Think Thrice

Should you join a startup? Heuristics to simplify tough decisions

A friend recently asked for my thoughts on whether they should leave their job to join an early-stage startup. Their current position is great: working with an autonomous team of peers on important software used by many people. The main issue was that they weren't learning as much anymore, and joining a startup is just damn exciting.

We discussed the situation, their thinking, and tradeoffs. It became clear that trying to project the different outcomes of how their career would unfold if they stayed, joined the startup, or looked for other opportunities was unproductive. There were simply too many variables at play.

How can we cut through the complexity of tough decisions like this? Are there a few parameters that integrate many - not perfectly, but well enough - that we can focus on? Is there one decision we can make that eliminates thousands?

During our discussion, we arrived at some key heuristics that helped clarify their decision in a holistic way:

Regret minimization

My friend is a generally optimistic and risk-averse person. This might be contradictory, but I think it means that while they believe things tend to work out in the long run (and have for them, as a matter of their life experiences), they have difficulty trusting that belief as the basis of an actual decision "in the now".

So I asked: "When you're 80, do you think you'd regret not taking this opportunity?".

After a brief pause, they answered unequivocally: "Yes."

I've found that thinking retrospectively about regret as your future self can provide a more balanced perspective (i.e. in alignment with your considered values) than thinking prospectively about uncertainty as your current self.

Why? When you think backwards, the potential experience has already occurred (with distance to process it), which helps you evaluate how you'd feel about it more accurately. Whereas when you think forwards, you tend to get distracted by unrelated events in your life (e.g. someone was mean to you, a mistake you made) and thinking about whether to have the experience at all.

Thinking backwards seems to help get rid of noise.

Negative visualization

While evaluating the experience from the future, it's useful to imagine that the worst-case scenario plays out. This helps establish that the worst-case scenario wouldn't be catastrophic and gets you to think about whether you'd enjoy the process regardless of outcome.

In my friend's case, the worst-case was that the startup goes out of business and their equity is worthless. This would mean some "lost" compensation relative to if they stayed at their current job or joined an established company, but not to a degree that would jeopardize their long-term goals.

This logic can be framed as a simple bet: would you actually pay $X for a roll of the dice? My friend's answer was "yes".

Imagining the worst-case made it clear that the lower bound on key parameters (type of work, culture match, learning opportunities) was actually quite high because of some past experience working with the founders. So, we reasoned that even if the worst-case outcome occurred, they wouldn't regret having the experience because of the learning they'd carry forwards. And the opportunity cost was minimal - they'd be able to access their old (current) environment and network easily.

This made it clear to us that this "risky" decision wasn't so risky after all.

Learning as an investment

It's useful to view learning as an investment. By investment, I mean that you put down some optionality now (typically because you'll make less money or work more hours) to acquire knowledge that could yield disproportionately more optionality in the future. At an early startup you'll learn a lot by making decisions, because there are so many to make and so few people to make them. And because fewer people have been early employees, these experiences are more valuable in the market.

My friend also values learning but was concerned about compensation. I didn't think it was a tradeoff: the learning was compensation in the sense that they could probably "liquidate" it if they joined any established company in the future. They weren't really losing money in the same way you don't lose money when you buy a stock - it just gets bound up. The stock could go underwater, but we established that they were comfortable with that.

While they had learned a lot from their current job, it was clear that they were well within the zone of diminishing returns. The learning opportunities at the startup would be decisively greater in both quantity and quality. Reframing learning as "delayed compensation" reduced the opportunity cost of joining a startup by examining the expected value of those experiences in the future.

The goal here isn't to trick yourself, it's just to have a more holistic consideration. You still have to know what you want to learn and how you expect that to unlock opportunities for your future self.

Working with interesting people

If you work with people you find interesting, you're going to learn from them, you'll enjoy the experience even if things don't work out, and you won't regret learning and having fun. Groups of interesting people collaborating with one another tend to create useful, valuable things. However, it's difficult to know exactly what those things will be from the outset (think about how many startups have pivoted to success) via some structured analysis, so you have to balance doing due diligence with overanalyzing.

I think you should prefer to work with interesting people over working with a particular technology or domain. I made this mistake in graduate school: an interesting topic won't overcome a stifling environment because meaningful progress requires long-term cooperation. In contrast, passion is infectious; an environment of interesting people can push you to explore and do things you wouldn't otherwise consider.

This wasn't particularly relevant in my friend's case as they've been fortunate to have always worked with interesting people. The startup and their current job are basically equivocal here, in the best way. But their experiences and success highlights how working with interesting people tends to makes the lower bound on things very high.


To simplify complex decisions down to 4 questions, ask yourself the following:

  1. When I'm old, will I regret not doing this?
  2. If things don't work out, would I still enjoy the experience?
  3. Will I learn a lot?
  4. Will I be around interesting people?

Part of the reason I'm writing this is as a reference for my future self to read the next time I'm making a tough decision. If things work out, I can expand on my experiences and why these heuristics might work. If they don't, I can expand on the nuances I think they miss (they're heuristics, after all).

Get the next post?