đShould your company be optimizing or discovering?
A great [founder] doesn't question themselves, just their approach and level of effort. My review of Matt Lerner's "Growth Levers"
Introduction: A Personal Perspective
Throughout my career as an individual contributor, I've observed two distinct operational modes in software development: optimization (which I previously called "polishing") and discovery (formerly "researching"). These modes represent fundamentally different approaches to problem-solving and development and as we will learn in this article produce entirely different outcomes!
In this post I look at Matt Lernerâs new book âGrowth Leversâ. We dive into his concise expression of how established companies must operate differently than establishing ones. Finally we look at leadership principles presented in the book, and how leaders need to hold a conflicted viewpoint of confidence and humility.
Understanding the Two Modes
In optimization mode, success criteria are clearly defined: high test coverage, zero lint/compile warnings, and strict type safety. This mode focuses on refinement and improvement of existing systems and processes. Conversely, in discovery mode, the primary measure of success is shipping features and delivering results. Consider this analogy: When mining for gold, should you meticulously optimize your blasting technique, or should you first identify promising locations and optimize later?
Technical Implementation Considerations
Python development provides an excellent example of these dual modes. Via import typing, developers can establish strict contracts between functions, allowing tooling to enforce specifications. However, this raises an important question: If the specification itself isn't solidified, does enforcing strict typing add value?
In discovery mode, when exploring novel approaches, static typing might unnecessarily constrain progress, particularly in solo development. This isnât to say however that typing is always bad in discovery mode. In team settings, types serve as "executable documentation," helping communicate intent effectively. I find the composition and experience of the team balanced how much of âmove fastâ vs âcheck your typesâ is appropriate.
Extending the Concept: From Individual to Organization
Matt Lerner's book "Growth Levers" extends these observations to team and organizational levels, identifying crucial differences in inputs and desired outcomes between these modes. I summarize these modes into these two pseudo-math expressions
Optimization Mode
Discovery Mode
According to Learner, in Discovery mode (what I previously called research) time is the precious resource. You must leverage each unit of time into the greatest learning you can. I think of the units of this expression as facts per hour (fph). The goal is to produce exponential gains. Optimization mode on the other hand, money is the resource to be leveraged, optimizing the $1 of cost to turn into the subsequent $1 of revenue. Unit here are dollars per dollar (dpd). Optimization can only find marginal gains, to achieve exponential gains, you need novel approaches, ergo writing new procedures and playbooks.
Practical Applications in Startups
Minimum Viable Testing (MVT)
Lerner offers an insightful refinement to the startup concept of Minimum Viable Product (MVP) by introducing an even lighter-weight approach called the Minimum Viable Test (MVT). Among the MVT strategies he highlights, I found the landing page test and the false door method particularly interesting.
The landing page approach involves creating a simple webpage that collects email addresses from interested users, informing them that the product is still in development and adding them to an interest list. As an engineer I like building things. Testing if we should even build it is not a first step in my default operating mode. Hence I need to expciltlt test if the idea is even worth build! Step 0! Russell Brunsonâs books have some interesting approaches to designing compelling landing pages.
The false door strategy takes this a step further by presenting users with a button or interface element that appears to offer a feature, but instead of providing functionality, it captures valuable user feedback when clicked. When users interact with this false door, they're prompted to explain what they expected the feature to do or how they thought it would solve their problem. This approach is particularly powerful because it captures user insights at the exact moment they're actively trying to solve a problem, providing authentic, contextual feedback without the investment of actual feature development. Also, phrased properly, a false door can show customers you are actively working on the product to improve their experience, and you care about their input into that process.
Experiments and User Interviews
The value of any experiment lies in its ability to generate actionable insights and surprise! Lerner's approach to analyzing MVT results aligns well with the customer development principles outlined in "The Mom Test" by Rob Fitzpatrick. After each test, teams should collectively evaluate their findings by asking fundamental questions: Is there measurable value in what we've learned? Should we expand, repeat, or scale this test? Have we discovered any activities we should stop, or hypotheses we can now eliminate? Most importantly, should we adjust our key performance indicators based on these new insights? This systematic approach to experimentation, focused on testing our boldest assumptions about users, forms the foundation of effective customer discovery. If you're interested in a deeper dive into how these principles connect with setting meaningful KPIs and conducting effective user interviews, feel free to drop me a note.
Cognitive Challenges and Leadership
Through his interviews with Nobel laureates, Lerner uncovered a compelling mindset around experimental failure. Great scientists maintain unwavering confidence in themselves while rigorously questioning their approaches and effort levels. Science itself is the process of learning what can be learned, just as entrepreneurship is the process of building what can be built. This insight reveals a fascinating paradox that all founders must embrace: maintaining absolute confidence in their current answer while remaining equally confident in their ability to find a better one. The interplay of these seemingly contradictory mindsets enables founders to navigate through uncertainty. The implications become clear when explaining complex systems, as we naturally fill knowledge gaps with generalizations and assumptions. This creates an illusion of knowledge where presenting solutions inadvertently strengthens our confidence in these assumptions, unconsciously elevating our generalizations to the same level as hard data. That's why vision must remain malleable and learning continuous. Discovery and Optimization modes emerge not as linear steps but as circular arcs of continuous improvement, occasionally spiraling into entirely new realms of possibility.
Questions emerge as one of the most under-appreciated yet powerful tools in a leader's arsenal, specifically questions centered around user behavior and user understanding. When teams find themselves caught in circular debates about different approaches, it often signals underlying disagreements about customer needs rather than technical solutions. Leaders can cut through this confusion with precise questions:
"Which customer said that?" grounds discussions in real feedback rather than assumptions.
"What are our riskiest assumptions about our customers?" helps teams identify and test critical hypotheses.
"What's the fastest way we can test that?" drives teams toward action and learning.
In one-on-one meetings, consistently asking "What's important and how is it going?" creates a natural cadence of accountability while keeping focus on discovery.
This question based leadership approach proves particularly valuable during the inevitable long stretches where progress feels elusive, which is the reality for most startups. Since success typically arrives not as steady progress but as years of learning suddenly culminating into explosive escape velocity and growth, leaders must use these questions to maintain team direction and morale.
Conclusion
The fundamental distinction between discovery and optimization modes lies in their potential impact. Optimization mode, while essential for established processes, yields incremental gains through careful refinement and efficiency improvements. Discovery mode, however, opens paths to outsized returns by enabling teams to uncover entirely new approaches and opportunities. Success in discovery mode demands that leaders embrace the scientist's dichotomy: maintaining unwavering confidence in their ability to find answers while simultaneously holding their current solutions lightly enough to refine or replace them. This delicate balance of confidence and humility enables teams to navigate the uncertain territory where breakthrough innovations emerge. Just as scientists don't question their fundamental capabilities but rather their methods and effort, effective leaders must cultivate both the conviction to pursue bold hypotheses and the flexibility to adapt when evidence suggests better approaches. In doing so, they create environments where transformative discoveries become possible, leading their teams beyond incremental improvements toward truly exponential gains. I thank Matt Lerner for his excellent book and for taking the time to codify his insights into a well written and enjoyable book.