Spenser Skates - Challenging Big Ideas - [Founder’s Field Guide, EP. 42]
What this covers
Patrick O'Shaughnessy interviews Spenser Skates, co-founder and CEO of Amplitude. The talk begins with his programming background, including two Battlecode wins at MIT, and moves to building a product analytics company. Skates describes product development shifting from executives declaring which feature will work to quantifying each feature's impact, a shift he frames as Mad Men vs Moneyball. He says no major SaaS vendor yet sells to the chief product officer. The Calm case study is the concrete example: the strongest predictor of long-term engagement was whether a user set a daily reminder, a feature an engineer had added on a whim.
The middle of the conversation turns to equity. Skates traces grants to Hewlett-Packard in the 1950s, then criticizes the 90-day exercise window, which he says favors the cash-rich over those who stay. He also criticizes one-year vesting at companies like Coinbase, Stripe, and Lyft. He sketches a governance idea in which shareholders gain votes the longer they hold, and he calls dual-class structures unprincipled. He also explains why he took Benchmark's lower Series A offer, and how founders can judge which investors add more than cash.
The most contentious stretch covers the IPO. Skates cites Jay Ritter's data on roughly 25% average underpricing and calls a traditional IPO a breach of a CEO's duty. He favors direct listings, though he concedes they cannot raise primary capital, and Patrick pushes back that the comparison may be apples-to-oranges. He also warns CEOs and investors away from SPACs. Shorter detours cover Norman Borlaug and Google as an extension of the mind.
Skates argues that the traditional IPO process systematically underprices companies because its advisors and ecosystem serve interests other than the existing shareholders, and that equity and vesting norms should be designed to reward long-term ownership.
- Bankers, lawyers, and public market investors have incentives that diverge from the company's existing shareholders, and an average of about 25% IPO underpricing transfers roughly $200 billion since the 1980s.
- Equity should be made generous, transparent, and long-dated, as shown by Amplitude's 10-year exercise windows and its opposition to one-year vesting.
- Data-driven product analytics, applied to user-level behavior, reveals engagement drivers that founder intuition misses, as in the Calm reminder example.
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Hewlett-Packard was the first company to make equity grants a standard form of compensation in the 1950s, doing so partly to incentivize company building and partly to give workers part ownership of the means of production, a blend of capitalism and socialism.
“the very first company to make it a standard thing to give you equity grants was, at least in the value, was Hewlett Packard back in 1950s.”
The 90-day exercise deadline exists because employee options are incentive stock options that cannot be held after employment ends, and the workaround recommended by lawyer Craig Spencer is to convert them to non-qualified stock options, which can be held indefinitely.
“you can just convert those ISOs to Ensos for non-qualified stock options, and you can just hold them indefinitely.”
Norman Borlaug's high-yield, disease-resistant wheat strains, developed in the 1950s and 1960s, quadrupled Mexico's wheat production within a few years and helped make it food self-sufficient, and the same approach was then replicated in India, Pakistan, and Africa, helping to disprove the Malthusian famine predictions of Limits to Growth.
“the wheat production in Mexico quadruple in a few years, and then they became self-sufficient in terms of food.”
Research shows that people recall information they can Google less well, but this reflects outsourcing memory to the web, which makes society smarter; Spenser views tools like Google as extensions of the mind and expects all software to become similar extensions.
“we've outsourced a huge part of our brains to the web and we can look up all sorts of things really, really easily”
A significant downside of direct listings is that they cannot raise primary capital at listing, so companies must raise private capital before listing or after, and Spenser concedes this is a genuine weakness in the direct listing process.
“one of the biggest downsides of the direct listing process is you can't do a primary capital raise.”
A CEO's job as a fiduciary is to get the best possible price for existing shareholders, so a traditional IPO that leaves large first-day price pops is a breach of that duty, and any CEO who chooses a traditional IPO over a direct listing is breaching it.
“As a fiduciary to amplitude, my job is to get the best price for equity on behalf of my existing shareholders.”
The traditional IPO process is structured so that investment bankers, public market funds, press, and lawyers, whose incentives differ from the company's, push the company to lower its stock price, which is a principal-agent problem that works against the company and its existing shareholders.
“The whole process is set up to give you a bad outcome. You hire consultants, your investment bankers, to advise you”
SPACs circumvent the SEC's disclosure protections because a reverse merger into an already-public shell lets companies make unconstrained forward-looking forecasts, such as projecting $100 million in revenue with none today, so Spenser advises CEOs and public investors to avoid them until the rules are sorted out.
“it's basically an end run around all the structures that the SEC has put in place.”
The common arguments for underpricing IPOs, such as controlling price, keeping the stock rising to please employees, and avoiding pricing out long-term holders, are all flawed, because price discovery needs an auction and markets do not reward monotonic price increases.
“You don't know your price either. That's why you have an auction process to discover your price.”
Jay Ritter's research on IPOs back to the 1980s found about $1 trillion raised in total with average underpricing of about 25%, meaning roughly $200 billion was effectively given from companies to institutional investors, which Spenser calls the largest arbitrage opportunity in all of finance.
“On average, there's about a 25% under-pricing. So what that means is of that $1 trillion, about $200 billion was inefficiently pricing given away”
Dual-class stock structures, which give pre-IPO shareholders 10 to 20 times the votes of later buyers, are bad corporate governance because there is no principled reason pre-IPO founders' shares deserve more influence than public or institutional investors' shares; the argument for founder long-term views is only grandfathering.
“I think it's actually really bad from a corporate governance perspective.”
A large first-day stock price jump after an IPO, such as a 100% pop, indicates that the offering was mispriced because one side lacked correct information about the company's value, which is a sign of market inefficiency that the company's existing shareholders bear.
“whenever you have big price movements, that's a sign of a massive inefficiency in the market”
Investment banks in traditional IPOs often threaten to delay or cancel the offering if the CEO pushes the price too high, effectively holding the company hostage, and this is not explicit collusion but a consequence of how the process is structured.
“a lot of these bankers will threaten to hold up the IPO and say, 'Hey, if you're too high on the price, we're not comfortable going out'”
Traditional IPOs deny retail investors allocation, which transfers tens or hundreds of millions of dollars from companies to institutional public market investors, a transfer Spenser says is hard to justify as anything other than cronyism.
“in a traditional IPO process, you don't give allocation to retail investors”
Patrick argues the IPO versus direct listing comparison may be apples-to-oranges because IPOs raise primary capital from the company while direct listings only let existing shareholders sell, so the comparison may not explain underpricing.
“in the IPO process you're raising money... a direct listing, you're not.”
The standard 90-day window to exercise options after leaving a company forces departing employees, who may need tens or hundreds of thousands of dollars to exercise, to either pay a large sum into an illiquid private company or forfeit their equity, so it mainly rewards those who are cash-rich rather than those who stay.
“You're only really incentivizing the people that don't have money, and it's like it's almost a form of wage slavery or equity slavery.”
A major part of Calm's value to users was building a healthy meditation habit, not guided meditation alone, and the daily reminder was the most undervalued feature even though an engineer added it on a whim.
“A huge part of the value was helping you develop a healthy habit around meditation. They hadn't even realize that without it”
Spenser chose Benchmark's lower-valuation Series A offer over more generous offers because Benchmark's portfolio, which included Twitter, Yelp, Zendesk, New Relic, and Snapchat, suggested that roughly half its companies become massively successful, versus about one in 30 for a typical VC fund.
“probably about half their businesses go on to be massively successful, like incredibly massively successful. And that's a way better hit rate than any other VC fund.”
Major SaaS companies each sell to a specific buyer, such as Salesforce for sales, Adobe for marketing, ServiceNow for IT, and Workday for HR, but no one has built software for the chief product officer, which Amplitude is positioned to fill.
“No one's built something for the chief product officer.”
Short one-year vesting schedules at companies like Coinbase, Stripe, and Lyft turn equity into effectively cash compensation, cutting total ownership upside from three years of grants, and make employees mercenary rather than long-term owners.
“When you get to the one year point, it's like they're exchanging equity as basically cash. It's a little bit mercenary like”
Most venture capitalists add little beyond cash, but a few exceptional investors can change a company's trajectory, and founders can identify them by examining the full portfolio, including failures, and by doing reference checks on what the investor has actually helped companies achieve.
“So the lesson is, most VCs are undifferentiated. There are a few exceptions and you can tell by looking at their portfolio.”
Product development is moving from intuition-driven 'Mad Men' pitching, in which executives declare which feature will work, to a 'Moneyball' approach in which the impact of specific features is quantified; leading companies such as Facebook, Zynga, and Netflix were already taking the quantitative route rather than relying on a Steve Jobs-style brilliant idea.
“the world of product was going through this transition from a Mad Men phase where you literally had people in suits with cool ideas”
Corporate governance should reward long-term ownership by giving shareholders more votes in electing the board the longer they have held their shares, an idea that Delaware corporate law and the SEC have not yet adopted.
“I would love there to be a way for companies to basically the longer you've held the shares, the more votes that you get in electing the board.”
When Calm moved the daily reminder prompt to the first sign-in screen, the share of users setting a reminder rose from about 3% to more than 50% overnight, and user retention tripled, showing that a feature buried in settings was the most undervalued part of the product.
“They went from 3% to more than 50% of their user base setting a reminder overnight. The retention of their user base tripled”
For the meditation app Calm, the strongest predictor of long-term engagement was whether a user set a daily meditation reminder, a result the team did not expect and that contradicted their hypotheses about user types.
“the number one predictor was whether someone decided to set a daily reminder to meditate.”
Amplitude was the first Valley company to grant all employees 10-year unconditional exercise windows regardless of tenure or situation, a policy that required persuading the board and that other companies have since begun to copy.
“We were the first company in the valley to give 10-year unconditional exercise windows to all employees”
While building Sona Light, the two-person team built its own analytics system, which consumed roughly one-third of its development time and distracted them from voice recognition, the core business; they should have outsourced analytics to a specialist.
“We probably spent one-third of our development time just trying to build analytics.”
Product usage is not a linear sequence of clicks but more like a random walk, where every user's journey is unique, so analyzing it requires tools that make it easy to query a single user's history quickly.
“Products are not, 'Oh, click, next, next, next, done.' They're not linear things, it's more like a random walk.”
A large share of a meditation app's value comes from helping users build a healthy habit rather than from the guided sessions themselves, and this habit value was the most undervalued feature until data revealed it, although the lesson applies only to some categories such as fitness and lifestyle.
“a huge part of the value was helping you develop a healthy habit around meditation.”
Scaling functions such as finance, sales, and engineering has been done thousands of times, so founders should learn from experienced outside executives and stand on their shoulders instead of trying to reinvent each function, which is the most common founder instinct and a major mistake.
“you could just take their knowledge and stand on their shoulders to get further ahead more quickly.”
Founders should choose investors based on the quality of the portfolio they would be joining rather than on the best headline terms, since a strong portfolio of companies is worth more than a higher valuation.
“I ended up signing with them and it was even better than I could have expected.”
Eric Vishria of Benchmark was so involved in Amplitude that he interviewed hundreds of candidates, referred about a quarter of the early team, and was the most helpful investor in every fundraise, and Spenser says the company would be about a tenth of its size without him.
“We'd be a 10th of the size if it weren't for him. It's been like having another co-founder in the business”
Benchmark partner Eric Vishria's Series A investment in Amplitude, despite offering the lowest valuation of five offers, changed the company's trajectory, and Skates says Amplitude would be a tenth of its size without him.
“We'd be a 10th of the size if it weren't for him. It's been like having another co-founder in the business”
Amplitude built its own database with all of a single user's data stored in the same partition, so that a user's full journey can be scanned quickly to answer any question, although Spenser says he does not recommend other companies take this costly route.
“we ended up deciding to write, which I don't recommend anyone do but we did it, a custom database.”
For a SaaS company going public, the price investors will pay is about 90% of what matters, and it is largely determined by factors outside the CEO's control such as interest rates and tech-stock sentiment; Amplitude's $4 billion round cleared at a price that was too low for $5 billion and too high for $3 billion to draw enough investors.
“If we hadn't been at three billion, there would have been 20 investors that were in. If it had been at five billion, there would have been zero.”
The first few years of a startup are nothing like the expected trajectory, because founders must work through five or six failed ideas, such as the Sona Light voice recognition app that went through Y Combinator and failed, before finding the product that leads to a real business.
“We worked on five or six different ideas that didn't turn out. We did one app called Sona Light”
Even the highest cash salaries, around $10 million a year, are negligible compared with the wealth that early equity can create, which has produced employees worth tens of millions, hundreds of millions, or billions of dollars.
“that's nothing, that's absolutely nothing. With equity, you can get people worth tens of millions or hundreds of millions or billions in some cases.”
Reading about Norman Borlaug's agricultural work led Spenser to conclude that the highest-leverage way he could do good was to build software in Silicon Valley, since technology is what has most improved the lives of both the affluent and the poorest.
“I said, 'Okay, what do I know how to do?' I know how to build software, so what's the biggest and highest leverage place I can do that?”
Teams that refuse to show their strategy on the scrimmage server are making a mistake, because a win confirms why the strategy works and a loss reveals why it failed, and with around 300 competitors, one opponent's gain is negligible.
“There's like 300 different teams competing on this thing. You're going to learn two things.”
Most customers of product analytics only need basic questions such as retention, funnel drop-off, and feature-usage differences between groups, which can be answered with simple correlations; sophisticated AI matters only for a small part of the product.
“Most customers on Amplitude are just doing the basics.”
The argument that pricing too high will drive away long-term holders misunderstands markets, because a CEO should want the buyer who pays the most, and a high price today signals expectations of a high price in the future rather than predicting a fall.
“Who do I want as owner of Amplitude? Whoever wants to pay the most.”
In the MIT Battlecode competition, a good-enough navigation algorithm paired with an AI that adapts its strategy to each map beats teams that over-optimize a single navigation algorithm.
“Really, it was about creating one that was good enough, and then having your AI adapt to whatever the right strategy was on the map.”
Venture capitalists and existing Silicon Valley players, such as Andreessen Horowitz, who defend the traditional IPO have convinced themselves because of their own Stockholm syndrome, even though the traditional IPO harms founders, employees, existing investors, and retail investors.
“there's like a Stockholm syndrome in Silicon valley where the people that have convinced themselves the most, are the venture capitalists”
Software will move from static, identical experiences for every user to proactively adapting to each person, which would eliminate frustrations such as hour-long password resets, anticipate needs such as rides or groceries, and make people more leveraged.
“Over the longterm, you're going to see software that will adapt more proactively to you as a user.”
Amplitude was extra slow to hire outside executives, which held the company back, but the delay gave early team members a great deal of ownership and agency and made them deeply invested.
“we were extra slow. That definitely was a mistake, and in some ways held us back.”
Most employees did not understand how equity works and expected hidden gotchas, so explaining equity transparently and granting it aggressively became a recruiting differentiator that helped Amplitude break out of a crowded market.
“nobody had any idea how equity really worked and a lot were expecting to be taken away to some gotcha.”
Inexperienced entrepreneurs who guard their ideas are wrong on two counts: no one will steal a good idea, and sharing it both forces it to be tested and makes it better through feedback, so good ideas have to be pushed onto people.
“Firstly, no one's going to steal your idea.”
Requiring employees to return cash compensation 90 days after leaving would provoke regulatory and public outrage, so the same clawback applied to equity should be considered equally unacceptable.
“imagine you gave cash to employees and then 90 days after leaving, they had to give it back to you.”
The kindest thing anyone has done for Spenser was MIT classmate Matt Huang giving him free advice several times a week for about a year, when neither of them was successful, which Spenser believes kept him from failing longer and helped him reach his current position.
“he would G Chat me or call me a few times a week for about a year, even though we were nobody and nothing at the time.”
Amplitude aims to become the data infrastructure that powers personalized software, moving from 20 of the Fortune 100 toward a majority of them, and reaching the scale of Salesforce, which is above $20 billion, from its more than $100 million in ARR last year.
“We're in 20 of the Fortune 100 today and I would love for us to get from that to a majority of the Fortune 100.”
The feature that proved most important to Calm's user base was built by a single engineer on a whim because he personally wanted it, not because of a company-wide strategy among the roughly 50 people at the company.
“one engineer just put in the feature on a whim. They wanted it personally”
Winning the MIT Battlecode AI competition depends on understanding the game's mechanics and iterating on strategy, not only on being a strong programmer, which is why the competition closely resembles starting a business.
“What makes you good at that is not the same thing necessarily as being a good programmer.”
A person who wants to be a long-term owner should seek as much equity as possible as early as possible, and Spenser believes the argument favors even longer vesting than four years.
“you want to get as much equity as you can as early as you can. Which is why I love I think there's an argument to do them even longer”
Amplitude was founded in 2012 on the belief that every company would need quantitative product analytics, launched in 2014, and grew to a $4 billion valuation round led with Sequoia.
“We looked at that and we said, 'Hey, every single company out there is going to need this.' That was 2012. We launched the company in 2014”