How to evaluate a startup?

What all are the questions that one should ask in order to do the understand the potential of the startup so that there is less risk of layoffs when joining there, given the uncertain times.

1. What does your path towards profitability look like? - This one is the most important: In this economic climate, startups need to have good fundamentals. Money coming in should be more than money going out.
2. How is your revenue growth over time? - If a startup isn't growing, then it is dying.
3. Is customer retention healthy? - Even if you have good growth, it's hard to survive with high churn. This is called the "shark fin effect".
4. How has headcount grown over the past year, and what's the plan for its growth next year? - This is similar to #2 in that if it's flat, that's a bad sign. However, if it's too much (e.g. planning to grow 5x in headcount next year), this could be a sign of irresponsible spending.

All that being said, there's not much stopping the interviewers from stretching the truth when you ask these questions (and they probably will). Similar to interacting with people in general, it's up to you to assess the veracity of their responses and look for signs like body language.

Do the best you can to get information through Crunchbase, tech news, your network, and asking these questions, but when it comes to joining a startup, there's always some element of a leap of faith.

Once you get the interview, follow the advice from here to pass: "How to prep for interviews at a startup?"

If you do end up going to a startup, here's some useful resources I recommend:

• Session #8 - Hypercharging Your Growth At Early Startups (w/ Staff Engineer Seed Zeng)
• "What's the difference between Big Tech and startups in terms of learning and growth?"

To add on to Alex's points.

Customer retention

ask for Net Dollar Retention metric. Google around for what "good" looks like (at least 110%, ideally above 150%)

Revenue growth

Revenue should triple or double year over year. Source: https://www.saastr.com/is-it-now-triple-triple-triple-double-double-double-t3d3-for-top-tier-saas-startups-probably/

Join a startup after they have product-market fit to derisk. That means the product has to be already in a good shape and people are buying (evidence from ARR, annual recurring revenue). The next step is to build out missing features and do scaling work: in software and in sales, marketing.

One heuristics of a growth phase B2B company (pretty sure chance of equity being worth something) is the hiring of a lot of sales and marketing people. In my limited experience, for B2B company, the sales headcount (business development, account manager, solution engineers...) is at least double the engineering headcount. You can see the company's Linkedin profile for this.

Joining a startup before product-market fit means accepting that there can be years of iterating on a product that does not sell well, or the company folding. You would get more equity though.

Overvaluation

Google for "revenue multiple", it's roughly the x in the formula below:
revenue * x = valuation

For example, if revenue is 100mil, and valuation is 10 bil, then revenue multiple is 100.

As of Dec 2022, this is too high, meaning if the startup were to be public right now, the market would value it at a price way below 10bil. And your equity is worth less than the startup claims.

I don't know what's a good multiple for now, many things affect it, but my impression is it should be below 50, 20-30 is better.

I believe selecting a startup to join is not easy at all, but it's easier than commonly thought. Hope the above points help.

email me at john dot snow dot alias at gmail dot com if you want to chat.

Thanks Alex and Rahul for the great forum. I asked and received several answers so far about career growth as an engineer.

Forgot to mention one of the "easier" ways of knowing if a tech startup is legit: Seeing how many of their engineers come from FAANG.

Of course, the following are also true:

• A startup can do well without any of their early engineers coming from FAANG
• There are tons of bad engineers at FAANG.
• There are many incredible engineers who never work for FAANG.

But in general, talent tends to follow the money and the average FAANG engineer is extremely skilled relative to the general engineering population. If a startup can convince several engineers of their caliber to join them, there's a very above average chance they're cooking something impressive.

I worked at Robinhood, and I saw this first-hand. So many of the first 50 engineers were ex-Google, ex-Meta. They brought in the best practices from those companies, shaved off the garbage big company red-tape and bureaucracy stuff, and very successfully grew the early core of Robinhood around it.

Read the Star Principle by Richard Koch.

For you to join a startup should be:
• in an industry growing 10%+ per year
• in a market growing 10%+ per year
• growing as the market segment leader at 10%+ per year

If you are wrong about the company, you'll have the industry and market segment knowledge to transfer into the market segment leader.

Right now... tech is slowing up, oil and energy are growing. When oil and energy dive, jump back into tech.

• Quality of people joining the company. If the smartest people in a domain are joining, there's a good chance there is some magic at the company, and you'll learn from incredible people.
• Quality of the founders: if the founders are well-respected/famous/serial entrepreneurs, it effectively puts a "floor" on the exit of the company. (At minimum, it'll get acqui-hired.)
• Try out the product! This is easy to do if it's a consumer product, e.g. Snap or Airbnb.

In terms of questions to ask:

• What are the plans for the company, and for the team, over the next 3-6 months?
• What are the biggest risks to the business? How are they trying to mitigate them?

I actually think it's hard to deeply understand the culture from a 30 min phone call. If the company is in hyper-growth, e.g. doubling or tripling each year, the culture will evolve over time.