Expensify’s Firing Strategy

David Barrett —  July 11, 2016 — 8 Comments


Nobody likes to talk about firing.  It’s not something to celebrate: if you need to fire someone it means you screwed up.  Either you hired the wrong person, or — more common — you hired the right person, but failed to enable their success.  Either way, the blame falls on the company (not the individual), so it’s no surprise that companies tend to avoid talking about their failures.

But despite that discomfort, firing is an important topic because in the long run, careful application of firing is actually more important than hiring.  To understand why, consider this simple chart:

image (3).png

The red line indicates how much total productivity your team has to give.  All things being equal, your potential productivity increases linearly — every time you hire one person, they bring one more “unit” of potential productivity.  However, in a collaborative team where everybody talks with everybody, it also means everybody distracts everybody — just a little bit.  This little bit adds up over time.  Even worse, while productivity increases linearly, overhead increases exponentially (orange line): with every new hire, not only do you distract one more person, but one more person distracts you.  The net of both of those is that total output is equal to your potential productivity minus your overhead — meaning after a certain point, the total output actually goes down with each hire.

Why Hiring Reduces Output

This might sound impossible, but think back to when your company was really small.  Remember when every new hire felt like they made the company just jump forward in productivity?  But over time, it just feels like… something is dragging on the company.  The hiring keeps happening, and even if you maintain very high standards and everybody is working super hard (both of which get increasingly difficult at scale), it just feels like things slow down.  This feeling isn’t in your mind: it’s because the unavoidable overhead of collaboration comes at the expense of your ability to ship product out the door.

Indeed, if we throw arbitrary numbers at this and assume everybody contributes one “unit” of productivity, but at the cost of 1% overhead of everybody they collaborate with, this means that the most productive team you can possibly have is limited to 51 people — and they only have the total output of 25.  Every person you add to the team after that actually reduces total output, to the point where if you have 100 people, none will have any time to ship product as they’ll all be completely consumed in collaborating with their peers.

Sound impossible? Think of it differently: 2 people can pick a place to eat rather quickly; 10 is hard, but doable. 100 — impossible. Nobody will ever eat.

Like most things, there’s a simple solution to this problem: keep your teams really small, thereby minimizing the collaborative overhead.  This sounds really great, and nearly everywhere you can read of the virtues of organizing your company as a series of small teams, each with a very clear mandate and a very specific focus.

What people tend to forget in that organization is: who makes the teams, and who decides what they do?  After all, the point of the small teams is to eliminate cross-team communication — so how do you ensure all those teams are working on the right things, and doing so in a way that ultimately comes together to make a single product?  The answer is: hierarchy.  Behind every small team company is a large management hierarchy making all the big decisions — meaning all these great teams are inexorably constrained to only dealing with small problems.

Don’t get me wrong: “small” is relative.  A small problem might be “scale this cluster of servers up by 100x” — and that can be a real challenge and a lot of fun.  But in the grand scheme of things, any company that truly needs to scale some system up by 100x must by definition be undergoing such a rapid amount of growth that there are much, much bigger problems to be faced — problems like “which new features should we build?  which should we cut?  which markets should we charge into versus retreat from?”  Or even “which clusters of servers should we scale up 100x, versus just cut entirely (along with the people who maintain them) and migrate to a third party?”  Those are the “big” questions, and by definition, any company organized around small teams has intentionally removed most of the brains of the organization from any room where those discussions take place.

Said another way: you can’t have a small team laser focused on a small problem, and also expect them to have the time, context, or incentive to also help with the big problems.  You can’t have it both ways.

Expensify is #OneTeamOneDream

We like to have our cake and eat it too, so we take an entirely different solution to the “small teams / big problems” conundrum: one large team.  Everybody is actively encouraged to collaborate with everybody else in the company, with such techniques as:

  • Everybody is subscribed to (almost) every internal mailing list on day one; it’s up to you to decide what you want to filter out.
  • Everybody is welcome to (almost) every meeting; it’s up to you to decide if it’s worth your time to be there.
  • Everybody is invited to our annual offshore (and holiday party, and onshore — with travel covered for you and your significant other from anywhere in the world), with active attempts to randomize seating arrangements to maximize cross-team exposure.
  • Everybody participates in the hiring process, and contributes feedback on the candidate’s progress through the day.
  • Everybody is compensated using the same process, without regard to team or title.

That all sounds great, in theory.  It makes for a workplace that is incredibly innovative and collaborative, where the best ideas are distilled from the greatest number of minds.  It’s the key to why we are the technology leader in our space, and growing faster than the rest of our market, while also being profitable and having an amazing time doing it.

But recall the problem with large teams: the exponential increase in overhead caused by adding someone to a fully-connected collaborating team can, at some point, equal or even exceed the linear productivity the new hire adds to the team.  Keeping teams small is an easy way to avoid this, at the cost of reducing collaboration and creating a thick management hierarchy.  But Expensify solves this a different way by building one large team consisting of people who are:

  • Unusually talented, so as to obtain the maximum linear productivity gain
  • Unusually humble, so as to minimize the exponential increase in overhead

(Said another way, people who will Get Shit Done, and Not Ruin It for Everyone Else.)

Granted, we don’t really solve the fundamental constraint that drives a finite team size — the basic math behind that is inescapable.  We just push the problem far enough into the future that we needn’t solve it today, because even small changes can have profound effects here.

Going back to the previous chart, if you have a team of people who are even 25% more productive than the average — with 25% less overhead — you can grow a team that produces 2x the net output before succumbing to overhead:

image (4).png

Granted, now you can have a 84 outputting with the productivity of 53 — that’s better, but still not great.  But then again, we only assumed 25% more productive, and only 25% less overhead.  The real question is: just how much more productive can a person be, and how little overhead is possible?

The 10x Rule

Everybody has heard the common trope: great programmers are 10x more productive than average programmers.  We’re told this in college, we read it online, and we even see these superstar people around us in our careers.  So when I started Expensify, I thought:

Why not only hire people 10x better than everyone else?

It’s such a simple idea, it seems deceptively obvious.  After all, if you could do it, then your productivity chart would look like this:

image (5).png

If you have the right team, your productivity is so high, the overhead of collaboration is insignificant.  You can have a team of pretty much any size, and they can all talk to each other freely — sharing the best ideas, learning from each other.  There is no need to create a thick hierarchy to make decisions for them, because they are capable of making big decisions themselves.

The key questions are: is this even possible, and if so, how?

Building a 10x Team

At Expensify, we’re betting it is possible, and that it can be accomplished with two main tactics.  The first tactic involves identifying superstars early in their career, and then backing them to the hilt.  Attracting, retaining, and inspiring people who can perform at 10x their peers is critical to creating the steep linear increase in productivity necessary to get the team off the ground.

However, as necessary as it is to get the right people in the door, that alone isn’t sufficient to ensure long term team growth.  Because no matter how steep your linear productivity increase, it will ultimately succumb to the exponential increase in overhead if you leave it unchecked.

This is why over time, even more important than your ability to hire, is your discipline to identify those who are driving up the exponential overhead curve — single-handedly draining more productivity from the organization than they contribute themselves — and let them go.

When you have a small team, one bad apple isn’t so bad: maybe they sap 2% of everybody’s attention, but if there are only 10 people, that’s only 20% of one person’s time consumed.  If they’re at least as good as everyone else, they are still bringing 80% of one person’s productivity to the table, paying for themselves and more.  For this reason, small teams are surprisingly tolerant of disruptive people.

But as the team grows, 2% of 50 people’s time adds up to 100% of one person — and that one person is draining as much from the company’s productivity as they individually bring.  For a while, that’s actually tolerable: these people are seen as merely “coasting”, where they cost as much as they bring, but no more.  As the team grows, however, 2% of 100 people’s time adds up to 200% of one person: someone who is “coasting” now will inevitably create serious drag later.

Furthermore, in practice, the bad apples don’t cost just 2% — they cost something like 10%, of everybody’s time, and upwards of 70% of some people’s time.  You try so hard to ignore this because oftentimes, the most disruptive people are also some of the greatest individual contributors.  You want it to work so badly, which is why you invest so much time (and allow yourself to get so distracted) trying to make it work.

But that’s ultimately a losing battle, because as the team grows, no seriously disruptive individual can contribute enough to compensate for the amount of accumulated distraction they create in the team overall.

So while starting a 10x team is largely a matter of hiring, growing a 10x team is as much a matter of firing — identifying those who are creating more disruption than they are bringing, making early attempts to correct the problem, recognizing quickly if those efforts aren’t working, and acting decisively on the results.

The Performance Improvement Plan (PIP)

When we had hired our 100th person into Expensify, over 60 were still working at the company.  That means up to that point, 40 people had either quit or been fired.  (And of those who quit, nearly all did so when they found their peers increasingly unwilling to tolerate to their disruptive attitudes, and thus less fun to work with — misery loves company, and if your peers aren’t as miserable as you, you leave to find others who are.)

At the start, the terminations were pretty messy.  We didn’t have a good sense of what was our fault, their fault, how to anticipate and avoid problems coming down the pipeline, or how to deal effectively with problems when they appeared.  There were many dark days.

But over time, we started to see patterns.  Each new problem seemed similar to a problem we’d seen it the past.  It enabled us to anticipate situations earlier, and thus deal with them proactively rather than just reacting after they arrived.  Most important of all: we started to understand the root causes behind the problems, and thus how to engage with someone early to either bring them back onto a constructive path — or skip ahead in line and encourage them to find another opportunity (with a substantial severance package to sweeten the deal).  All of this has culminated in a technique we call a “PIP”, short for Performance Improvement Plan.

When we see someone following a path we’ve seen before, a path we know isn’t going to lead to great places for them or us, we apply the following general process:

  1. Do we genuinely feel there is a chance to correct the problem to everybody’s satisfaction?
    1. If no, let them go right away, with an appropriately generous severance package based on the circumstance.
    2. If yes, have we already dropped gentle hints and given them the ability to self-correct, such that they won’t be surprised if confronted on the issue?
      1. If no, raise the problem informally in our next 1:1.  (We refer to this as a “pre-PIP talk”, meaning a conversation that is laying the foundation for them to understand, accept, and successfully respond to a full PIP if we need to do it.)
      2. If yes, have we already outlined in writing the exact problem explicitly?
        1. If no, provide a written PIP, which includes a detailed problem description, probationary period, metrics for success to satisfy the PIP, and consequences of failure.  Then, ask: are you able and willing to go through it?
          1. If no, let them go right away, with a generous severance.  It’s a difficult, but incredibly mature choice to recognize that the odds of succeeding with the PIP are too low to justify taking everybody’s time.
          2. If yes, start the PIP and have weekly updates.  Make it clear that a PIP is being held out as an opportunity to stay, not merely a chance they’ll get fired.  This is a subtle nuance, but an important one: a PIP is a serious investment in the individual, and if substantial effort and progress isn’t being shown, they will be let go prior to the PIP’s conclusion.
        2. If yes, then they’re already aware of the problem and have been given the opportunity to correct it, and will be let go without another PIP.

This process is still evolving, and each time we go through it, it gets a little more refined.  Furthermore, as we get confident in the effectiveness of the program, we are tending to employ it earlier and more often — always with a primary aim to create mutual success, but with a backup plan to let them go if success remains elusive.

PIPs are never fun, for anyone involved.  But they are critical to keeping that exponential overhead curve from rearing its ugly head and undermining the incredible productivity of the 10x team overall.  Because at Expensify, it’s not enough to merely succeed at building a successful business.  We’ve already done that.

We want to build a successful business AND continue to attract, retain, and inspire the best team in the world — a team that gets better over time, rather than accepting the dystopian attitude of “every day is better than the next”.

I fully expect our most glorious days are ahead of us, not behind us, and I can’t wait to get there.  If you’d like to be there for those glory days, I’d be honored if you’d consider a gig with Expensify.  Drop us a line at jobs@expensify.com — I can’t wait to hear from you!


David Barrett


Founder of Expensify, destroyer of expense reports, and savior to frustrated employees worldwide.

8 responses to Expensify’s Firing Strategy


    One of the few original takes I have read on this topic. Thank you for posting! A few questions I have: 1) Does the official PIP stresses the employee to the degree where it is hard to perform, and thus improvement is doomed? Roughly what % of employees who go on a PIP emerge successful vs leaving?
    2) Do you think 10x performance is possible in all roles?
    3) Do you have an opinion on a “healthy” level of turnover?

    PS – fantastic product, congrats!


    Re: “Does the official PIP stresses the employee to the degree where it is hard to perform” – That’s left up to the individual to decide. However, we try to make it a real choice by providing the option of a generous severance if they choose not to go forward. This way people only go forward if they really believe it can work, and are committing to making it work. And because of that confidence/commitment, the organization can do its part to make it work too. The worst is if someone goes forward in a half-hearted way while casually looking for another job; that just wastes everybody’s time.

    Re: “Do you think 10x performance is possible in all roles?” – First, “10x” isn’t a hard and fast number — it’s for illustrative purposes. But do I believe every role has top performers that can dramatically outperform the average or worst performers? Definitely.

    Re: “Do you have an opinion on a “healthy” level of turnover?” – I’m not sure of the data here, honestly, and I’d welcome any thoughts you have on this. Thanks for asking!


    Do you have exit interviews? Do you have appeals? How can you be sure “disruptive people” aren’t just like, people who bring up workplace issues? How do you, a cis white man, account for intrinsic bias? Have you listened to people who are not cis white men share their stories about being fired because they “made people uncomfortable” by objecting to their colleagues’ offensive behavior? There is so much nuance here, I’m struggling to understand how you manage to make decisions that affect people’s careers without having considered it.


    These are all good, difficult questions without easy answers.

    Re: “Do you have exit interviews?” – Yes, it’s obviously important to get the feedback of the person leaving, whether they’re quitting or being let go.

    Re: “Do you have appeals?” – This is part of the Performance Improvement Process. We sit down with the employee and walk through all the concerns in detail, with a particular goal of understanding if there’s additional information or context that’s relevant that we didn’t know. It’s not a kneejerk thing — it’s a very methodical procedure to ensure everybody has all the information, and every opportunity to raise more.

    Re: “How can you be sure “disruptive people” aren’t just like, people who bring up workplace issues?” — Oh, it sounds like we have different ideas of what it means to be disruptive. Raising workplace issues is fine and healthy, even if they’re difficult concerns without clear resolution steps. I’m more referring to disruption coming from failure to collaborate effectively, such as not considering the opinions of others, or actively undermining the collaboration process itself.

    Re: “Have you listened to people who are not cis white men share their stories about being fired because they “made people uncomfortable” by objecting to their colleagues’ offensive behavior?” – Again, it sounds like you’re approaching this from the opposite angle of how it typically plays out in practice. This isn’t about stamping out dissent from those who raise concerns about co-workers. This is about identifying if someone is creating concerns in others, determining if those concerns are valid, and (if so) running a process to address those concerns. The best decisions come from a diverse team that feels safe in raising dissenting opinions in a respectful environment, and this is about ensuring everybody is aligned toward that goal.

    Re: “There is so much nuance here, I’m struggling to understand how you manage to make decisions that affect people’s careers without having considered it.” — Yes, I absolutely agree there is a tremendous amount of nuance here, and that it affects peoples’ careers in tangible ways. But so does inactivity. The reality is a collaborative environment is a very delicate, special place that needs to be protected. There are no simple answers or generic solutions; everything needs to be taken in isolation with the best principles of fairness in mind.

Trackbacks and Pingbacks:

  1. Why Is This Ad Promoting Getting Your Coworkers Fired? | Fast Company | Business + Innovation - July 15, 2016

    […] ad is the brainchild of expenses management startup Expensify. It is intended to recruit people to work for the […]

  2. Why Is This Ad Promoting Getting Your Coworkers Fired? – Fast Company | Farmissy - July 15, 2016

    […] ad is the brainchild of expenses management startup Expensify. It is intended to recruit people to work for the […]

  3. Why Is This Ad Promoting Getting Your Coworkers Fired? | Grabber - July 16, 2016

    […] ad is the brainchild of expenses management startup Expensify. It is intended to recruit people to work for the […]

  4. Startup Loves Firing People - August 15, 2016

    […] Expensify blog  […]

Have something to say? Share your thoughts with us!

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s