What I Learned Working in a Hypergrowth Startup: On Deck

My learnings from working at On Deck during its hypergrowth phase when the company went from 15 to 200 employees in less than a year.
February 25, 2024
Facebook LogoTwitter Icon
Black LinkedIn Icon
On Deck team photo infront of a cabin in tahoe

Hi, I’m Kieran. I worked at On Deck during its hypergrowth phase from September 2020 - January 2023. As employee #15, I contributed to the startup’s growth from no revenue to a $20 million run rate in less than a year. In that time, On Deck hired ~175 people. I led the team that built the admission infrastructure to support and scale 20 fellowships for founders, job seekers, capital allocators, operators, and creators. 

The exponential growth and expectations of a well-funded startup trying to scale that fast creates so much tension on the entire organization and team – you are forced to adapt how you work to the new rules, boundaries, and playing field. My friend encouraged me to write about what it’s like to work at a hypergrowth startup in case it’s helpful to people evaluating whether to join one and how to excel in this environment. Some of my learnings will apply across all startups experiencing this type of growth, while others may be more specific to On Deck.


newton's cradle showing momentum

Startups live or die based on their ability to generate momentum. In the early days, startups are tasked with creating momentum out of thin air – selling a product before it’s built, convincing team members to join without a fully refined vision, etc. It’s thrilling and one reason why many people who join a startup end up working at startups for the rest of their careers. At a hypergrowth startup, the peaks and valleys are more pronounced. At On Deck, I remember them vividly.

The peak:

  • Scaling to a $20 million run rate
  • Raising a $20 million Series A led by Keith Rabois at Founders Fund, then a pre-emptive $40 million Series B a couple of months later led by Tiger Global Management
  • Raising a $100 million fund led by Tiger Global Management to create an accelerator from scratch and invest in hundreds of founders

The valley:

  • Tiger Global Management breaking a signed term sheet
  • Laying off ¼ of our team (~72 people) in May 2022
  • Not cutting deep enough and having to lay off another ⅓ of our team (~73 people) in August 2022
  • Several senior leaders stepping back from the startup

When a hypergrowth startup starts to spiral, it really spirals. To paint the picture, a few months before the first round of layoffs, I moved to work on admissions for the accelerator side of the business. In less than a month, we designed an application, created an admission process, hired investment partners across different industry verticals and geos, implemented a 48-hour SLA for each application, and built custom infrastructure – a combination of no-code and code. It was a heroic team effort that allowed us to conduct diligence, fund, and support 100s of founders from a stack of 7,500+ applications that would come in over the next few months.

While I was focused on that side of the business, On Deck hired a new team that changed the previously set up admission process we were running across our ~20 fellowships, the revenue-generating side of the business, for job seekers, capital allocators, operators, and creators into more of a traditional sales process. Rather than candidates selling us on why they were a good fit for a fellowship, we started to sell them.

When we missed our targets on the revenue side of the business and conducted the first round of layoffs, I moved back to oversee admissions across the board for the fellowships. This entailed working with new fellowship Directors and direct reports and teaching people how to run an admission process. I have a lot of empathy for the people I worked with during this period – they were hired under different circumstances, expectations, and for different skill sets. It was completely foreign to them.

This was the first time in my life, at work, where I felt like I was losing no matter how hard I worked or which levers I pulled.

The excitement from seeing low-hanging fruit and an opportunity to make more revenue, change the process into one that I knew would close the best candidates, and nail the details to create a delightful candidate experience quickly turned into the feeling of defeat and the reality that I was not capable of doing more to help the team.

It’s like the pushing the boulder up the hill metaphor that startups talk about – but the boulder was now 3x the size, and the hill was greased. There is just so much more debt – from the infrastructure to the clashing politics, to the drop in team morale from the layoffs – that made it much harder to operate and build what I wanted to build the second time compared to when On Deck was still a small startup.

We started to see better results each week, which earned us more buy-in from different stakeholders. We salvaged as much revenue as we could in the short quarter. But at the end of the day, it still wasn’t enough. We were burning too much money relative to our revenue and had to do the second round of layoffs or risk running out of money—the team’s morale sunk even more.

Momentum is important at a hypergrowth startup because the force required to overcome downward momentum is 10x more.

Where to focus – the core product & business

chart showing google's revenue by product
Google’s revenue by product. Image Credit: Four Week MBA

When I interviewed at On Deck, many people mentioned that shiny object syndrome was one of the significant risks to the business. While true retrospectively for On Deck, many early startups like to market this as their weakness. The weakness comes off as a sign of strength in some regards – a signal that the team is ambitious and wants to accomplish a lot – rather than for what it really is, an existential threat to the business, like regulatory hurdles, technology risk, or market risk.

I fell into this trap and didn’t see it for what it really was when I was interviewing. A startup without focus is a sign of immaturity and should be treated as a red flag to candidates, especially if the startup has enough data to think that the core product can be a venture-scale business. In the next section, entitled product market fit, I’ll go into more detail about why I believe this to be true.

I don’t blame the founders or other leaders at the company for having shiny object syndrome because I had shiny object syndrome – I think we all did. When we launched ODX, On Deck’s accelerator program, the shiny new object, almost every early employee wanted to be on the new team. It was because many of us joined wanting to help founders, so we saw this as a way to get closer to our personal missions of helping these people rather than our other customer personas. My secondary reason was that I felt I could learn more by working with people I admired, like Erik Torenberg, Sam Kirschner, Minn Kim, Gonz Sanchez, Trish Kennelly, and Shawn Xu, on the new thing than continuing to hammer away at my previous role.

Don’t get me wrong, I learned a lot, but I think I learned more in my previous role. In it, I became obsessed with helping people learn how to run an excellent admission process, building infrastructure, designing different criteria to vet for various customer personas, and understanding how to measure and improve our network quality as we scaled. Through my obsession, I went from someone who knew nothing about admissions to developing a strong conviction and set of beliefs from a combination of my own intuition, data, and feedback loops that I was collecting around how to build an admission system to support a 100-year institution – the initial vision for On Deck.

I now realize that I and other startup employees, especially young and ambitious ones, move on too quickly to the next thing. Find out for yourself, but I now believe the best way to learn is to focus on your core responsibilities and by executing against a metric – whether that be conversion, retention, revenue, etc. – over long periods, even if it gets to the point of decimal percentage point increases. The knowledge you’ll obtain from making these improvements to the system will be unmatched, especially in the later innings.

Because of the implications momentum has on hypergrowth startups, it’s even more important that your A players work on the core product and business. Very successful big companies actually organize much better than startups in this regard – the best engineers at Google have always focused on Google Search, Gmail, and other core products, not experiments like Google+, Google Glass, etc.

On Deck as the example, here is a framework that you can use to think about your role and what to work on if you’re integral to the startup’s success:

  • If ODX worked and On Deck failed, then we failed to build a venture-scale business
  • If On Deck worked and ODX failed, then we built a venture-scale business

When you think about it like this, it’s clear what I, and many others at On Deck, should have worked on.

So, do everything you can to focus on the core product and business, and insist that other leaders do as well.

Product Market Fit

poster that says "make something people want" in orange with Y Combinator logo
Image Credit: Y Combinator

I teased this earlier, but even if you believe you have product market fit for your core business, you need to stay focused on it because it can change. The market, customer preferences, updates to the product, pricing, competition, and other factors can all dictate how much product market fit your business actually has.

At On Deck, it happened gradually, at least for our core product, the founder fellowship, which made it hard to see. I think we had it in the very early days when cohorts were small, and most fellows came in via referrals from our network. But, once we tied employee compensation to revenue and the size of the cohort, the quality bar started to slip. As a result, the fellowship produced notable outcomes less frequently, and the perception of the network quality started to decay.

For the other fellowships, we should have seen it earlier, as most of our applications were coming in through inorganic methods like newsletter sponsorships, paid ads, and one-time launch announcements rather than organic methods like referrals. The net promoter score also fluctuated widely across fellowships – largely dependent on the quality of the fellows as judged by the fellow filling out the survey and the operational talent of the Director to build out the curriculum to support an excellent experience.

Understanding product market fit is an acquired skill set. It’s not as simple as tying it to net revenue retention, GMV, user growth, or other core metrics from the industry you’re building in. I wish it were that easy – but it’s a combination of quantitative and qualitative data that you have a better chance of seeing if you’re close to the customer.

Most people, even career startup journeymen, will never see true product market fit. And, even if they see it once, it doesn’t make it any easier to evaluate whether you have it in future startups. Even successful founders like Justin Kan, who sold Twitch to Amazon for $1 billion, misinterpreted product market fit for his next startup, Atrium, and failed.

You are likely dead if you misinterpret your product market fit at a startup with expectations of exponential growth. The only hypergrowth startup I’ve seen pull this pivot off in recent years is Brex. I talk about how they iterated here (#5 – under “ability to iterate”) and correctly read their fleeting product market fit. You are still in trouble at a startup with linear growth, but you have a better chance and usually more time (since you’re burning less money) to try to find it again for your product or a new one.

Stay skeptical and aware of your product market fit.

The environment – everything is on fire all the time

dog cartoon saying "this is fine" while sitting in burning house

In a hypergrowth startup, everything will feel like it’s on fire all the time. This is normal. Startups undergoing this growth rate tend to outgrow processes, infrastructure, and even new teammates in months, not years.

At On Deck, I remember early on experiencing this as it always felt like a constant battle of whether I could update the documentation of new improvements and experiments within admissions in time for new fellowships to use for their inaugural cohorts. It usually meant more work for me if I didn’t update it in time.

When reflecting on operating in this environment, there are several things that I would recommend:

  • Sometimes let the fire burn. It’s natural to want to fix things if you have strong ownership of your work, but sometimes it’s more important to focus on the larger problem down the road than the one in front of you today.
  • Build systems that work for today and a year from now. The best operators can see around the corner and design systems that will scale to meet additional demand.
  • Understand the root of the problem. Take the time to understand why the fire started in the first place and set up safeguards so they don’t happen again.

And most importantly:

  • Every startup has fires. When times get tough, remember fires from hypergrowth > fires from no growth.

Community as a business

S curve for # of people in the community on x axis and community's utility on the y axis

This learning is specific to On Deck and other community-first businesses. I created this chart – that’s directionally similar to the one that David Booth, On Deck’s CEO, shared at a meeting, to sum up how to build a healthy and valuable community that scales. It’s the one graphic etched in my memory.

In this example, I will reference a founder community because it’s the persona I understand best. Communities for job seekers, capital allocators, operators, and creators will also follow the graph, but with different nuances.

In the early days, it’s relatively simple to control the quality of a founder community by being very selective and hand-picking people from your network who you know are good. Once people in the community start to accomplish impressive feats like raising money from well-respected investors, the credential of being in the community starts to become valuable as it provides signal to help other founders in the community raise money, make key early hires, and land deals with enterprise customers. Basically, people think you’re good just for being associated with the community.

When the community grows, you lose the exclusivity you had in the early days. As a result, people will claim that the quality of the network has worsened because you’re being less selective. In most cases, you are being less selective unless the number of applications you’re receiving to your community grows at a similar rate to the number of people you’re admitting into the community. And even then, people will still believe this. Even communities with more brand power, like Y Combinator (YC), received pushback from alums when they decided to increase their cohort size.

What YC and other communities that have scaled successfully figured out is that as the community scales, you need to increase the actual utility of the community. YC did this by productizing the things that founders would have a better chance of getting with the added liquidity – co-founders (YC Co-Founder Matching Platform), jobs (YC Work at a Startup), data on investors, insights from other YC founders (Bookface), distribution (Hacker News, Launch YC), etc.

The entire growth curve shifts to the left in a startup with exponential growth expectations. You have less time to scale the community’s utility to match the number of people in the community. You also don’t get the benefit of having significant enough outcomes to anchor your brand like YC or Stanford had before they scaled their communities.

Other venture-backed communities like Chief are facing similar challenges. Bootstrapped communities like Hampton should fare better off when it comes to scaling as long as they are patient. They have no external pressure or expectations from investors who invested in a hypergrowth startup to take more people into the community for the sake of revenue or other arbitrary metrics and, as a result, have more time to build the products needed to increase the community’s utility as it scales.

Other challenges community as a business has that I saw firsthand at On Deck were the operational resources and tooling required to run and scale them, bad actors, adverse network effects, forum moderation, unit economics, and more.

Community as a business is tough to make work on a venture scale timeline.

Asymmetric bets

chart showing asymmetric upside vs loss of invested capital
Image credit: Alex Oppenheimer

If you know what the power law is, then you know small wins don’t mean much in the land of venture-backed startups. You need a multi-billion dollar exit for everyone to make money. With this expectation comes a certain level of risk a startup needs to be comfortable taking. 

I refer to these risks as asymmetric bets. The upside from each bet you make needs to improve your chances of getting clarity and derisking the feasibility of your path to building a multi-billion dollar startup. Asymmetric bets apply to product decisions, who you hire, what distribution channels you build out, and more.

These bets must add up so your startup has more leverage to create a better product and user experience. The ultimate goal is to capture more value and build a defensible business model.

At On Deck, we built a lot of leverage from early bets but failed to capitalize on it. We had a ton of data on what founders needed, what startups job seekers were applying to, what resources were valuable to people wanting to accelerate their careers but couldn’t build the products that would align with what they needed most, and where we could capture the most value.

I think part of the reason for not executing on this was that it was tough to uncover and build conviction in the insights we were collecting across 10s of tools we were using to manage each fellowship, and we defaulted too much to what others had done before us.

If we could do it all over again, I wish we had made more asymmetric bets from first principles on things that we felt were valuable to our customer base, not what others have found in the past to be valuable to theirs.

When it comes to making asymmetric bets, always default to first principles and innovate.

On learning and reflection – enjoy the ride

team picture in the woods in front of our wooden lodge cabin
On Deck’s team retreat in Lake Tahoe

Most of my learnings above are written from a place of regret. It actually took me many months before I fully reflected on my experience because I knew I would look back and focus on the things that I could have done better or would have done differently in hindsight rather than all the good moments—one of my many character traits that can both be good and toxic.

But, even though we didn’t get the outcome we all wanted, I’m very positive about my time at On Deck. First and foremost, I met incredible people – coworkers, managers, fellows – that I’d love to team up with again in the future. Second, getting to experience hypergrowth at least once in your career is a privilege. The learning rate is unparalleled because you see the feedback loop from your decisions much quicker than in a linear growth environment.

Startups are complex, but hypergrowth startups are especially so. Joining a hypergrowth startup can be a career trajectory-changing move for the right person. But, make sure it’s something you want as it will require personal sacrifices. I’m grateful to Erik Torenberg, David Booth, Julian Weisser, and Brandon Taleisnik, who interviewed me and allowed me to work at On Deck.

Enjoy the ride.

Read next