The hiring process steps behind near-zero attrition, and why remote is its own function: Allan Christensen of Doist

Doist's hiring process steps are unremarkable on paper: at least three interviews, a test task, and a numeric score. What makes them worth studying is the result. Over the company's first seven years, four people voluntarily left.

Allan Christensen is COO of Doist, the company behind Todoist, a task manager with more than 400,000 customers, and Twist, an asynchronous communication platform. Doist has been remote since 2007, with people in more than 35 countries, and is that rare software company that grew without institutional investment.

He is in Copenhagen. He also began his career building pipe organs.

Pipe organs to Chile to a to-do app

Christensen's grandfather founded a pipe organ building business, and his four sons joined it, one of them Christensen's father. So when Christensen began his apprenticeship in 1996 he was the third generation, with some expectation that his turn to lead might come.

It did not, because he already knew he wanted to live abroad and learn other languages. As his apprenticeship finished he applied to three pipe organ companies in California and moved to the United States in 2000, to work in Santa Cruz.

The detail he enjoys: he regularly drove over Highway 17 to San Jose without knowing what Silicon Valley was, and only realized a decade later, once he was in technology, that he had lived there.

After six years of building organs he went back to university for an engineering degree, and started a side business selling football merchandise online, because he had always wanted to run something. Then Berlin, an MBA, and a procurement internship at Volkswagen. A Chilean classmate invited him to write his thesis for his company in Santiago, and after a couple of years in Germany he went.

Which is where he met Amir Salihefendić, the Danish-Bosnian founder of Doist, who was in Chile with the Start-Up Chile program.

The meeting story is the best origin story on the show. They first met at a startup party, which Christensen attended largely to speak English. Then came the annual expat football match between the British and the Danish, and Christensen needed players. He knew Amir was Danish and played.

On the drive to the match, the driver asked Christensen whether he had found work after his thesis. He said no. Amir, in the back seat, overheard, and asked what his strengths were. Christensen half-joked that he would skip to his weakness and get it over with. They laughed, and Amir asked nothing further.

Ten days later an email arrived asking whether he wanted to join as a business developer for a to-do app, and Christensen realized he had been in a job interview. They started meeting at a Starbucks.

For the record, they were on the same team against the British, so he could not have let Amir win.

The COO scope, and how it shrank

The division of labor was obvious from day one. Amir is a Python developer and was doing everything except support, for which he had hired one person in Poland. Christensen did everything else.

That meant business development, marketing and public relations outreach, some of it sent from Amir's address at Todoist. Their first TechCrunch coverage came from being the first to-do app supporting HTML5, which he describes as a big moment. Plus recruiting, finance, and support when needed.

The pattern he describes will be familiar to any early operator: he owns it until it reaches a level of complexity where they decide they need a professional, then they hire someone to lead it. His scope has narrowed steadily over the years as a result.

What sits with him now: people operations, customer success, finance, compliance, internal project management, and a head of remote.

Why remote is a function rather than part of HR

Michael's question is a good one, because most companies that take remote seriously still park it inside the people team.

Part of the answer is opportunistic. Chase Warrington had been head of business development, was an American who had lived in Europe for five or six years, and had started a podcast about living abroad. When they discussed how his role might evolve, the fit was obvious.

The strategic reason is more interesting. Doist's stated vision is to build the future of work, and the way they work is the core of that. Before the pandemic, telling recruits they could work anywhere with an internet connection was a real edge. When remote became a commodity, that edge eroded, and Christensen's read was that unless they kept pioneering it they would become one of the many.

So investing in a dedicated function was how they stayed ahead of a curve they had helped create.

And it is not only remote. It is asynchronous work, which is what Twist exists for.

What the role actually does internally is concrete. Monthly hangouts that mix people into groups of three for an hour of conversation about anything other than work. And the retreat program.

Three kinds of retreat

Doist runs three types.

The big annual company-wide retreat. Mini retreats that are function-specific, where small teams sometimes bundle together, so finance, people ops and product might travel as a group, while the Apple and Android teams each go on their own. And an annual leadership retreat.

The operational detail worth stealing is that Chase built templates for roughly 15 locations around the world, each with the full package worked out: the coworking setup, where to stay, what to do. Choosing a destination stops being a project each time.

What Twist exists to prevent

The origin of the product explains the working style.

Doist had been using another project management platform, decided to stop investing in it, and looked at where everyone else was going, which was Slack. They tried it, and Christensen's verdict is blunt: real-time chat is a conveyor belt with information coming at you constantly, and it was not them.

So they built something shaped to their own needs.

His definition of asynchronous work is that you respond when it fits your workflow, which is what protects the time to do deep work. He cites research suggesting it takes something like 23 minutes to return to flow after an interruption.

The design consequence is deliberate: there is no green presence indicator in Twist showing that you are online or typing. The intent is to give people time to think.

Where the line between sync and async sits

Christensen's rule starts from a remote-first premise: decisions need to be documented online. Offline discussions should be recapped so everyone has the same context, which they do not do by default but treat as a habit for anything important.

So anything containing a decision, a discussion, or the reasoning behind a solution goes into threads.

Synchronous time is for one-on-ones and team meetings, and even those are documented afterward with decisions recorded. Twist does have a real-time chat feature for smaller things.

His characterization of what synchronous conversation is actually good for: brainstorming and working out how to proceed. Then you take it back to async as a discussion or a proposal.

The line I would put on a wall: decisions are rarely made in meetings, and when they are, they are still posted to the platform so they can be challenged.

Why so many companies went back to the office

Michael sets up the macro picture from Emma Goldberg's reporting in the New York Times, which shows the share of businesses whose employees rarely or never work from home has returned close to its pre-pandemic level, despite the enormous shift during the pandemic itself.

Christensen's answer starts with an admission rather than a defense. People are social, and remote is not for everyone. Doist learned this the hard way, having hired genuinely talented developers who did not thrive, primarily because they lacked proximity to a mentor.

He also acknowledges Doist is at the extreme end of the scale, and that its position was not chosen on principle. They were in Chile, none of them wanted to stay indefinitely, they did not speak Spanish, and the people they hired through the startup program were also going to leave. Remote was not optional. Every process since has been built around it, so it now feels natural.

His prediction for everyone else is unglamorous and probably right. The optimum for the average company is somewhere in the middle. He does not expect everyone to want full remote, but the flexibility to work a day or two from home is here to stay.

And he is honest about the cost. When you go remote, things move more slowly. The quick conversation across the table now requires a meeting or has to be written down. What you get back is that anything written and shared internally can be referenced later without re-explaining it. You trade speed for control over your time, and he thinks the trade appeals to more people over the long run.

The hiring process steps

This is the section to take notes on.

Christensen was the last interviewer for the first fifty or so hires, and enjoyed it, partly because he got to tell the whole story. Bootstrapped meant they could not compete on salary or benefits, so the pitch was different: a flat hierarchy where the best argument wins, and the chance to have real impact if the project worked.

The framework came from Laszlo Bock's Work Rules, specifically the practice of scoring candidates on a scale of one to four, with a threshold average required to be considered hireable. Doist built its own version.

What they were protecting is stated plainly: get the right people on the bus in the right seats and you can build anything. And if someone joins who does not reflect the culture, deal with it, so the company does not drift onto the wrong cultural trajectory.

The process itself: at least three interviews plus a test task, then scoring. Scoring above the threshold means hireable, which is not the same as being hired. It opens a conversation about the pros and cons.

The evidence he offers is the retention number. Across the company's first seven years, four people voluntarily left.

As the company grew and functional heads were empowered, he moved out of hiring committees. He describes the loss with some feeling, and his self-assessment is worth quoting: he cannot build products, that is not his thing, but culture and people are his currency.

The retreat that told him it was working

His illustration of what the hiring produced is a three-year sequence at the annual offsite.

The first was in Menorca, about 30 people, and overwhelming, because he had worked with these colleagues for years and met almost none of them.

The second year they were 40, and you could see people grouping by language: the Spanish together, the Portuguese together, the Russians together.

By Iceland the year after, people were mingling across cultures, languages and functions. And by the third, people boarded the bus and sat wherever there was a free seat, did the same at restaurants, with what he describes as genuine curiosity about the range of people in the room rather than any instinct to cluster.

That is when it clicked for him that they had built something unusual.

Career progression without a hierarchy to climb

With a flat structure, Michael asks how anyone progresses.

Christensen is candid that this took several attempts. As recently as 2019 there was no career framework at all. And his observation about introducing one is honest: once you add it, you are asking for trouble, in the good way.

The structure has four levels: C level, heads, leads, and individual contributors. They built their own mastery track with five levels across various functions, but it remained subjective, felt dull, and did not excite anyone.

They have since adopted Pando. Each person now has a rubric with competencies and levels, can request feedback, and can log goals and achievements, which gives managers far better material on how someone is actually developing.

The visibility problem stable squads created

The reason the framework mattered is a specific structural side effect, and it is a good cautionary tale.

Doist works in stable squads. You are on a squad for a project, usually six months, and your manager may not be on it.

Previously they rotated squads roughly monthly, but lost so much to context switching that they moved to stable six-month squads at the start of the previous year.

Which surfaced a problem nobody had predicted. Managers reported that they could see communication and contribution in Twist, but for an engineer they could not see what the code looked like or help with it. Working remotely, you cannot walk over and ask what someone is working on.

Before the new framework, the compensating mechanisms were mandatory monthly one-on-ones where the individual had to bring things they wanted feedback on, plus squad leads offering input. But there was no system, so it depended on someone reaching out.

Two things hold it together culturally. Independence is a core value, and they hire proactive people who pick up the ball. And they have a hands-on manager philosophy, where every manager is expected to be working on projects themselves.

They are now adding forced monthly feedback sessions to generate the content that serves both the individual's growth and the manager's ability to help.

The trademark case

Christensen's crazy story is about being slow to register trademarks in the EU.

In 2018 two separate accusations of trademark infringement arrived from two different people in Switzerland at the same time, which was itself odd. The first came from a woman claiming her surname was Todoist, and who had coincidentally registered both Todoist and Doist as Swiss trademarks the previous year. They settled that quickly for a small amount.

Assuming the second was connected, they expected the same. It was not.

The second party claimed to have registered Twist in the EU in 2010, and asked for no damages, only that Doist pull the Twist app from the app store. Digging in, Doist found they did not hold the EU trademark at all, only a Swiss one registered the previous year. Doist promptly registered it in the EU. They also discovered Siemens held it for Germany, and acquired that.

Everything so far had been conducted through lawyers, so Doist tried reaching out directly on the assumption that reasonable people could sort it out.

The reply expressed disappointment that Doist had launched Twist in Switzerland without permission, and requested 220,000 euros, with a parenthetical clarifying that this was not in dollars.

Doist declined, having very few Swiss customers. In court they learned the other side had no product and no business, just a single page with a mockup that resembled a software product, essentially a flyer.

They made one more attempt, offering 10,000, which apparently insulted the other party so much that they raised the demand to 350,000.

So it went the whole distance. Amir traveled to Geneva for a hearing. The court ruled in Doist's favor and declared the trademark registration fraudulent and invalid. The other side attempted an appeal to the Swiss Supreme Court, which was rejected.

Christensen's takeaway is how far people will go without a case.

Michael's parallel is instructive in the other direction. Facing a similar situation early on against a company known to be litigious, he simply changed the product name.

The 5 things I took away from this conversation

1. Hireable is not the same as hired. Doist scores candidates and treats clearing the bar as permission to have the real conversation, not as a decision. That separation is what stops a strong interview performance from carrying someone through on momentum.

2. Making remote a function is how you keep an advantage you already have. Allan's reasoning is subtle. The remote edge was real when it was rare, and it eroded when everyone had it. Rather than accept that, they invested in the thing that had made them different. Most companies would have folded it into HR and let it become maintenance.

3. Decisions get posted even when they are made in a meeting. Not so there is a record, but so they can be challenged by people who were not in the room. That is a different and better reason for writing things down.

4. Stable squads made people more productive and their managers blind. This is the operational trap I will watch for. Reducing context switching was correct, and it quietly broke the feedback loop, because a manager who is not on your squad cannot see your work. The fix had to be a system, not more diligence.

5. Three years of watching where people sit. Allan's measure of whether the culture was real was not a survey. It was noticing that by the third offsite, people got on the bus and sat wherever there was a seat instead of clustering by language. Slower than an engagement score, and considerably more convincing.

FAQ

What are the steps in a good hiring process? Doist's version is at least three interviews plus a test task, followed by scoring each candidate on a numeric scale adapted from Laszlo Bock's Work Rules. A candidate above the threshold is considered hireable, which triggers a discussion of pros and cons rather than an automatic offer.

How do you hire for culture without it becoming bias? By making the criteria explicit and scored rather than leaving them to impression. Christensen's stated aim was to keep the company from drifting onto the wrong cultural trajectory, which requires knowing what the culture is specifically enough to assess against it.

What does a head of remote actually do? At Doist, the role covers both external advocacy and internal optimization: designing the practices that make a distributed company work, running the monthly small-group hangouts that connect people across teams, and building and running the retreat program, including templated logistics for around 15 locations.

How do you handle career progression in a flat organization? Doist has four levels and initially built its own mastery track, which proved too subjective to motivate anyone. They now use a platform-based framework where each person has a competency rubric, can request feedback, and can log goals and achievements, giving managers concrete material to work from.

How do you give feedback when a manager cannot see the work? This is a specific problem created by stable project squads, where your manager may not be on your team. Doist's answer combines mandatory monthly one-on-ones, squad leads providing input, and a documented competency framework, plus scheduled feedback sessions so the loop does not depend on individuals remembering to ask.

Also mentioned

• Doist, and its products Todoist and Twist
• Work Rules by Laszlo Bock, the source of Doist's scoring approach
• Pando, the career framework platform built for remote companies
• Start-Up Chile, where Amir Salihefendić was when the two met
• Chase Warrington, Doist's head of remote, and the retreat templates behind the program
• Emma Goldberg's New York Times reporting on the return to the office

Listen to the full episode

Allan Christensen on Between Two COO's

Between Two COO's is hosted by Michael Koenig. Subscribe on Apple Podcasts, Spotify, or wherever you listen.