Destructive creation: when feedback loops go?wrong
In his bestselling book, Eric Ries says that the “Build-Measure-Learn feedback loop is at the core of the Lean Startup model”.?
You’ve probably heard of Build-Measure-Learn. It’s a technique for continuously testing and incrementally improving a product. Build-Measure-Learn and other experiment-based approaches are a powerful model for rapid iteration on your software product. That’s because they’re designed as a negative, ie, self-correcting, feedback loop. But in the right conditions, iterative product development can flip to become a positive loop. And that’s where things go terribly wrong for products.
What on earth am I talking about? Let’s take a step back, into the wonderful world of systems thinking.?
Feedback Loops
This isn’t a Systems Thinking 101 article (see the reading list at the end if you want more!), but it’s important to get on the same page about why feedback loops matter.?
Summary: Negative feedback loops are self-correcting and move a system toward stability or equilibrium. Positive feedback loops are self-reinforcing and move a system toward instability or chaos.
All systems, natural or human-designed, have negative feedback loops. “Negative” here does not mean “bad”. Negative feedback loops are self-correcting. You might know them by other names, like “checks and balances”. A negative feedback loop brings things back toward stability. Negative feedback loops have three parts:
Does that sound a bit familiar? It’s not that different from Build-Measure- Learn. “Build” is the response, “measure” and “learn” are monitoring. In this model, the goal sits outside the loop, in your experiment design (we’ll come back to that in a minute). By measuring and learning, you should stop building the things that aren’t moving you in the right direction.?
A positive feedback loop is self-reinforcing. It will ultimately destabilize or even destroy a system, moving it toward chaos. Colloquially, you might call positive feedback loops something like “downward spirals” or “vicious cycles”. An example of a positive feedback loop from nature is melting sea ice: melting of sea ice leads to reduced reflection from the white surface, which leads to faster melting of sea ice, and so on.?
It seems counterintuitive that a negative feedback loop could turn into a positive one. The key here is that a negative feedback loop must have a goal. If you are developing your product without a clear “why”, you’re building in a vaccum, and all your measurement is noise.?
Flipping the?Loop
Let’s dig into what that looks like.?
Let’s say you have a well-defined goal for your product. For very early-stage startups, goals tend to be simpler?—?seeking product-market fit, establishing traction, ensuring that you have a sustainable business. For businesses at any stage or scale, you may have one metric that matters, or a North Star metric, or a clearly defined statement of purpose or vision and mission. As you iterate on your product, it should be clear whether what you built is moving your measures in the right direction in relation to your goal.?
Perhaps you’re starting to have an uneasy feeling at this point. Let’s say your goals are not so well-defined. Maybe you have implicit goals, like “grow” or “get more revenue”. Maybe your goals are vague, like “win”, or “be the best”. Maybe your goals are clear, but there are lots of them, either concurrently or consecutively. As Richard Rumelt observes, “Many companies treat strategy as a way of presenting to the board and to the investing public their ambitions for performance, and they confuse that with having a strategy”. Ask yourself: with an unclear goal, how can your negative feedback loop self-correct? What is it correcting toward? Do you have a goal, or just a pile of ambitions?
This is where your product iteration can flip into a destructive positive feedback loop. You build something. You measure. You don’t achieve your ambition. You build more. You still don’t get “results”. The board is unhappy. You hire some more people so that you can build more. You gather more data. You hire more people so that you can use the more data more effectively. You build more.?
A common example of can happen in companies that seek to differentiate via customer experience. Because “experience” is inherently subjective, it’s very easy to keep building with the goal of “better”. But if you’re not aligned on what that really means?—?the “why” of your business or your product?—?you can do infinite amounts of work, and hire infinitely large teams, and achieve nothing but growth.
But wait, growth is good, right? The board says yes, growth is good. Growth will turn those results around. Growth aligns with your ambition. Growth will enable “better”. You build more. Piles more. And you hire! So much! Congratulations, now you’re in Growth Mode!
Why are you in Growth Mode? Because you had to keep building. Why? Because you weren’t achieving your ambition. Growth Mode feels like success, or at least, like progress. But Growth Mode is also a positive feedback loop. The more you grow, the more you have to grow. And if this is happening without clear goals, it’s very hard to design a negative feedback loop to kick in. You’ll deal with one eventually, though?—?they’re built into the system. A down round, a negative valuation, a market correction. Then you’ll respond, with layoffs and cost-cutting and hiring freezes. Stability will come, but it will hurt.
You can’t iterate your way to immunity from the business cycle. But you can make your iterative product development processes into effective negative feedback loops. Many teams will naturally gravitate toward gathering more or better data, or tightening the way they measure individual experiments. That’s important, and deserves attention. But the effort will be wasted if broader goals are not clear and aligned; it’s just more grist to the positive feedback mill. If your experimentation or iteration are consistently failing to deliver, this should be a signal to leadership that something is wrong upstream. The good news is that you’re zooming out to an area of higher leverage. The bad news is that the work is hard.
The Strategy?Lever
Your feedback loops need clear goals to operate effectively. As a leader, the realization that your team isn’t properly aligned around a goal is always humbling. It’s also really, really common.?
If you’re a leader in an organization where the wheels seem to be coming off your product development cart, you need to act to interrupt the positive feedback loop. You should start by examining your product strategy. Don’t tinker around the edges of the team’s work. Your job is to make sure that they can do their jobs. If your product is flailing or failing, your lever is strategy. Given how much bad strategy is out there in the world, I don’t feel unreasonable in asking you to take a long, hard look at yours. I’ve included some resources below that should help you assess your current strategy, improve upon it, or craft a new one.?
Developing a clear strategy and putting in the work to ensure that your team is aligned around it is the best way to ensure that your product build cycles continue to do the right kind of self-correcting work in your system.?
Resources
The Lean Startup , Eric Ries
A/B Testing: How to Get It Right, Ron Kohavi and Stefan Thomke
Leverage Points: Places to Intervene in a System, Donella Meadows
Why bad strategy is a 'social contagion', Richard Rumelt
Thinking in Systems: A Primer, Donella Meadows
Keep writing about this! Orgs that lack a clearly defined north star and a strategy for achieving it gain nothing by implementing iterative development.