Choosing What to Build Takes Opposing, Extreme Mindsets
Two schools of thought dominate how startup founders and product leaders decide what to build: 1) wait for a stroke of genius (I’ll call this the savant style) or 2) do research and repeatedly speak with customers to improve the product based on their feedback (iteration). Where I often observe a clash is when a savant founder is paired with an iteration-focused product leader or vice-versa. The savant demands conformity to the vision regardless of what the users say, and the iteration-trained product team asks for data that doesn’t exist and refuses to get involved in what are perceived to be pet projects. When iteration and data leads, the savants grow impatient and their powerful ideas go to waste. I know I’m creating enemies in both camps by saying this, but the issue is that one philosophy isn’t actually better than the other; both are required for massive levels of success. The solution is to know what type of uncertainty each style can reduce and where it belongs in the lifecycle of a product, and then finding a way to use both models to achieve the best result.
A company with too much savant influence is characterized by taking on numerous only semi-related initiatives each of which has some serious unanswered questions with regards to how exactly the world is better for customers once the vision is instantiated. A company with too much iterative instinct can quickly cease to be innovative and may find that it cannot compel new users to join as it begins to focus almost exclusively on keeping existing users happy instead of simultaneously growing. To avoid either fail mode, lean on the savant style for deciding which initial problem to solve, and use the iterative style for making the best version of a solution to that problem.
Let’s make this concrete with some examples. At mePrism, our customers were tired of having their data used without their consent by the largest tech companies on the planet. The problem was clear, but only in retrospect. It’s not exactly something you would stumble across in a user interview if you started with interview questions generally related to data privacy. Sure, it’s possible, but it’s unlikely that a team of product researchers would organically stumble upon this particular problem and then find enough signal to decide that building a consumer Data Bank is a) the most important problem and b) the right way to solve it. In this manner, someone at the company has to make an authoritative decision very early on in the startup’s lifecycle to effectively start product development at the company by tackling a particular problem, and that decision will be flawed at some level, but it’s also an essential step for moving forward.
Once the startup chooses that problem and its initial solution design, deploying an iterative approach toward mastering the solution in the customer’s eyes is not only the best next step, it is also essential. Carrying forward the mePrism example, choosing which data sources (ex: Google data, Facebook data, Netflix data, etc) to expose to the users in what order is a perfect product research challenge. Analysts and Product Managers can identify which data sources improve the experience, value proposition, retention, and new user on-boarding the most, and make a data-driven recommendation. Conversely, choosing the data source in a savant style is likely to ignore important constraints and possibly use up the team’s time and energy on ones with lower returns.
I like to think of this trade off between methodologies as existing on a company value function with total value produced on the y-axis and the various incremental product decisions displayed on the x-axis.
Fundamentally, savant style puts you on the curve and iteration advances you toward the optimal spot on THAT curve, but only on THAT curve. Jumping between peaks is most straightforward through another savant inspiration.
When beginning with savant choice, you don’t know exactly which part of the function you land on and how valuable its maximum theoretically is, but you start with a problem that can be very powerful indeed. Once you’ve landed on the curve, product research and customer iteration improves the state of that decision until it reaches one of the local maxima. This is what I mean when I say that iteration eventually maxes out its return and then a team needs another savant-like inspiration, perhaps based on loose signals attained from interviewing and iterating, to make the next leap.
When these philosophies sing in concert and everyone at a company understands which sort of decision is being made, product development can move lightning fast, but when decisions that should be data driven get confused with ones that are more savant driven or vice versa, the entire machine slows to a crawl and even sometimes moves backwards.
