Business questions to ask before taking a startup data role

If you join a startup as an early employee, you’re essentially an investor. But unlike capital investors, you can’t diversify your portfolio as a full-timer. You need assurances that your time investment is likely to yield a positive return. Ideally, this would be a better return than the return on any other use of your work time.

Good startups guarantee a return for early employees by paying a competitive base salary. Typically, compensation also includes equity in the company, which vests over time. However, unlike equity in publicly-traded companies, there’s a good chance that your startup equity will be worthless.

To help you assess the value of your startup equity, this post presents questions from the Product & Business Model section of my Data-to-AI Health Check for Startups. In creating the health check, I realised I could have titled it “questions I should have asked past employers”, i.e., you get to learn from my mistakes! The health check has seven other areas, which I may cover in future posts.

Before proceeding, note these assumptions:

1. You don’t need to take a specific job offer urgently, i.e., you have multiple options or time to search for better options. If there’s urgency, you can take a suboptimal role, build your skills and savings, and aim to generate better options on your next search.
2. You’re not seen as too junior to be asking deep questions. If you’re early in your career, consider if early-stage startups are for you: My view is that you may be better off working with an established company, where there are more structured mentorship opportunities.
3. The startup isn’t proposing an unpaid equity-only position. If you’re expected to work full-time without a salary, you’re a founder.
4. The startup is small enough for you to speak with the founders as part of the recruitment process (maybe <100 employees).
5. The founders aren’t refusing to answer your questions. If they are, move on.

Investor-level product & business questions

Most of the following questions are typically answered by a pitch deck, so asking the founders to take you through the pitch may be the most time-efficient way of getting answers. Depending on the stage of product development, you may also be able to gather some answers yourself from the company’s website.

While these are investor-level questions, your assessment of the answers should be different from that of an investor. You don’t have dozens of other startups in your portfolio – only one. Better make sure it’s a good one.

Q1: What is your company’s purpose? What problem are you solving and why? You’re looking for a plausible story and solution. It’s important to understand founder motivations, and assess whether you want to spend many of your waking hours bringing their vision to life.

Q2: What does your product do? If the product is already live, a demo would be best to answer this. Otherwise, wireframes and other plans would be good enough. If the product isn’t live yet, watch out for unrealistic plans. You don’t want to work on a product that will never get released.

Q3: What are the relevant market sizes (TAM/SAM/SOM)? TAM is total addressable market – the total market demand for the product, which helps assess the growth potential. SAM is serviceable addressable market – the market demand the product can plausibly fulfill, which helps assess revenue targets. SOM is serviceable obtainable market – the part of the market that the startup’s product can capture, which helps assess short-term growth potential.

Q4: Where do the problem, market, and solution sit on Jason Cohen’s problem flowchart? While TAM/SAM/SOM are useful high-level metrics, the problem flowchart goes deeper into assessing the viability of a startup. A surprising number of startups skip such assessments, and fail as a result. You may regret joining startups that make such preventable blunders.

Q5: What is on the product roadmap for the next 6-12-24 months? In startup-land, plans for 12-24 months are in the realm of wishful thinking, but it’s good to have an idea of the general direction. Knowing what’s on the roadmap for the next six months will help you assess whether you want to come on board.

Q6: What is the business model, i.e., how do you make money? Together with the other questions, this will help you assess the viability of the business. You should be especially wary if the founders haven’t figured out how to generate revenue yet, which means they’ll have to raise money to keep paying you. If they’re not seeing healthy growth in other key metrics (e.g., number of active users), they’ll struggle to raise more funding.

Q7: What is the competition? How’s your product differentiated in the eyes of customers? How hard is it for competitors to copy you? Founders should have solid knowledge of the competitive landscape, and be able to explain why customers choose their product over the competition – and why they’ll continue to do so. Steer clear of founders who exhibit a low understanding of customer wants and needs. The company’s value ultimately comes from making something people want.

Q8: What are the key business metrics (definitions, values, and trajectories)? This is especially pertinent if you’re the type of data person who’s going to get deep into business metrics as part of your job (a data scientist/analyst, as opposed to a data/AI/ML engineer). But regardless of role, it’s important for you to know how the business is performing. You should be confident that startup executives are measuring the right things.

Q9: Since the last raise, how has the company performed against its goals? This includes goals that are covered by the key business metrics, as well as product development milestones. Repeatedly failing to achieve self-imposed goals is often a red flag – the goals may be unrealistic, and the business may not be viable.

Q10: How much runway is left before another raise is needed? This is critical for employees to know. For example, if there are only three months left before the startup runs out of money, you may be out of a job pretty quickly. Note that the question still applies if the startup is bootstrapped (i.e., self-funded or funded by revenue) – money needs to come from somewhere to cover your salary.

Data-to-AI product & business questions

While the above questions should be asked by any early startup employee, you should also get answers for the following questions if you’re considering a data/AI/ML role. If you’re the first data hire, pay specific attention to answers that indicate that the startup isn’t ready for a data hire, or that you may have to wear hats you’re unwilling to wear. For example, if you’re passionate about advanced AI/ML modelling but there are gaps in data engineering and basic analytics, you’re likely to be the one doing the data work to address those gaps.

Q11: What is the data intensity of the product on a scale of 1-5? High data intensity typically requires low-latency processing of large volumes of data with more than one database server. With high intensity, a slowdown in data processing would noticeably affect key business metrics. High data intensity means that solid data engineering skills are required for success – it’s important to ascertain that founders are aware of this requirement.

Q12: Is advanced AI/ML core to the product? What if you used simple heuristics? One issue with AI/ML is the hype. AI is indeed transformative and exciting, but using AI isn’t always required for the product to succeed. In the words of Google’s first rule of ML: “Don’t be afraid to launch a product without machine learning”. As a data professional and an outsider, you are in a good position to assess whether advanced AI/ML has to be core to the product. The answer should only be yes if it would make a difference in the eyes of the customers. Using AI/ML too early is often a premature optimisation. You should assess whether the added complexity of dealing with MLOps is justified.

Q13: Are you planning to increase data intensity or advanced AI/ML use? Why? This question is similar to the one about the product roadmap, but specific to data/AI/ML. Again, the Why is key – ensure that there’s a solid business case for increased data/AI/ML complexity. In a healthy startup, increased complexity is driven by customer need, not by excitement about shiny tech.

Q14: Are any decisions routinely blocked or delayed by limited access to data? This question helps assess gaps in data collection and quality, as well as the company’s culture around the use of data. It should also help you understand what sort of work is likely to be needed, e.g., even if there are plans to use more advanced AI/ML, the reality of data gaps may mean that plenty of data engineering work is needed.

Feedback welcome

If you found the above questions helpful or if you have any other feedback, I’d love to hear from you. I’m planning to evolve my Data-to-AI Health Check over time and post more on the other areas you should ask about. Subscribing for updates is the best way to get notified when it happens.