I’m in the camp that a startup is a temporary organization, operating in conditions of extreme uncertainty, with a goal to search for a business model before the money runs out. And that environment requires a TOTALLY different toolbox than the one you need in an execution focused environment (e.g. selling industrial ammonia, building a skyscraper, or rolling out your product to the nth new market). They are not the same problems, and confusing one for the other has frustrating and existential consequences for a team.
Leadership in these conditions also requires adaptation. Guidance comes in the shape of guardrails and coaching not roadmaps and approvals. Effective guardrails look like: Clear company mission and vision, healthy high level quarterly goals that are results not feature focused, clear personas, unique product principles, development environment geared towards rapid experimentation, universal understanding of the problems we’re solving together, rapid regular face to face sessions with customers, scalable documentation like component libraries and style guide, and group design sessions.
There are a lot of tools in the box, but here are the most important ones. Much has been written and said on the lean approach to product development so I won’t waste much time reiterating.
Problem/Solution Fit > Product/Market Fit> Scale… IN THAT ORDER!:
First, validate that the problem you think you’re solving exists in the real world and that a solution is...possible. Next, figure out if that problem can be solved with a product (repeatedly) people can and will use, and that a pool of humans exists that are willing to pay your for it.
THEN...figure out how to grow it.
Doing this out of order is like feeding a Mogwai after midnight; something will grow but you’re not going to like trying to control it.
Experiment Driven:
When the path forward is obscured and always changing, the approach to progress needs to be bent towards who can find the fastest way through, not who thinks they have the best map. In that world, the experiment and what you learn (not the feature), is the best measure of progress.
Results Focused:
Creating clear goals with teams in the shape of precise and public metrics, not features, is a great way to create both an environment of accountability as well as one that nurtures creative thought and increased risk taking. Having teams own and publicize the nature of their experiments takes ego, personal failure, and confirmation bias off the table and replaces it with real conversations about what to think about and do next.
Build a Learning Machine:
Your ability to learn and grow is constrained by exactly that. You should invest in your delivery and learning process like you think about investing in product. A tuned up learning machine lets you ship early and often, experiment with split testing in the wild, and gets you in front of real users for qualitative formative feedback before you code a thing.
Users Drive the Boat Chief:
All organizations have “stakeholders”, but ultimately, it’s your customer that runs the show in the long run.
But listening to customers is NOT simply doing what they ask or compiling to a list of demands (although this is fertile territory to begin to understand their pain). Users are not designers and they’ve proven pretty ineffective at articulating how to solve a problem and its relative importance. The best way to keep customers happy is by deeply understanding the scale and frequency of their problems and designing imaginative, efficient, and delightful ways to solve SOME of them.
Autonomous Teams:
These types of cross functional pods are the fastest and most effective way to build great products almost entirely because they are self-sustaining and self-regulating. Small groups of engineers, designers, product managers, marketers, and who ever else has a front-line stake in the success of the product need to be working together unencumbered. Our goal in creating and shaping these teams is to provide them with whatever it takes to immediately get out of their way. It's like a controlled explosion. Buckle up.