5 Data Science Startup Red Flags You Need to Be Aware of Before Applying | by Madison Hunter | Oct, 2022

Startups aren’t all casual dress, dogs in the office, and catered lunches

One of the exciting parts of being a data scientist is the prospect of getting to work for a variety of companies, from those that are well-established to those that are just starting out.

Tech startups are enticing places to apply to because we often hear about the luxurious bonuses that can come with the job: free meals, quiet rooms, casual dress, flexible work options, dogs allowed in the office, and more.

However, beyond the fact that the kind of tech startups described above are few and far between, there are also several red flags that you need to be aware of before applying to these types of companies. While it could be argued that these red flags apply to all companies, they tend to be unique features of startups that you won’t be aware of until you’ve worked for one.

1. A lack of credible monetization led by people without a solid track record in the industry

Startups are often developed from a beer napkin agreement between young, ambitious techies who may have cut their teeth in the industry for a year or are straight out of university and want to create work for themselves.

Because of this, the first red flag of data science startups is easy to spot: a lack of credible monetization for a time long enough to get the company on its feet that is managed by people with a solid track record for excellence in the industry.

Startup capital is everything a startup needs to get the business moving and with it comes a requirement for the startup to become verifiable before the money runs out. Without properly managed startup capital, you can’t pay for office spaces, licenses or patents, salaries, marketing budgets, hosting fees, and more.

However, it’s not enough to acquire startup capital. You also need someone with a solid track record ensuring that the money is being spent appropriately and always with the intention to move the company forward. It’s easy to spend hundreds of thousands of dollars on office space and salaries without knowing what you’re actually getting out of it.

This means that when you’re applying to a startup, you need to interview your interviewer to determine how much money the company has, how long it’s expected to last, who’s running the show, and how the money is expected to be spent. It’s a very real fear that you could be hired by a startup and then suddenly not have a job a few months down the line when they run out of money due to poor management or irrational spending. While working for a startup is always a gamble, your job security for your contract period should not be.

Luckily, data science startups shouldn’t have that much overhead. Much of the work can be done from home and conference spaces can be rented out when the team is wanting to get together. Besides salaries and potentially office space, the only other costs will likely include software, servers, and marketing. Occasional costs will crop up for travel and events, though these may be low depending on the type of clients or partners the startup is targeting.

The key will be determining what percentage of the monthly budget is being spent on which cost and whether or not you think that’s reasonable given the needs of the company to move forward. Put simply, a few thousand dollars set aside every month for catering can be a red flag when only one thousand dollars are being set aside for marketing. Priorities speak when you see how much is allocated for each.

2. No obvious market, direction, or customer base

Lots of young, ambitious individuals will create startups just so they can brand themselves as “founders” and begin to live that startup lifestyle that has been so glamorized in the last 20 years. Unfortunately, it’s not uncommon for up-and-coming startups to have been founded without a clear market, direction, or customer base in mind for their products or services.

For data science startups, this should be an easy decision to make when developing the startup: provide data science products or services to customers in a specific niche. As easy as it is to say, it’s another thing to come up with the right answer.

It can be easy for startups to become distracted or to learn that a market they once thought was there no longer exists. For example, a data science startup that begins by saying that their target market is small businesses who need dashboard solutions to learn about their clientele and the efficiency of their processes could be swayed by the idea of creating custom full-scale data solutions for finance companies. Alternatively, a data science startup could go into the market thinking they could provide small businesses with data reporting solutions only to find that most small businesses are monopolized by a competitor or are happy using out-of-the-box solutions that can be purchased for a mere $100.

Therefore, it’s critical to determine whether or not you buy into what the company is selling when you’re doing your pre-interview research. That way, you’ll know whether or not there’s a real market for the product or service that you will be helping to provide. Startups who claim their proprietary AI can solve a company’s business problems using only a few months’ worths of data may be more smoke and mirrors than substantial results.

From my experience, it’s vital to remain focused on one specific niche that you know your startup can support. It can be easy after a while to become disenchanted, bored, or distracted by the latest news you’ve heard about a new area that needs data science services. Instead, it’s better to stay the course, focus on what you’re good at, target an enticing niche from the beginning, and build your reputation as being the company that services that niche. This focus will not only help keep the startup driven, but it will also help with keeping investors happy (point 1 above) and will help keep product or service development moving in the right direction (point 3 below).

3. The key product is still in the initial phases of development

If this is the case, the startup you’re applying to may be a sinking ship.

Due to the strict budgetary constraints that come with startup funding, there’s a certain cadence of experimentation, validation, and iteration that must be maintained for the company to keep moving toward becoming profitable. Even if this means that the startup must pull the plug on a product after years of investment and development, there must always be a sign of moving forward and not dwelling on an idea that just won’t manifest itself the way it was intended.

For example, Ford and Volkswagen just pulled the plug on their joint partnership to develop self-driving cars after concluding that the “large-scale profitable commercialization of self-driving cars was further out than expected.” While it was a $2.7 billion loss for Ford, they realized that it was better to cut their losses despite having been invested in this project for over two years.

Back to startups, the sign of a good data science startup is one that has a few iterations of an idea in the wastepaper basket but that is always moving forward with trying to make one idea work (without beating a dead horse). Data science products aren’t very difficult to come up with and produce successfully. More often than not, data science startups will be focusing on providing dashboards, reporting, AI, or other similar services to companies. There’s not much room for elaboration beyond tailoring these services or products to specific industries, nor are these particularly difficult services or products to develop and provide. Therefore, it should be pretty easy to see whether you’re walking into a good startup or one that is questionable.

From experience, I know that it’s always necessary to cut your losses, no matter how hard you worked or how much money and time you spent (i.e., Ford and Volkswagen). Yes, it’s a loss, but that leaves you open to trying new ideas that just might be the ones you were looking for.

4. A lack of staff

I once worked for an established tech company that was running on a startup model — in other words, we did the most amount of work with the least amount of staff possible. It casually resulted in me doing the job of seven other people, working for 180 days straight, staying up until midnight to hit deadlines, and not getting the feedback I needed to do my job until the very last minute.

A small startup isn’t necessarily a red flag. It only becomes a red flag if you notice that everyone is doing the job of at least three other people and only getting the pay of one. While many data science functions don’t necessarily need a big team, it’s vital that each person in the startup is doing enough work to move the company forward but not so much work that deadlines are being missed or the quality of work decreases due to overwork and fatigue. For example, a data science startup may find that having a couple of software engineers on staff to handle the production side of things can help free up their data scientists for more data-related tasks.

Due to budgetary constraints, many startups will try to keep a shoestring staff to keep costs low. However, what they risk doing is burning out vibrant employees, creating a toxic work environment, and not pushing the company to its full potential. Yes, the salaries may be lower in the beginning to afford all of the required staff members, but it will pay dividends later on as the startup flourishes and becomes self-sustaining. It’s better for data science startups to have a full team of data engineers, data scientists, business analysts, software engineers, and marketing specialists than to have a few people handling all of these aspects (who may not even be very good at all of them).

It’s vital to broach the topic of staffing early on in the conversation. Having an idea of team member workloads and responsibilities can help you decide whether the team is running short-staffed or conservatively.

5. “We’re a family”

For some reason, startups that described themselves as “one big happy family” became desirable to us as workplaces. I believe that this is a conspiracy where, in reality, startups planted the idea in our heads that we wanted to work for companies who stated that they were “more like a family” and we bought into it, the same way we bought into casual dress, free lunches, dogs in the office, and team bonding nights.

While most data science startups are close-knit ventures with a smaller staff than most tech startups, that doesn’t mean that the workplace needs to have a “family” mentality. Yes, you may be more closely aware of your co-worker’s personal life due to the proximity in which you’re working with a smaller number of people, but that doesn’t mean that your professional and personal lines should become blurred.

If a startup describes its company culture as being “like a family”, don’t hit “submit” on your application submission — just run for the hills.

Workplace family culture is a toxic effect that creates an exaggerated sense of loyalty that can become harmful, can create a power dynamic where employees can get taken advantage of, and causes personal and professional lines to become blurred.

Enough said.

What you need to look for in a data science startup is a team who can work together professionally, values new ideas of how to handle menial data problems, and prioritizes healthy work-life balances. None of this “family” BS.

Startups aren’t all casual dress, dogs in the office, and catered lunches

One of the exciting parts of being a data scientist is the prospect of getting to work for a variety of companies, from those that are well-established to those that are just starting out.

Tech startups are enticing places to apply to because we often hear about the luxurious bonuses that can come with the job: free meals, quiet rooms, casual dress, flexible work options, dogs allowed in the office, and more.

However, beyond the fact that the kind of tech startups described above are few and far between, there are also several red flags that you need to be aware of before applying to these types of companies. While it could be argued that these red flags apply to all companies, they tend to be unique features of startups that you won’t be aware of until you’ve worked for one.

1. A lack of credible monetization led by people without a solid track record in the industry

Startups are often developed from a beer napkin agreement between young, ambitious techies who may have cut their teeth in the industry for a year or are straight out of university and want to create work for themselves.

Because of this, the first red flag of data science startups is easy to spot: a lack of credible monetization for a time long enough to get the company on its feet that is managed by people with a solid track record for excellence in the industry.

Startup capital is everything a startup needs to get the business moving and with it comes a requirement for the startup to become verifiable before the money runs out. Without properly managed startup capital, you can’t pay for office spaces, licenses or patents, salaries, marketing budgets, hosting fees, and more.

However, it’s not enough to acquire startup capital. You also need someone with a solid track record ensuring that the money is being spent appropriately and always with the intention to move the company forward. It’s easy to spend hundreds of thousands of dollars on office space and salaries without knowing what you’re actually getting out of it.

This means that when you’re applying to a startup, you need to interview your interviewer to determine how much money the company has, how long it’s expected to last, who’s running the show, and how the money is expected to be spent. It’s a very real fear that you could be hired by a startup and then suddenly not have a job a few months down the line when they run out of money due to poor management or irrational spending. While working for a startup is always a gamble, your job security for your contract period should not be.

Luckily, data science startups shouldn’t have that much overhead. Much of the work can be done from home and conference spaces can be rented out when the team is wanting to get together. Besides salaries and potentially office space, the only other costs will likely include software, servers, and marketing. Occasional costs will crop up for travel and events, though these may be low depending on the type of clients or partners the startup is targeting.

The key will be determining what percentage of the monthly budget is being spent on which cost and whether or not you think that’s reasonable given the needs of the company to move forward. Put simply, a few thousand dollars set aside every month for catering can be a red flag when only one thousand dollars are being set aside for marketing. Priorities speak when you see how much is allocated for each.

2. No obvious market, direction, or customer base

Lots of young, ambitious individuals will create startups just so they can brand themselves as “founders” and begin to live that startup lifestyle that has been so glamorized in the last 20 years. Unfortunately, it’s not uncommon for up-and-coming startups to have been founded without a clear market, direction, or customer base in mind for their products or services.

For data science startups, this should be an easy decision to make when developing the startup: provide data science products or services to customers in a specific niche. As easy as it is to say, it’s another thing to come up with the right answer.

It can be easy for startups to become distracted or to learn that a market they once thought was there no longer exists. For example, a data science startup that begins by saying that their target market is small businesses who need dashboard solutions to learn about their clientele and the efficiency of their processes could be swayed by the idea of creating custom full-scale data solutions for finance companies. Alternatively, a data science startup could go into the market thinking they could provide small businesses with data reporting solutions only to find that most small businesses are monopolized by a competitor or are happy using out-of-the-box solutions that can be purchased for a mere $100.

Therefore, it’s critical to determine whether or not you buy into what the company is selling when you’re doing your pre-interview research. That way, you’ll know whether or not there’s a real market for the product or service that you will be helping to provide. Startups who claim their proprietary AI can solve a company’s business problems using only a few months’ worths of data may be more smoke and mirrors than substantial results.

From my experience, it’s vital to remain focused on one specific niche that you know your startup can support. It can be easy after a while to become disenchanted, bored, or distracted by the latest news you’ve heard about a new area that needs data science services. Instead, it’s better to stay the course, focus on what you’re good at, target an enticing niche from the beginning, and build your reputation as being the company that services that niche. This focus will not only help keep the startup driven, but it will also help with keeping investors happy (point 1 above) and will help keep product or service development moving in the right direction (point 3 below).

3. The key product is still in the initial phases of development

If this is the case, the startup you’re applying to may be a sinking ship.

Due to the strict budgetary constraints that come with startup funding, there’s a certain cadence of experimentation, validation, and iteration that must be maintained for the company to keep moving toward becoming profitable. Even if this means that the startup must pull the plug on a product after years of investment and development, there must always be a sign of moving forward and not dwelling on an idea that just won’t manifest itself the way it was intended.

For example, Ford and Volkswagen just pulled the plug on their joint partnership to develop self-driving cars after concluding that the “large-scale profitable commercialization of self-driving cars was further out than expected.” While it was a $2.7 billion loss for Ford, they realized that it was better to cut their losses despite having been invested in this project for over two years.

Back to startups, the sign of a good data science startup is one that has a few iterations of an idea in the wastepaper basket but that is always moving forward with trying to make one idea work (without beating a dead horse). Data science products aren’t very difficult to come up with and produce successfully. More often than not, data science startups will be focusing on providing dashboards, reporting, AI, or other similar services to companies. There’s not much room for elaboration beyond tailoring these services or products to specific industries, nor are these particularly difficult services or products to develop and provide. Therefore, it should be pretty easy to see whether you’re walking into a good startup or one that is questionable.

From experience, I know that it’s always necessary to cut your losses, no matter how hard you worked or how much money and time you spent (i.e., Ford and Volkswagen). Yes, it’s a loss, but that leaves you open to trying new ideas that just might be the ones you were looking for.

4. A lack of staff

I once worked for an established tech company that was running on a startup model — in other words, we did the most amount of work with the least amount of staff possible. It casually resulted in me doing the job of seven other people, working for 180 days straight, staying up until midnight to hit deadlines, and not getting the feedback I needed to do my job until the very last minute.

A small startup isn’t necessarily a red flag. It only becomes a red flag if you notice that everyone is doing the job of at least three other people and only getting the pay of one. While many data science functions don’t necessarily need a big team, it’s vital that each person in the startup is doing enough work to move the company forward but not so much work that deadlines are being missed or the quality of work decreases due to overwork and fatigue. For example, a data science startup may find that having a couple of software engineers on staff to handle the production side of things can help free up their data scientists for more data-related tasks.

Due to budgetary constraints, many startups will try to keep a shoestring staff to keep costs low. However, what they risk doing is burning out vibrant employees, creating a toxic work environment, and not pushing the company to its full potential. Yes, the salaries may be lower in the beginning to afford all of the required staff members, but it will pay dividends later on as the startup flourishes and becomes self-sustaining. It’s better for data science startups to have a full team of data engineers, data scientists, business analysts, software engineers, and marketing specialists than to have a few people handling all of these aspects (who may not even be very good at all of them).

It’s vital to broach the topic of staffing early on in the conversation. Having an idea of team member workloads and responsibilities can help you decide whether the team is running short-staffed or conservatively.

5. “We’re a family”

For some reason, startups that described themselves as “one big happy family” became desirable to us as workplaces. I believe that this is a conspiracy where, in reality, startups planted the idea in our heads that we wanted to work for companies who stated that they were “more like a family” and we bought into it, the same way we bought into casual dress, free lunches, dogs in the office, and team bonding nights.

While most data science startups are close-knit ventures with a smaller staff than most tech startups, that doesn’t mean that the workplace needs to have a “family” mentality. Yes, you may be more closely aware of your co-worker’s personal life due to the proximity in which you’re working with a smaller number of people, but that doesn’t mean that your professional and personal lines should become blurred.

If a startup describes its company culture as being “like a family”, don’t hit “submit” on your application submission — just run for the hills.

Workplace family culture is a toxic effect that creates an exaggerated sense of loyalty that can become harmful, can create a power dynamic where employees can get taken advantage of, and causes personal and professional lines to become blurred.

Enough said.

What you need to look for in a data science startup is a team who can work together professionally, values new ideas of how to handle menial data problems, and prioritizes healthy work-life balances. None of this “family” BS.