CEO vs COO: the conversation most founders postpone. Micah Rosenbloom of Founder Collective

The CEO vs COO question is usually settled badly, or not at all. Micah Rosenbloom has been on both sides of it: once in a company that got it wrong, and once in a company that got it right and sold to 3M.

Rosenbloom is a managing partner at Founder Collective, the seed firm behind Uber, ThredUp, HotelTonight, Venmo and PillPack. Before that he co-founded Brontes Technologies, a dental technology company built on real time 3D imaging, where he was COO through an acquisition by 3M. He is the first working venture capitalist to appear on the show.

Getting it wrong the first time

Rosenbloom's first company was a dot com, founded by four people around his dining table in Los Angeles, where they divided the jobs. He took the president title, which made him second to the CEO.

His retrospective is unsparing. They were not good at delineating titles, there was too much ego, and too much overlap in who did what. The company raised a lot of money and failed. He is careful not to blame the structure for the outcome, noting the era did plenty of the work, but he made mistakes and so did everyone else.

The lesson stuck. Years later he watched the same failure recur across a portfolio, and he can put a rate on it. Two or three times a year, among the companies his firm works with, a founder leaves or a founding team blows up, and it traces back to a foundation that was never properly laid.

He identifies two versions of the mistake. In the first, founders simply postpone the hard conversation, reasoning that if the company is worth something they will sort it out later, and if it is not then it never mattered. In the second, which is more insidious, they divide the titles but skip the conversation underneath. The labels exist and the belief does not. Nobody actually accepts that when it comes to it, the CEO decides.

Getting it right the second time

Brontes started at MIT, where Rosenbloom and his partner Eric met a professor and a postdoc working on real time 3D imaging and looking for a commercial application. The professor's assessment of them, which Rosenbloom repeats with some amusement, was that he had pictured MBAs as aggressive, arrogant but smart business people, and that they fit the bill.

They spent about a year winnowing market opportunities before landing on dentistry, helped by connections to Invisalign, which wanted a way to digitize a dental impression directly from the patient rather than through the tray of material most people have had in their mouth at some point. The first slide of the pitch deck was a picture of Egyptians making gold crowns, to make the point that the process was that old. Uncomfortable, slow and inaccurate, and ready to be replaced by a scan.

The founder conversation happened in a conference room at Harvard Business School, and it happened on purpose. Both of them could not be CEO. They put it on a whiteboard, because they put everything on a whiteboard, and they were transparent about who was better at what. Eric was the stronger strategist, better at positioning the company for fundraising, and had a read on the market and a maturity that made him the right CEO.

They considered co-CEOs and rejected it. Rosenbloom, who has since invested in companies that tried it, is against the structure.

What he says next is the honest part. Eric told him he was good with people, good at operating and managing, and should be COO. Rosenbloom describes taking a slice of humble pie, and he took the weekend to think about it because it was a hit to the ego. He came back seeing it as the right division of responsibility.

The rule they set is the one most teams never state out loud. They would consult each other on big decisions, and if it came to a genuine disagreement, the CEO's call was final.

What the job actually was

With the decision settled, the split was clean. Eric took the board, fundraising, strategy and much of product. Rosenbloom took manufacturing, engineering, which reported to him through the CTO, hiring, and keeping the trains running.

His practical description of the role is that he was the one who always knew how many units would ship and when, which features would be ready, and what was going wrong. When a customer problem escalated, he took the call. There were occasions that required the CEO's title, but the default path for a frustrated customer ended at him.

He is direct that the title can be a catch all, and that what it meant at Brontes was specific to Brontes.

He also thinks of himself as having been the company's cultural focal point, which he treats as part of the job rather than a side effect. He ran the all hands, with Eric taking the big picture and functional heads reporting in. The culture was open enough that someone in engineering could stand up and say they were worried about sales. There was a golden bowling ball, spray painted after a team outing, that moved from desk to desk when someone did something exceptional.

Recruiting as the main event

In the early years, Rosenbloom spent more than half his time hiring, and he names it as his single biggest contribution to the company.

The work was networking into communities of engineers and salespeople to find out who was good, building rapport with recruiters, and selling the dream repeatedly. He describes it as a sales function, driven by the belief that good people make it possible to get more good people.

Two hires illustrate the timescale. Their head of marketing, a career dental lab industry veteran who called himself the OG lab guy, took two years to recruit away from an incumbent. So did the operations leader Rosenbloom eventually brought in, who became COO after Eric left.

The compounding benefit is one that operators underrate. When you personally recruited the head of marketing or operations, you have a bond with them and they are grateful you brought them in. That rapport is available later, when you need it.

He is clear the role does not stay fixed. The first couple of years were recruiting. The next were supply chain and customer problems. His image for it is a hitter who slots in wherever the team needs them.

Selling the company without stopping the company

Rosenbloom went through the 3M transaction as a co-founder and COO, and his account is one of the more useful things in this episode for anyone facing the same thing.

Complexity went up sharply. Regulatory, engineering, analysts with questions, a data room, hundreds of documents to assemble, and buyer scrutiny that produced constant requests for data nobody had needed before.

The division of labor is what made it survivable, and he frames it as a separation of church and state. Someone has to run interference and handle the day to day of the deal. Someone else has to stay focused on selling the big opportunity, because a deal process offers endless reasons to get nervous, and it is easy for a transaction to drown in caveats. Keeping the CEO on the enthusiasm and the strategic case, while the COO feeds the operating team and moves the mechanics along, is exactly what good bankers try to create. Rosenbloom's view is that it works better when it comes from inside the company.

There was also the ordinary problem of secrecy. Suits arrived from 3M and nobody could be told, so the process was disguised, helped by a main conference room set apart from the rest of the office. He assumes people suspected anyway.

Underneath it, someone had to keep the business running, because the deal might take months and might not happen. In the early stages that was his job while Eric worked out whether there was a deal worth doing.

Losing your scrappiness on purpose

The most self critical part of the conversation is about what happened after the acquisition.

With 3M's resources behind them, they built a training center. Five dental chairs, flat screens everywhere, a $700,000 3D printer on display. 3M wanted the showpiece and they had the space in Boston. Rosenbloom's retrospective is that they did not need it, and that the time spent ordering equipment and assembling it was time a startup would never have spent. They also launched an orthodontic product quickly, largely because they could.

His summary of the trap is that a COO is a resource gatherer, and then one day you have all the resources, and the frugality that got you there quietly leaves.

Michael's contribution is a line from Automattic: we ain't Google, so let's be frugal.

Rosenbloom extends the point to today's heavily funded startups and thinks they will lose it too. His head of finance used to say he was always the no guy, and Rosenbloom's response was that he needed him to be the no guy, because it is genuinely hard. His example is a conference. One person asks and it is important, so yes. Next year three people ask. The year after, twenty. He calls it the tyranny of incrementalism, and says it happens all the time.

What an investor looks for in a founding team

Rosenbloom watches for the same yin and yang he had with Eric, and thinks he is somewhat unusual in weighting it so heavily.

If two founders are in the pitch and one does 99% of the talking, he starts to worry, because the complementarity is not there. When the division of responsibility is clear, he feels like he is getting two skill sets for one investment. He invests in solo founders and says it is simply harder, with a caution that sometimes a solo founder has not found their partner and sometimes there is too much ego, and too much attention on keeping equity rather than getting enough people on the bus to make that equity worth something.

He notes how much the perception of the COO role has changed. Ten years ago people asked why a ten person company would have one. Nobody asks now. What he evaluates instead is whether the founder is self aware enough to hire against their own weaknesses, which sometimes means a COO and sometimes means a finance or product partner.

Speed is not conviction

Asked what a COO joining an early stage company should look for in investors, Rosenbloom names operational empathy first, and then makes an argument worth sitting with.

He thinks the market is currently confusing speed with conviction. An investor who says they love it and will write the check today reads as belief. An investor who asks to see product data, wants to meet again next week, and suggests coffee reads as hesitation. Rosenbloom argues the opposite is more often true. The one digging in is more likely to have real conviction when the check arrives, and more likely to understand and help the business afterward.

He understands the appeal of the fast yes, and acknowledges founders do not always have a choice. The distinction he wants people to hold is between an investor making a bet and an investor becoming a partner. Later, bets are fine. Early on, investors can help or harm a company, and the ones who help are frequently the harder sell.

The 5 things I took away from this conversation

1. Decide who decides, and say it out loud. Not the titles, the tiebreak. Rosenbloom's two companies differ on exactly this point and so do their outcomes. His estimate that a founder blow up traces back to this two or three times a year across a single portfolio should be enough to make anyone have the conversation this week.

2. Taking the second seat well is a skill, and the ego cost is real. He took a weekend to accept being COO and says so plainly. That honesty is more useful than the tidy version, because the people who cannot get past that weekend are the ones who end up in a co-CEO arrangement that fails later.

3. Recruiting is not a phase, it is the leverage. More than half his time in the early years, two year campaigns for individual hires, and a lasting bond with everyone he brought in. Operators who treat hiring as an interruption to the real work have the ratio backwards.

4. In a sale, split the roles the way bankers would. One person carries the story and the enthusiasm. Another carries the data room and keeps the company running. Doing both yourself is how deals stall and businesses drift at the same time.

5. Resources erode discipline, and someone has to be the no. The $700,000 printer on display and the conference that grows from one attendee to twenty are the same failure at different scales. Frugality is not a stage a company passes through, it is a practice that has to be actively defended once the money arrives.

FAQ

What is the difference between a CEO and a COO in a founding team? In Rosenbloom's case the CEO owned strategy, fundraising, the board and much of product, while the COO owned manufacturing, engineering, hiring and delivery. The more important distinction was decision rights. They consulted each other on big calls, and the CEO had the final say when they disagreed, which they agreed to explicitly rather than leaving implied.

Why do co-founder relationships break down? Rosenbloom's diagnosis is that the hard conversation gets postponed or faked. Some teams never divide authority at all. Others assign titles without anyone genuinely accepting what those titles mean, so the disagreement that eventually arrives has no resolution mechanism. He sees the consequences two or three times a year across his firm's portfolio.

Should an early stage startup have a COO? Rosenbloom says the question has changed. A decade ago people asked why a small company would have one. Now he treats it as an asset at almost any size. What he actually evaluates is whether the founder is self aware about their own gaps and has hired against them, whether that person is a COO or a finance or product counterpart.

What does a COO do during an acquisition? Two things at once. Run the mechanics of the transaction, which means the data room, hundreds of documents, and coordinating regulatory, engineering and finance for buyer scrutiny. And keep the operating team focused on running the business, since deals take months and may not close. Meanwhile the CEO stays on the strategic case rather than getting pulled into the minutiae.

What should founders look for in an early stage investor? Operational empathy, and evidence of real diligence. Rosenbloom argues that founders misread speed as conviction, and that the investor asking harder questions and taking longer is often the one with genuine belief and the ability to help later. The distinction is between someone making a bet and someone becoming a partner.

Also mentioned

• Founder Collective and its portfolio, including Uber, ThredUp, HotelTonight, Venmo and PillPack
• Brontes Technologies, the MIT spinout in real time 3D dental imaging, acquired by 3M in 2006
• Invisalign, whose interest in digitizing dental impressions pointed Brontes at its market
• The founder partner model, whose participants included Chris Dixon of Andreessen Horowitz, Bill Trenchard of First Round, Scott Belsky of Adobe, and James Tamplin of Firebase
• Fred Wilson's AVC, on whether one job per person is a fading norm
• Airtable, part of Founder Collective's internal operating stack
• Axios Pro Rata by Dan Primack, the deal newsletter referenced in passing
• Micah Rosenbloom on Twitter at @michajay1

Listen to the full episode

Micah Rosenbloom on Between Two COO's

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