talking about “Darwinism” in the 1860s and, soon after, people started applying the term to non-biological processes like capitalism.

But what is new is the rate of change; the pace of the game; the speed of iteration. It’s not about how quickly you make specific changes; it’s about how quickly you can equip your company to deal with constant change. Your company’s vitality is a direct reflection of its velocity. And velocity is a result of tight feedback loops across your organization.

Identifying and shortening your organization’s feedback loops is the most effective way to keep your business reactive. It’s simple to shorten feedback loops when they’re confined to a single engineer on a single computer. But at scale, this becomes much more challenging — and much more important. At its heart, the DevOps movement is committed to reducing the feedback loop between traditional developers and operation engineers.

What Is a Feedback Loop?

There are two components of a feedback loop: an action and a reaction. What makes it a “loop” is when the results of the reaction feed back into the process. There may also be several steps between the action and the reaction but, for simplicity’s sake, we’ll think of these as part of the original action.

When we’re identifying feedback loops, there are two things we should care about: 1) what the actions and reactions are, and 2) the cycle time of the entire feedback loop. Cycle time is comprised of:

  • choosing an action
  • performing the action
  • waiting for a reaction
  • measuring the reaction

Diagram-01.jpg

Before we discuss how to locate feedback loops in your organization, let’s look at some concrete examples to see how they really work.

Biology Loves Feedback Loops

Homeostasis is a biological system’s tendency to maintain a set of conditions by continuously adapting to internal and external changes.

Reflexes, too, are extremely short (and finite) feedback loops. When you touch a hot pan (action), the heat triggers sensory neurons in your skin. These travel as electrical signals to your spinal cord, where they’re processed by interneurons (reaction). These interneurons relay a signal to motor neurons in your hand, stimulating muscles that cause you to snatch your hand away (action). When you snatch your hand away, the sensory neurons stop screaming, and the interneurons stop telling your hand’s motor neurons to move. This is why your hand will only jerk back a short distance.

All of this happens without hitting your conscious mind — because that would take too long. If you are on fire, your body wants to exit that situation as quickly as possible. No need to run the decision by the bureaucracy of the brain; that would just increase the loop’s cycle time.

This isn’t unique to humans; it’s true of anything that can reproduce. All life rests on the backs of busy little feedback loops, adapting their hosts to external input and optimizing them for replication.

If we zoom out, we can see giant feedback loops operating across species: over millions of years, billions of organisms have changed in response to external stimuli. Some lifeforms, like bacteria, have shorter feedback loops than others; this is why antibiotic resistance is increasingly concerning. Or the flu virus, which reinvents itself every year like Burberry’s.

As humans, we tend to think we’re the best at everything. But our reproductive feedback loops are terribly slow: it takes nine months for a baby to be born, and years before that baby can even reproduce. Our edge is our intelligence: we’ve been able to create faster feedback loops with our tools. With language and writing, we exchange knowledge, and change behavior — all within a single generation.

By shortening feedback loops in this way, we’ve been able to create the three most important inventions of humanity — poetry, particle physics, and Soylent.

Systems As Self-Regulating Organisms

Great, so I’ve rambled for a bit about biological feedback loops, but what does any of this have to do with product development or business more generally? Let’s look at some areas where short feedback loops improve a product’s vitality.

Communication

A (good) conversation between two people is the product of a glorious feedback loop. You say something, the other person responds, and your next sentence should be informed by what that person said. Intelligent speaking requires active listening.

This is why it’s uncomfortable when someone is dominating a conversation: by not allowing other people to speak, the discussion loses relevance to every other participant. Conversation hogs ignore reactions in favor of their own actions, and that makes people feel ignored.

This is why quality stand-up comedians don’t just memorize their routines; they observe their audience and adapt to the room. The best teachers don’t lecture to students for 75 minutes; that’s not even a complete feedback loop — it’s just one prolonged action. Instead, they use the Socratic method to ask questions and create dialogue. With each sentence, a student exposes holes in understanding, providing the teacher with opportunities to course correct.

Media

An episode of South Park can be created in a week. This kind of production time is rarely seen in the animation industry, with most shows requiring months of work. The show’s creators believe this short turnaround makes the creative process more reactive, resulting in a consistently topical show. As vitality goes, it’s a winning strategy. Saturday Night Live has been delivering shows weekly since 1975 and shows no signs of stopping.

Musicians, too, have learned that audiences remain more engaged if they release a monthly single instead of a yearly album. This sort of musical titration fans the flames of fandom and helps artists gauge their audience’s engagement. Based on those reactions, musicians can pivot, incorporating feedback on style or sound into future decisions.

Finally, the recent surge in livestreaming (particularly in the gaming industry) seems like a logical conclusion to the world’s obsession with short feedback loops. What could be shorter than seeing something happen in real time? Streamers can immediately respond to viewers, modifying performances based on audience feedback. This allows viewers to feel involved in the product’s creation. Even a small amount of agency can increase emotional investment and loyalty.

Software

It shouldn’t come as a surprise that software engineers are quite familiar with short feedback loops. The very act of writing code is one big feedback loop:

Diagram-02.jpg

The beautiful thing about software is that it’s generally cheap to change. If you’ve got a kid with a difficult personality, you can’t update their source code. The only way to iterate on that product is to have another kid and update your parenting style.

On the other hand, finding a bug in software doesn’t require recalling the entire product; just fix the bug and move on with your life. Software engineers inherently understand the power of efficient feedback loops; it’s why we spend so much optimizing routine tasks.

A growing trend in the video game industry is to provide “early access” to titles. This allows developers to incorporate feedback from real users into design decisions. Some of the most successful games on the digital marketplace Steam are results of continuous development cycles. The value of short feedback loops should be apparent: open conversation between a developers and users results in a more enjoyable product.

DevOps Is Powered By Feedback Loops

Software is eating the world. And, like most processes, software development benefits from short feedback loops. So where should you start? The answer to my rhetorical question is “DevOps”. This movement concerns itself with improving how software is developed and delivered. By optimizing feedback loops between developers and operators, companies reduce the cost of shipping code and measuring user response.

This isn’t the only feedback loop you need to worry about. But it’s a feedback loop that can be automated, leaving you to focus on feedback loops relevant to your business. Before you can optimize those, however, you must locate them. You must…

Recognize Before You Optimize

Finding feedback loops in your organization starts with identifying the value you bring to customers. Use the four pieces of a feedback loop to generate questions about your process:

  • How do we make product decisions?
  • How do we change the product?
  • How do we deliver the product to users?
  • How do we know what users think of the product?

Diagram-03.jpg

Asking these questions should lead you to a single feedback loop at the core of your business. And this might not exist yet! If that’s the case, you should immediately focus on building out the missing functionality. This “primary” feedback loop defines your mission and highlights weaknesses. While there will certainly be other feedback loops, you should automate those until you’ve optimized your primary feedback loop.

A Contrived Example

Let’s say you’re running a paper clip company. Your job is to produce quality paper clips at an affordable price. Your primary feedback loop consists of the production of paper clips (action) and measuring customers’ responses (reaction).

Let’s say you start your business by manufacturing one million plastic green 1⅛” paper clips. When you bring these to market, you realize that everyone hates green and loves magenta. But you still need to sell this extra inventory, so you begin a social media campaign to move public preference from magenta to green.

Your company becomes well-versed in color theory, running centuries of art through machine learning algorithms to measure spectrum distribution over time. You discover that switching color bias is notoriously hard — particularly when the colors are complementary. So you go out of business.

This Is Clearly Silly

But there’s truth in the mess: people get distracted by problems they don’t need to solve. Had you made one hundred paperclips of each major color and measured which ones sold, you would have a better understanding of what people wanted. You could be helping thousands of folks keep their paper together — while still remaining profitable!

You should have asked questions about the primary feedback loop:

  • How long does it take to make a paper clip?
  • How long does it take to ship them to suppliers?
  • How often do suppliers have to restock?
  • How do you know which paper clips are popular?

Each of these questions leads to a metric worth optimizing. And you might discover that you lack the tooling to even measure anything! But that’s fine: even achieving awareness of your process is a worthy task.

Avoiding these questions is understandable; there’s often a Hard Truth hiding behind them, and sometimes the Hard Truth is that you’re building a product that no one needs or even likes. But you must ask and answer these questions if you want your business to succeed.

Optimization is a Feedback Loop, Too

Once you’ve identified your feedback loops, you’ll need to start thinking about optimizing them. As we’ve discussed, investing in your organization’s adaptiveness is one of the best things you can do for your company’s health.

But change doesn’t happen overnight! We’re all human, relying on patterns and structure to get through our days. Process optimization requires fundamental change, and even people who want to do that will have to push past some serious inertia.

The trick here is to introduce changes incrementally, according to DevOps best practices. If you’re trying to foster a testing culture, for example, mandating that everybody write tests Starting Today might receive some pushback. Instead, ask engineers to write a test for every bug they find. Each new test exposes gaps in code coverage, informing what areas of the codebase should receive next priority.

After sprints or outages, run retrospectives to explicitly capture feedback from stakeholders. By doing this, you walk away from mistakes with information on how to prevent them in the future.

Finally, actual user feedback should be prized above all else. Your “experts” won’t think like regular consumers because they’re creators.

Sweeping changes rarely work as expected; people are too stubborn and will find ways to resist commands from on high. But making small, incremental changes to your process will give your team some breathing room while your culture changes. This allows change to happen organically, while still prioritizing the most fragile parts of your product.

Survival of the Fastest

Software delivery is probably not your primary feedback loop. And it shouldn’t be; that’s our job. CircleCI spends its time devising ways to get code out of engineers’ heads and into users’ hands — without sacrificing speed or security.

We think that people make better decisions when they’re informed. And becoming informed requires making decisions — executing before you’re convinced of success. We’re here to help you make more decisions, more often. We can’t see the future, but we’re pretty sure it’s going to be different than the present. As the world moves faster, your organization needs to keep up.

And keeping up means increasing your product’s resilience to change. That change might be political, environmental, technological, or cultural — the whole point is that you won’t know when an asteroid is going to cause an extinction event. Don’t let your company go the way of the dinosaurs just because it moved like one; be the small mammal which survives an impact because it was able to adapt.

Identify your feedback loops and shorten them. Your descendants will thank you.