Your product is a joke
Improv comedy is a kind of theatre where performers create unscripted scenes based on prompts from the audience. I've been doing improv for 3 years and I often catch myself thinking that the framework used in improv applies to product pretty well.
Build a base reality before you get too creative
The base reality in any improv scene is the "who, what, where" of the characters on stage. For example, a base reality could be two friends (who) having lunch (what) in a park (where). When you're on stage with eyes peered on you, there's a natural rush to be funny but we need to resist. The base reality needs to be laid first and in fact, it doesn't need to be funny. It needs to be clear and serve as a neutral background for the funny things to contrast against.
The key product lesson is that you should understand the world you are in before you do anything else. This is the base reality of your product. Who are your customers? What are their problems? What do they do about these problems? What part of the existing solutions will you not change? Resist the natural rush to be creative with features, designs, engineering - and first observe and understand your base reality. The base reality shouldn't be innovative and you should be careful about contriving it to be what you wish. Be specific and truthful. Here are some examples of base realities.
Stop and get really explicit about your game
Once a base reality is established, improv performers look to find the game. This is the unusual thing that stands out, and everything funny revolves around. It's really important for improvisers to explicitly frame the game (with a shocked reaction, a confused question, a dramatic pause and so on) so everyone (including the audience) understands what's unusual.
It's similarly really important that you pause and get explicit about your product game. Be clear and aligned as a team about your unusual insights and beliefs. These will likely be your most critical assumptions and form the basis of your innovation. Your insight could be a specific take on the problem, the technical approach, the distribution, the trends in the space, and so on. Here're some examples.
Look to explore or heighten, but maintain focus
There's a natural tendency in improv to bring in lots of games because we're afraid about not being funny enough. However, the funniest scenes are funny not because they have multiple games, but because they take a focused game to places no one (even the performers) expects. Improvisors do just that by exploring the game ("if this is true, why is it true?") or heightening it ("if this is true, what else is true?").
This is an important lesson with product too. Resist the temptation to bring in lots of games because you want to hedge your bets or because it all looks exciting. Do less, and take it places no one (including you) can expect. Wonder "why" to explore the deeper rationale behind your key insights and reason from first principles. Wonder "what else" to heighten your key insights by deducing second (and n) order impact and peripheral ideas.
Don't use not knowing your game as an excuse to not start
Some improv scenes are initiated with a generic line and performers extract the game organically. e.g. "I can't believe it's midnight" is an intriguing start to a scene but there's no obvious game. In contrast, some improv scenes are initiated with strong game right away. e.g. initiating the scene with "No, you're an accountant, you can't just become a lion tamer". Both ways can lead to hilarious scenes.
Likewise, some products are initiated with a rough idea. This is in the camp of Eric Ries' model, where you're lean, get feedback, and iterate quickly. The idea is to treat the path to product market fit as a series of experiments with hypotheses. In contrast, there is Keith Rabois' model, where you have a strong vision from day 0 and not much changes from then. The idea is that you have a master plan from the start, and you get heads down on executing it. Check this post by Casey Winters comparing these models with far more nuance.
Of course, ideally you should find your product game as soon as possible to leave more room to explore or heighten. However, you shouldn't use not knowing your game from day 0 as an excuse to to not get started. You should start building your base reality today and see what surfaces. OpenDoor succeeded with strong initial premise, but Slack and Segment discovered their game over time.
There are a few things we can learn from improv to build better product. Of course, these insights could be deduced and communicated even without the improv analogy. However, I think seeing strange connections in lateral crafts creates a much deeper appreciation for the insights. Plus, it's fun to geek out on two things you love, at the same time. Hope you found these comparisons as amusing as I did, and hope you have a few laughs with product and improv.