Chief operating officer duties are set by the CEO: Catherine Stewart of Shippo

The most quotable line in this episode is one Catherine Stewart heard from someone else: in tech, COO is whatever job the CEO does not want to do. It is the cleanest explanation of why chief operating officer duties vary so wildly from company to company, and it comes with a practical instruction attached.

Stewart was chief operating officer at Shippo, the shipping software company for e-commerce businesses, where the business grew three times faster than the US e-commerce shipping market and the valuation went from $220 million to more than $1 billion in 18 months. Before that she was chief business officer at Automattic, the company behind WordPress, and before that she was at Facebook, where she helped launch the Facebook Audience Network.

Publishing, McKinsey, and a $16,500 offer

Stewart wanted to work in publishing. Her first offer was from Oxford University Press for $16,500, which was a standard salary at the time and, in New York City, not a livable one. She went to McKinsey instead, reasoning she could return to publishing later, which she eventually did on the business side.

McKinsey taught her the generalist skill she has used since: dropping into a company's operations and working out where to increase revenue or reduce cost. The range of projects was wide. A MySpace engagement was her first exposure to tech, back when MySpace was bigger than Facebook. She also worked on Pfizer clinical trials, a detail that closed a strange loop years later when her son participated in one as a test subject for the COVID vaccine.

At Facebook she wore several hats across product management, marketing and finance, having joined to work on the three year plan. That breadth is what led to Automattic. Matt Mullenweg's pitch was direct: he needed someone who could worry about the money so the rest of them did not have to. She took what she calls the commensurate jack of all trades role, COO or CBO or CRO depending on the label, and stayed seven years.

She is refreshingly clear about where her interest actually lies. Finance is not her passion, and when asked whether she wants it reporting to her the answer is usually no, though she has run finance teams several times and can. What she loves is go to market, things measurable in dollars and cents, and building revenue that is sustainable rather than merely present this quarter. That means marketing, sales, customer success, business development and partnerships, plus the functions that support them.

What actually drove the growth at Shippo

Stewart joined as COVID was reshaping how businesses operated, which was both the opportunity and the problem.

Her description of the operational reality is one every operator should sit with. When your customer count doubles or triples overnight, support tickets follow, your product's capacity is tested, and things break. The challenge was absorbing that volume while retaining it, rather than watching people try the product during a surge and move to a competitor.

Her assessment of the outperformance is measured. The growth was partly fueled by COVID, which lifted the whole market. Beating the market came from running a tighter ship and giving customers an experience worth staying for.

The first 30 days: 50 interviews, then a public commitment

The onboarding approach Stewart describes is the most directly copyable thing in this episode.

Her first 30 days were a listening tour of roughly 50 interviews over about three weeks. Mostly employees, plus customers. Critically, she also spoke to prospects and to customers of competitors, to understand why they had chosen someone else. That last group is the one most new executives skip.

She distilled everything into five themes, then took them to the board along with her own plan and, notably, her own goals. Here is what I heard, here is what I will personally do, here is how you should measure me at three months, six months and twelve. She notes the twelve month goal usually mattered most, because the meaningful initiatives take that long to show up in the numbers.

Reducing employee churn was one of those goals, and she is honest about the difficulty. The output is highly quantitative, but the causes are fuzzy and the fixes take time.

Then she gave the same presentation to the entire company, pitched slightly higher level. Here is what I heard from you. If I got something wrong, tell me. Here is what I plan to do, here is how you can keep me honest, and this is my commitment.

Her reasoning is worth quoting in substance. Priorities are only real if you can talk about them. If she wants other people held to metrics, she has to be held to hers, including saying plainly when six months have passed and the number is not where she said it would be, without finding a scapegoat. The underlying principle is one she compares to parenting: people do not do what you tell them to do, they do what they see you doing.

Change management starts with admitting the problem

Stewart's first rule of change management is acknowledging where you actually are.

Companies are tempted to sugarcoat the status quo. Employees are on the ground and know better. If you tell people something is not a problem while they experience it daily, you lose trust quickly. Her preference is to name the difficulty, apologize for it, and say it will take everyone to change it.

The reward for that honesty surprised her. There are always detractors, but a striking number of people have already been feeling exactly that and are relieved someone said it.

Her example is the cultural shift at Automattic toward monetization. The company was deeply committed to free open source software, which can sit uneasily with a for profit company's need to generate profit. What she and Mullenweg worked out was how to talk about revenue so people understood why it was good for the mission rather than a betrayal of it. The framing they landed on was the Walt Disney line: we do not make movies to make money, we make money to make more movies. You cannot hire engineers or acquire companies that strengthen the portfolio without a healthy revenue stream.

The reception was a vocal minority of detractors and a much larger group that was glad, including pockets of people who asked how they could help. Stewart values the public debate, but is clear about the limit. If someone is unhappy enough with the company's direction that they cannot be productive, it may not be the right place for them. Private disagreement while producing excellent work is entirely fine.

Attrition, and the number that actually matters

Stewart has seen both extremes. Automattic's attrition was very low. Shippo's was high. Her advice is to establish a baseline by asking VCs in your industry and other operators, since tech runs higher than most sectors. As rough anchors, she calls 40 to 50% high and 4 to 5% low.

The rate, though, is the less interesting number. What she watches is regrettable versus non regrettable departures. Some people prefer voluntary versus involuntary, which she agrees is useful because it tells you about hiring effectiveness rather than retention, but regrettable versus non regrettable is the more meaningful cut.

Her point about underperformance is the one operators most often get wrong. Keeping someone too long, whether the issue is attitude, work ethic or quality, is demoralizing to exactly the people you most want to keep. She puts the two failure modes on equal footing. Having the wrong people stay too long is as damaging as having the right people leave too soon.

What hypergrowth does to a team

Facebook was around 2,500 people when Stewart joined, roughly a year before the IPO, and was doubling headcount most years she was there. Doubling while holding the talent bar is genuinely hard, and she says some years went better than others.

One year did not. The company relaxed its standards, and it produced an internal meme that treated hiring cohorts like vintages of wine, with advice circulating about avoiding engineers hired in the year the floodgates opened. The cost shows up later, either in performance management or in products built the wrong way.

She adds the compounding mechanism that makes it worse. In hypergrowth, the first thing a new hire does is hire. So someone who is not a great fit is soon recruiting more people like themselves.

The second challenge is layering. Titles are cheap to give out when a company is young, and what a VP means at a Series A company differs from what it means at Series D. The balance is bringing in outside talent with the experience to raise the level, while giving strong internal people room to grow and reasons to stay. She is firm that hiring exclusively externally and layering everyone is rarely right, and equally firm that there is no formula. It comes down to knowing the team, their strengths, and their actual career ambitions.

Defining the role before you take it

Stewart's account of what a COO is makes the variation concrete. With a technical founder as CEO, the COO may own everything non technical. Elsewhere COOs run product and engineering. In parts of health tech, the COO runs HR, finance, legal and back office functions while a chief business officer owns go to market. At some companies CBO means pricing, or special projects, or strategic planning.

Which means the job is defined by the CEO, and the conversation should happen before you join. Her questions are specific. What do you like doing with your time. Where do you feel you shine. Why are you hiring for this role. What do you want to spend less time on, or want to go better than it is going now.

Then check the overlap. Too much and it may not be a fit, unless the CEO genuinely wants to step back. Too little and you have gaps to bridge deliberately. Her summary is that there should not be many surprises once you start, and it is better for everyone if there are not.

Bringing your playbook without transplanting it

Asked whether she recycles policies from previous companies, Stewart's answer is that this is what learning means. Seeing the same problem before and knowing what worked is precisely the value of an experienced hire, and it lets a company skip the experimentation phase. Rather than trying six approaches, she can narrow it to the two that tend to work and let the team choose.

Org design is her example. There is no perfect structure, only several defensible ones that then have to be adapted to the company's specific situation, including legacy realities like a function sitting in an odd place because a particular person is good at it. But some constraints hold. You generally do not have legal reporting to the head of engineering, and there are reasons for that.

The most interesting part of her background here is the one she says people skim past on her LinkedIn: Random House. Working at the largest English language publisher, an old school company, gave her a reference point that Facebook could not. Random House moved slowly, and the shift from physical to digital books took years. It also had high quality managers, real investment in developing people, careers measured in decades, and deep industry expertise.

Her conclusion is to cherry pick deliberately across both. Move quickly like Facebook, and build the management depth of a company where people stay 20 or 30 years, rather than adopting either stereotype wholesale.

Launching the Audience Network in three weeks

The best story here is about how the Facebook Audience Network got built.

A tiger team put it together in about three weeks, and Stewart is candid that they did not get approval. The company was disorganized enough that they concluded they could build it and launch it quietly without getting caught, and that once it was working it would be hard to shut down. That is exactly what happened.

She describes it as cowboy and Wild West, with no bureaucracy in a position to stop it, and takes a real lesson from it: you can often do something with fewer people and less process than you think you need, then iterate.

She immediately qualifies it. That approach fits zero to one. A product already serving many customers, at the point of needing to scale, deserves more planning, so you build it once in a way that lasts rather than rebuilding every few months.

Keeping innovation alive without making everyone an entrepreneur

Michael asks how you preserve an entrepreneurial culture while adding the process a growing company needs. Stewart's answer starts by rejecting the premise that everyone should be entrepreneurial.

Risk operations and legal do not need that mindset. They need to work within it and avoid slowing things down unnecessarily. As she puts it, you probably do not want a lawyer hacking together contracts and shipping them at 80% good enough.

At scale, and Automattic was nearly 1,300 people when she left, coordination becomes the requirement. Teams need to know what other teams are doing so that a product marketing plan lands alongside the product.

Her mechanism for protecting innovation is separation. Small tiger teams with little bureaucracy and freedom to iterate, resourced with what they actually need, their own engineers, product manager and business people, though probably not their own finance person. Enough oversight, not enough process to slow them.

Her closing warning is about cargo culting. It used to be copy Google, then copy Facebook, now copy Stripe or Twilio. Approaches go in and out of fashion and all of them have costs. Move fast and break things works for moving fast, and Cambridge Analytica and the misinformation problems are byproducts of shipping before everything is understood. A company guarding people's money, with real compliance obligations, is held to a different standard.

The question she thinks defines a good operator: what is the minimum process I can get away with that still prevents this company from wasting enormous time on a costly mistake.

The 5 things I took away from this conversation

1. Run a listening tour, then publish your own scorecard. Fifty interviews is the easy half. Presenting five themes to the board with your own three, six and twelve month goals, and then repeating it to the whole company with an invitation to correct you, is the part almost nobody does. It converts a new executive from an unknown quantity into someone with visible commitments.

2. Interview the customers who chose your competitor. Stewart deliberately talked to prospects and to competitors' customers during onboarding. Internal listening tours tell you what your team believes. That group tells you what the market believes, which is frequently different and always more useful.

3. Regrettable versus non-regrettable beats the attrition rate. A company with 20% attrition losing the wrong people is healthier than one with 8% keeping the wrong ones. Stewart's framing that keeping poor performers is as damaging as losing strong ones is the sentence I would put in front of any manager avoiding a hard conversation.

4. Settle the CEO question before you accept the job. If the COO role is whatever the CEO does not want to do, then you cannot evaluate it without knowing what that is. Her questions are the right ones, and they are all askable in an interview, before the surprises become your problem.

5. Don't make everyone entrepreneurial, ring-fence the people who should be. This resolves a tension most scaling companies handle badly. You do not need a company of improvisers. You need small, well resourced teams with permission to move, sitting inside an organization that is otherwise coordinated on purpose.

FAQ

What are the chief operating officer duties in a technology company? They vary more than almost any other C-level role. Stewart's framing is that the COO takes on whatever the CEO does not want to do, which with a technical founder often means everything non technical. Elsewhere COOs own product and engineering, or only back office functions like HR, finance and legal while a chief business officer runs go to market.

What should a COO do in their first 90 days? Stewart's version was roughly 50 interviews across employees, customers, prospects and competitors' customers in the first three weeks, distilled into five themes. Those became a plan presented first to the board and then to the whole company, with her own measurable commitments at three, six and twelve months and an open invitation to correct her.

How do you measure whether attrition is a problem? Set a baseline by asking investors and other operators in your industry, since tech runs higher than most. Stewart treats 40 to 50% as high and 4 to 5% as low. More importantly, split departures into regrettable and non regrettable, which tells you whether you are losing the people you wanted to keep or removing the ones you should not have hired.

How do you keep hiring quality high while doubling headcount? Stewart's caution comes from watching Facebook relax its bar for a single year and live with the consequences, including an internal reputation attached to that hiring cohort. The compounding risk is that new hires immediately begin hiring, so a weak hire recruits more like themselves. The controls are honest attrition measurement and disciplined process.

How do you manage change without losing the team's trust? Start by naming the problem accurately rather than sugarcoating it, since employees already live with it daily. Explain why the change matters in terms people can believe, apologize for the current state, and ask for help fixing it. Expect a vocal minority of detractors and a larger group that has been waiting for someone to say it out loud.

Also mentioned

• Shippo, the e-commerce shipping platform, and Automattic, the company behind WordPress
• Facebook Audience Network, built by a tiger team in about three weeks without formal approval
• McKinsey, where Stewart learned the generalist operations toolkit, and Penguin Random House, where she learned management depth
• The Walt Disney line used to reframe revenue at Automattic: we make money to make more movies
• Stripe and Twilio, the companies Stewart says are currently copied uncritically, as Google and Facebook were before
• Coinbase and Robinhood, cited as businesses where moving fast carries a different level of risk
• Catherine Stewart on LinkedIn, and her site at CatherineTaylorStewart.com, built on WordPress

Listen to the full episode

Catherine Stewart on Between Two COO's

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