When Travis CI turned into a business with employees, one of our ideas was to not constrain people in how much time they take off for vacations. We didn't track the days people were taking off, and as the people running the company, we didn't actively encourage people to take times off. In short, we had an open vacation policy.

The cause was intended to be noble, as we didn't want to get into the way of people taking time off as much time as they need to recharge. I myself am a big fan of disconnecting for a vacation and staying away for more than just a few days to free the mind, gain new energy and fresh insights.

Two years later, this idea turned out to be a failure, and we're changing our vacation policy. Here's why.

Uncertainty about how much time would be okay to take off

When everyone keeps track of their own vacation days, two things can happen. They either forget about them completely, or they're uncertain about how much is really okay to use as vacation days.

Forgetting about them seems to be beneficial for a young startup company, at least on the surface. You want people to work as much as possible to push the product and company out of uncertain territory into profitability, right?

Wrong. What you will do is push people to the edge of burnout and unhappiness. They'll eventually leave your company.

This almost happened in ours, we pushed someone too far. They pulled the cord eventually, and we asked them to take off as much time as they need. We're sorry for this mistake, and we're thankful this person is still with us.

When people are uncertain about how many days it's okay to take off, you'll see curious things happen. People will hesitate to take a vacation as they don't want to seem like that person who's taking the most vacation days. It's a race to the bottom instead of a race towards a well rested and happy team.

I came across a passage in "Scaling up Excellence", an okay-ish but vague book on how to scale up a company (emphasis mine):

In Matthew May’s book The Laws of Subtraction, Markovitz describes how his team was burdened and annoyed by a convoluted HR system for managing vacation requests. He decided to ignore it and told his team “as long as they got their jobs done, I didn’t care how many vacation days they took each year.” It worked beautifully—he stopped wasting time on paperwork, his team felt respected, and they stopped gaming the system: “The number of vacation days that they took actually decreased.” Markovitz’s experiment succeeded because it created accountability. “My team’s focus shifted from figuring out how to beat the system to figuring out how to live up to the responsibility placed upon them."

I was horrified reading this, and it dawned on me how wrong we've approached our internal vacation policy. This text sums up exactly what's wrong with an open vacation policy. People take less time off, and it's celebrated as a success of giving people more responsibility.

Uncertainty about how many days are okay to take time off can also stir inequality. It can turn into a privilege for some people who may be more aggressive in taking vacations compared to people who feel like their work and their appreciation at work would suffer from being away for too long.

People would work on their vacation days

As part of my time working for a US company, I was exposed to a weird culture. When people announced they'd go on vacation, they'd tell everyone that they'll take their computer and phones with them, and that they will be available if anything comes up.

Earlier this year, I went on a three week vacation with my family. When we booked us a small house in France, my initial thought was: "How can I justify staying away from Berlin for so long? I know! I'm going to work while I'm there, at least for a few hours every day."

I've seen this happen in our company, and not just with me. The guilt of taking time off takes over, and you "just check in" or promise to be available if anything comes up. You respond to just one email or just one GitHub issue.

This ambiguity trickles through to everyone on your team. When someone starts checking in during their vacation, it lowers the bar for others to do it, and it increases the uncertainty of whether or not you should be checking in. When you as the leader in a company take vacations like that, you unknowningly set a bad example that others will feel compelled to follow.

Summing up the problem with checking in while you're on vacation, I quote from "Mission: Impossible II":

Mission briefing: "And Mr. Hunt, next time you go on holiday let us know where. This message will self-destruct in five seconds." Ethan Hunt: "If I let you know where I´m going, then I won´t be on holiday."

A vacation is a time to recharge, and your job as a company leader should be to remove any ambiguity of people thinking they're required to be available or reachable.

A company has to learn how to function when people are on vacation and unavailable, however important their role is.

After months of back and forth, I decided to do the only right thing I could think of. I took those three weeks in France off, fully and completely, without being reachable, and I told everyone about it upfront. And I encourage every single one of our employees to do the same.

The founders of the company only took little time off

In the early stages of a company, it can become all-consuming for the people with the biggest stake in it, the founders. Current technology culture celebrates people hustling hard, raising tons of funding, and in turn hustling even harder.

In short, we've set a bad example, and we ourselves didn't live up to the expectation of the open vacation policy. We took off less or no time in some years, always focusing on the hustle and on the idea that us being away would hurt the company or be a reason for stuff not getting done.

What did we change?

Starting in 2015, we've implemented a minimum vacation policy. Rather than giving no guideline on what's a good number of days to take off, everyone now has a required minimum of 25 (paid) vacation days per year, no matter what country they live in. When people want to take time off beyond that, that's good, and the minimum policy still allows for that. But it sets a lower barrier of days that we expect our employees to focus on their own well-being rather than work.

This policy is not just a guideline for our employees, it's mandatory for everyone, including the people who originally founded the company. As leaders, we need to set examples of what constitutes a healthy balance between work and life rather than give an example that life is all about the hustle.

Ensuring that everyone takes off the minimum number of days requires us to start tracking vacation days for everyone. Having numbers allows us to review everyone's vacation days on a regular basis, ensuring that the minimum time taken off is equal and that scheduling in vacation days is actively encouraged.

As Jesse Newland said in a talk at Monitorama:

Vacation is cheaper than severance and training.

We removed ambiguity of whether or not someone should check in by having explicit guidelines on what constitutes a vacation day and what doesn't. Our expectation is that when you're on vacation, you do everything but stuff that's related to Travis CI.

Instating an open vacation policy can be poison for your people's team and happiness, as it removes the lower barrier of what's an acceptable amount of time to be away and focus on recharging and your family, in short, your personal well-being. Your job as a company isn't to coerce your people into taking as little time off as possible, it's to make sure they have a good balance between work and life.

Our new minimum vacation policy is the first step to making up for our mistake, and we'll keep a good eye on how it works out.

My biggest of thanks to Jacob Kaplan-Moss for writing about the problems with an open vacation policy and possible solutions. His article has been great food for thought and a trigger for improvement in our own company's culture.

I'd love to hear about your experiences as a company leader, in particular about issues like vacation policies and team happiness. I'm also happy to talk about our culture at Travis CI. Email me!

I posted this on our company-internal blog at Travis CI to share progress on goals I've set myself over the past couple of months with my team.

They include some reflection on our development as a business and on what I got done and what I didn't manage to finish.

I figured I might as well publish these here as a public commitment to improving and to give some more insight as to how our company works.

Transparency and openness are some of the goals we aim for, and this fits in well with these ideals.

The Review

For the last three months, starting 1. September 2014, I had set myself three bigger goals to focus on for the following three months, ending on 30. November.

The goals were:

  • Set up our new office and move in with the Berlin team
  • Improve our customer onboarding process
  • Reduce customer churn by 0,5%

Part of setting goals is to have an honest retrospective on what got done and why something may not have gotten done, and what can be learned from that. That's the purpose of this post, to look at the goals, what motivated them, what I got done and where I failed and why.

Another thing about goals is that, while they give something bigger to work towards, their purpose doesn't have to be a finished result, but improvements. If a goal helps improve something, it's been worth setting, whether you've reached the specific target or not. It's not an excuse to set goals and not work towards them, but a guideline to help set and evaluate them.

New office

During our offsite the topic of a new office in Berlin came up. We quickly found a great space and signed a lease starting in September.

The process of getting it up and running has been slow, to say the least. We installed a kitchen, commissioned custom table tops for standing desks, but it took two months to install even a basic set of furniture, leaving the office empty and not usable at that time.

In October we got internet, a kitchen, and meeting room furniture. In November, the couch arrived, more desk frames and the remaining custom table tops.

While I was able to start working from our office by late November, it wasn't nearly ready for everyone on the Berlin team to work in.

Why has this goal not been met? I was travelling for two weeks in October, and I was preparing and giving a total of 8 talks in October and November, which took away a lot of time I could've devoted to the office.

By the first week, there's a full set of desks at the office, kitchen seating and a few monitors, so it's a close call, and progress has been made regardless of not reaching the finish line.

Looking back, we should've hired someone to manage the whole interior and getting everything set up and installed. Even then, it would've helped to be around and help make decisions.

In response to the excessive travel and time required for all those talks, I've also decided to go on a conference talk hiatus in 2015.

Improve our customer onboarding process

The idea was to improve our product's onboarding process to make it easier for our customers to get started, send better lifecycle emails, find ways to proactively reach out to customers struggling to get their builds set up.

I didn't get any idea implemented that I had on this over the last three months.

Why has this goal not been met? Mostly due to travels and conferences, I didn't have enough time to focus on it.

I also realized that this is not a task for one person, it's a product concern, which needs to be scheduled into the bigger plan for our teams to plan and implement. That needs to happen in a team that devotes time and resources towards this goal together.

Reduce churn by 0.5%

By the beginning of September, our churn was at 2.9%, according to Baremetrics. It went up and down since then, going up to 3.4% in late September, down to 2.75% in late October, now going back up to 3.5%.

There were a few things I got done that were related to this goal:

  • Allow resubscribing expired subscriptions. This wasn't possible before, so customers either didn't renew them when their credit card expired, or they had to email us to do it for them.
  • Add more people (admins in the organization) to the email that's send out when a credit card charge has failed. Churn is more than just people cancelling, it's also subscriptions expiring because of failed charges.
  • Visit customers in San Francisco. I met with 20 different companies while I was over there. While that has little direct impact on churn, it can help build a longer term relationship with customers, effectively increasing their lifetime value.

What had I planned beyond these things?

  • Add an upsell to our annual plans to the invoice emails.
  • Send an end-of-the-year email to advertise upgrading to the annual plan.

While the annual plans don't necessarily help to decrease churn, they can increase customer lifetime value and retention due to their annual nature.

The fluctuation in the churn, and it being higher now than it was in September suggest that this goal hasn't been met.

What I did get done makes sense regardless of this goal, though, as it can help in the longer term.

The fluctuation, at least over this short time frame, also suggests that maybe this is our normal churn and the pattern of it, and maybe churn is only one part that can be focused on. There's also lifetime value and monthly recurring revenue, which can be focused on.

Other things play into churn as well, making the current number a bit wonky, like our billing code creating duplicate customer entries in Stripe when a charge or something with the initial setup of the customer record failed.

Over the last three months, our MRR grew by 13.2%, whereas lifetime value is down 10.6%. Lifetime value is tied to churn, but there are other parameters to increase it, including improving your product, subscription plan upgrades, annual upsells, or changing your price structure.

For this month, I'd still like to focus on at least adding an annual plan upsell to our invoices, as the end of the year is a good opportunity to get some tax write-off for our customers with added expenses while helping increase their lifetime value for us.

What are my next goals?

My current time frame runs until February 16, when I'll go on an extended leave. Until then, I'd like to focus on two things:

  • Build a dedicated customer support team

    The motivation behind it has been discussed in a previous post. The goal is to help us define a hiring process, learn more about what it means to hire for diversity and what helps us find the right candidates.

  • Become a better manager

    Given how little experience I have, my assumption right now is that I'm bad at what I do, and that I need to actively work on getting better. Over the past couple of months I've been reading a lot of books on how other companies are being lead.

    But moving forward, I'm going to rely on outside experience to help in getting better in what I do or want to do. This is a long term goal, and one that's hard to measure, but it helps me guide in what kind of work I focus on.

    I'm starting to work with a management/leadership coach in December to help move forward with this goal.

Tags: goals

Listening to the “This American Life” episode on the GM/Toyota NUMMI plant recently, one particular part struck me as interesting when it comes to culture.

Culture is something that everyone would love to be able to easily replicate. Companies like Etsy, Netflix and others are forging ahead with openness, open source and empowering employees when it comes to their production systems.

NUMMI was an attempt to bring Toyota’s principles in building cars to General Motors, the automotive giant that was struggling hard in the eighties and was eventually bailed out by the American tax payers in 2009.

Toyota’s production line is famous for a simple tool, the Andon cord that allowed every worker on the factory floor to stop the assembly line whenever they encountered a problem. This empowered every employee to work towards a single goal: quality.

At NUMMI, this same system was implemented, and very successfully so. Every worker in the factory initially worked for two weeks with a team at Toyota in Japan to fully experience how teamworks looks like. It didn’t exist inside GM before NUMMI was conceived.

The Andon cord is an essential tool in learning and improving quality continuously. Every stop of the production line is an opportunity to learn and to improve the production process.

Before the NUMMI experiment, and in the rest of GM, the one goal is to never stop the production line. Quantity over quality, at all times.

Quality at NUMMI thrived, and GM looked into implementing this in more of their factories.

This experiment failed as there was a lot of resistance in management, amongst the workers and in the unions (all of whom had been fully onboard at NUMMI).

One bit in particular was interesting about the adoption issues.

The Andon cord was installed in other factories too, but when workers used it, they were reprimanded for stopping the production line. Managers were paid by volume of cars leaving the factory. In other factories, the cord was cut down so it was harder to reach.

I found this bit fascinating in so many ways, and it made me think about culture.

We’d love to just take a blueprint from another company and apply that to ours. But culture is something you need to work hard on, something that takes years of learning and improving to bring about, and it requires continuous nurturing to stay healthy.

You can’t just replicate culture.