Peter Reinhardt - Learning How to Sell – [Founder’s Field Guide, EP. 34]
What this covers
Patrick interviews Peter Reinhardt, co-founder and CEO of Segment, a customer data platform acquired by Twilio. The conversation starts with the product itself: a collection API reduced to two methods, identify and track, which let one pipeline replace many bespoke integrations. It then follows the company from classroom software through a deliberate six-month stretch after its open-source launch, when the team answered adjacent customer requests and killed many of its own ideas. Peter explains why the open-source library never competed with the hosted product. He also calls the Airbnb story of stubborn vision a poor model for founders.
The sales material is the most concrete. An advisor, Mitch, told Peter to ask an early prospect for $120,000, about 1000 times list price. The customer countered at $18,000, which showed Segment had priced on cost rather than value. Peter then works through the arithmetic of quota, where $1.5 to $2 million per rep at roughly 80% gross margin pushes deals into six figures. He argues Rev Ops is underappreciated and regrets staying product-focused between $10 million and $100 million in ARR. He also describes George Hu's method of drilling a plan down layer by layer until it reaches ground truth.
The last stretch turns more contentious. Peter reads Tesla as a battery company whose premium cars funded production scale. He claims the cohort of funded startups that buy SaaS flatlined around 2016 to 2017. He forecasts that privacy changes will concentrate advertising revenue in companies holding first-party data and hurt intermediary ad networks.
Reinhardt argues that durable B2B software companies are built by serving problems customers already have, not by imposing a founder's vision, and that sales economics, value-based selling, and revenue operations determine how far they can scale.
- Segment's early success came from customer pull after its own ClassMetric and analytics ideas failed
- Enterprise sales quotas, margins, and buying processes set the price and go-to-market model, not the vendor's preference
- Rev ops and bottleneck analysis, not product work alone, drive scale in later stages
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.
A leader's job is to inspire other people to do the work and explain why it matters, which is a fundamentally different skill from doing the work oneself, and academically trained founders often struggle with this shift.
“Your work is to inspire other people to do the work and then explain why it's important, which is a fundamentally different skill than doing the work yourself.”
Early Segment priced on cost basis rather than the value it created for customers, and Peter argues B2B sales should be built around value, because buyers care about the value they get and not the vendor's cost.
“the value that we were attaching in our minds was on a cost basis rather than a value basis.”
There is always one bottleneck somewhere constraining a business's growth, so leaders should isolate it and verify that changes actually resolve it.
“there's always one bottleneck somewhere that's constraining growth.”
In B2B sales, pricing should be anchored on the value the product delivers to the customer, because customers do not care about the vendor's cost basis, and founders who price on cost leave large amounts of revenue on the table.
“in B2B sales, value is what you should sell on and value is what matters to the customer. They couldn't care less what cost basis is.”
Tighter targeting raises customer acquisition cost for merchants, which means lifetime value must rise to compensate, so businesses will need to invest in keeping customers happy; Peter believes this may be a net positive for consumers, though the externalities are complicated.
“if your cost to a buyer goes up, it probably means that your lifetime value needs to go up in equal measure”
Tesla is effectively a battery company whose premium Roadster, Model S, and Model 3 sequence was a way to fund battery production scale, because batteries are expensive and the company needed premium markets to reach economies of scale before moving to utility-scale storage.
“Tesla is really a battery company and for them, it's a question of how do they achieve scale in battery production?”
The SaaS founding wave of 2012 felt like a Cambrian explosion with new interesting companies weekly on Hacker News, but that has faded into derivative and in-between products, and the cohort of funded startups that served as early customers appears to have flat-lined or declined around 2016 to 2017.
“the number of properly funded startups seems to have flat-lined or slightly started declining around 2016, 2017.”
Several companies' internal reports suggest the number of properly funded startups flatlined or began declining around 2016-2017, which removes the peer cohort that SaaS companies sell into and could impair new startups' ability to grow.
“the number of properly funded startups seems to have flat-lined or slightly started declining around 2016, 2017.”
Privacy changes will centralize advertising revenue in companies with their own consumer networks and first-party data, such as Google, Facebook, LinkedIn, Twitter, Pinterest, and Snapchat, while intermediary ad networks that connect third-party datasets are hurt.
“it will cause a centralization of ad revenue on the Googles and the Facebooks and LinkedIn and Twitter and Pinterest and Snapchats of the world.”
In B2B, a company is under contract to solve a business problem, so customer-driven building is largely cut-and-dry, whereas B2C may leave more room for a Steve Jobs-style founder to tell customers what they should want.
“In B2B we're under contract to solve a business problem. I think in B2B, it's pretty cut and dry. B2C might be different.”
Twilio COO George Hu's method of hierarchical unpacking drills any plan down through successive layers, from revenue target to bookings to rep count to hiring timing and recruiting capacity, until the team can verify whether the plan actually reaches ground truth.
“This unpacking of one intuition for where the bottleneck is likely to be. Two being very crisp about whether we really have solved that bottleneck or not.”
The Airbnb story of holding onto a belief that people would sleep on couches for years until it took off is a poor model for founders, since for every 10,000 successful companies like Segment there may be one Airbnb, and founders who bet on their vision with one shot on goal will usually fail.
“For every 10,000 Segment's out there, there will be an Airbnb, but do not bank on that.”
Effective enterprise sales involves getting the customer to articulate their own pain and value by repeatedly asking why, which clarifies the problem for both sides and speeds the sale, though the awkwardness of this approach is why most salespeople avoid it.
“You just get the customer to articulate it, clarifying for them, it's clarifying for you. The whole thing moves faster but it's very awkward.”
Good revenue operations does not squeeze 10 percent at a time from a sales team; by getting ratios right throughout the year so inbound interest flows through to closed renewals, it can produce a 2x to 5x improvement and show how to deploy newly raised capital.
“That's like a 2x or 5x potential improvement because as you start getting that stuff right it also shows you how you should deploy the capital”
Segment's data collection API needs only two methods, identify (who the customer is) and track (user X did Y at time Z), and this simplification let one pipeline replace many bespoke integrations.
“you actually only need basically two API methods. One is the identified call, which is who is the customer? And the other is the track call”
After the open-source launch, the team deliberately spent about six months solving adjacent customer requests, such as more sources and destinations, and killed many of its own product directions, before writing down a vision of becoming the company's customer record.
“we're going to spend the next six months just trying to solve their adjacent problems. We're not going to drop what we think this could be in the future.”
Sales advisor Mitch told Peter to ask an early enterprise prospect for a $120,000 contract, about 1000 times the public list price; the customer, Nat, countered at $18,000, which was still about 150 times the prior price, revealing that the company had been pricing on cost rather than value.
“He saved 85%. I was completely mind-blown because here we were charging 150x, literally 150x.”
Segment's open-source library never competed with its hosted product because deploying the library requires an engineer, while the hosted product removes the engineer from the loop so a marketer can turn on integrations, making the hosted version the actual solution.
“the open-source products by nature of being open-source doesn't solve the problem. And that's because you would have to deploy it yourself as a customer.”
Revenue operations is deeply underappreciated because it designs the territories, account tiers, vertical targeting, and headcount ratios of sales teams, such as sales engineers per account executive and SDRs per AE, which determine the economics and accountability of a large sales organization.
“Revenue operations, I think is deeply underappreciated.”
From $10 million to $100 million ARR, Segment kept focusing on product because that had gotten it there, which Peter now believes was the wrong choice; the company should have invested much more in revenue operations and go-to-market scale.
“I think that was probably actually the wrong choice.”
In SaaS, the analog to manufacturing cost is customer acquisition cost, so a new company should first target the customers with the lowest cost to acquire, which favors businesses with inbound demand and explains why venture capitalists look for passive, high-quota inbound sales models before outbound scale.
“you have to think about it as the cost to acquire the customer. You need to start with the customers that have the lowest cost to acquire.”
Enterprise sales economics require roughly $1.5 to $2 million in annual quota per salesperson, which forces deals into six figures once commissions, sales engineers, SDRs, and customer success are paid, and assuming roughly 80% gross margin; lower margins push the required quota higher.
“You have to be able to get to a quota per salesperson of $1.5 to $2 million a year very quickly.”
In SaaS, the scale economy or moat may be the sales team itself, since large sales organizations built through acquisition, such as Salesforce, gain customer lock-in from relationships in large accounts that are hard for rivals to replicate.
“I do wonder in SaaS if actually the scale economy or moat that gets created is actually the sales team.”
A company's go-to-market model should be set by how its customers want to buy, not by how the vendor wants to sell: developer-led customers want self-serve trials then sales, while enterprises want a full multi-vendor sales process.
“How do your customers want to buy? That is what matters here.”
Justin Kan's observation that first-time founders focus on product while second-time founders focus on go-to-market is partly right, because spending time with customers on go-to-market is the best signal of whether the product actually delivers real value.
“it's actually the best signal that whether your product is really the right thing.”
Peter's hardest personal change was giving up the habit of being right: as a top student he had to accept that he was wrong for the first two years, because in the real world you do not know the right answer until you ask someone else.
“you do not know the right answer until you go ask someone else. That's just not how school works, but that is how the real world works.”
Acquisition-built software suites face limits to their scale advantage because their cross-sell capacity is finite and their acquired products are not integrated, instead being separate products with a login layer and branding, which customers find unappealing and which invites disruption by integrated competitors.
“They're basically separate products that have a loss layer and branding slapped on top.”
Companies that try to clone Segment in-house underestimate integration quality work, since integrations like Google Analytics need years of refinement and each of hundreds of partner integrations keeps changing its API, so the maintenance burden is huge unless R&D is amortized over thousands of customers.
“when you have 20,000 customers using an integration and they're still finding things that can be improved about it”
The world does not care how a founder thinks it should work; it has its own problems, and a founder who humbly solves those problems gets rewarded while one who tells the market how it should work gets rejected.
“the world could not give a shit how you think it should operate. It really has problems that it wants solved.”
The most difficult part of the first chapter was keeping the founding team together through a year and a half of misery, which Peter says is why it survived; he believes many companies die because founders split up before reaching a breakthrough, and that people matter more than the idea.
“The idea mattered less than the people around the table.”
When Segment emailed its top 15 customers about raising the price to about $15 per month, one customer replied that the price was so low he might stop using the product because the company would not survive a year, which was the wake-up call that prompted price increases.
“this price is so low that I may have to stop using your product because there is no way you're going to be around in a year.”
Segment integrated a tool that shared third-party data but, within about two weeks, decided that practice was wrong and removed the integration, on the principle that end customers would not expect their data to be shared that way even if legal.
“Two weeks later, we were like that was a terrible idea. And we booted them off the platform.”
Tidy data, a long table of observations where each row is a thing done at a time, is the right starting structure because the current state of any user can be reassembled from event-level observations, while aggregates built first lose that flexibility.
“Basically, what it boils down to is a long table of observations of a thing at a time”
Kevin Kelly's idea of the technium frames technology as an independent system with needs; successful businesses serve those needs rather than trying to change the whole system.
“it's this independent thing that has needs, and like you just described, if you serve its needs, great things will happen”
A landing page for the open-source analytics.js library, posted to Hacker News as a test meant to kill co-founder Ian's idea, reached number one, earned thousands of GitHub stars and email signups, and drew demands for a hosted version within 24 hours.
“went straight to the top, got a few hundred upvotes on Hacker News, sat at number one for most of the day, got a few thousand stars on GitHub, got thousands of email signups.”
Twilio's communication APIs and Segment's customer data platform are complementary: Twilio can send messages but lacks data on who to message, when, and why, while Segment has that data but no channel to send through, so combining them creates new possibilities.
“Segment does like literally the inverse. Of all the data which should inform what message you send to who when and none of the communication channels to actually send that.”
Over the next 10 to 20 years, machine learning will blend role-specific applications such as support desks, CRMs, marketing tools, and advertising tools, making the pipeline of customer engagement very messy as recommendations determine which channel to use.
“I do think machine learning is going to make the blending amongst these different engagement layers, and tools very confusing actually for the next 10, 20 years.”
Data management platforms that aggregated and traded third-party data have largely collapsed over the last three years, and the Salesforce-acquired DMP faces a large class-action lawsuit, which vindicated Segment's first-party-only contrarian bet.
“the last three years basically saw the complete demise of DMPs, like their worthless assets.”
With roughly 300 downstream partners, at least one is almost always down or deliverability-impaired, so a data pipeline must buffer events and redeliver them when the destination recovers, or substantial data is lost.
“It's like we've got 300 partners. I guarantee at least one of them is down right now”
Privacy regulation such as CCPA and GDPR, plus Apple and Google moving to kill third-party cookies, are pushing companies toward first-party data, which Segment has always been built to support.
“Apple is making moves of killing off third-party cookies along with Google.”
Long-term defensibility for large companies comes from scale, either supply-side or demand-side, with network effects in consumer businesses, while early advantages like counter-positioning fade.
“long-term huge companies what's their defensibility? It's scale. It's either supply or demand-side scale.”
A small company's analytics stack multiplies pipelines as it grows: a website tracker feeds an email tool, then a second email tool, then a data warehouse, so each new use case requires re-plumbing the same data into another destination.
“Over time, the rat's nest of all these data pipelines multiples as companies become more complex”
Patrick's summary that sales should invert the usual pitch-deck approach, sniffing out pain points while letting the customer name them, was endorsed by Peter as matching how strong enterprise sales organizations already operate.
“Any great enterprise sales organization actually knows this but it was completely mind-bending and mind-altering for me”
Hardware companies should copy the premium-market-first pattern, starting with a high-price product to fund scale and then moving down the price curve, because hardware businesses are fundamentally economies-of-scale businesses.
“it's something that I think pretty much every hardware company should copy once you're a Roadster market because if you're in the hardware business, you're in an economies of scale market.”
Before Segment, most companies, and still most of the market, built each data integration in-house, with engineers assigned to every marketing pixel, Stripe pull, or Salesforce pull, creating a sprawl of independent bespoke pipes.
“Before us and frankly, still most of the market today, goes and builds this in-house.”
After ClassMetric failed, the founders called back investors from a $600,000 demo-day raise, asking what to do with the money, and most investors said they had invested in the team and told the founders to find something else, while two were paid back.
“we called back all the investors that we had raised money from a few weeks prior”
Peter initially opposed hiring a sales rep from Jersey with a slicked-back, intense look, but first sales hire Raph overrode him, and that rep became the most successful salesperson in Segment's history because customers loved him.
“I was wrong. He was our most successful sales rep for the entirety of Segment's history”
Explosive, uncontrolled interest is common in B2B founding stories, illustrated by Dropbox blowing up on Reddit, where every metric spiked after months of only about 10 visitors per day.
“Dropbox, another great example. They blew up on Reddit.”
Peter wrongly opposed hiring a sales rep who did not fit the engineering-company sales image, but that rep became Segment's most successful seller because he extracted customer value thoroughly and methodically.
“I was wrong. He was our most successful sales rep for the entirety of Segment's history”
A father's offhand remark that Peter would need to learn to finish something, after Peter started but never completed many projects around age 13, became a lasting lesson that finishing things matters.
“eventually you're going to have to learn to finish something”
Jeff Lawson's leadership is distinguished by deep care for culture, refusal to accept no as a final answer, and high energy that imparts momentum to a large team, as illustrated by how he eventually persuaded Peter to accept the Twilio acquisition after roughly 18 months of quarterly conversations.
“he's not going to take no, it might take a while, but he's not going to take no.”
Finishing what you start matters a lot, a lesson from a parent's tough-love remark when Reinhardt abandoned a new project every two weeks as a teenager, and such tough love is among the kindest things.
“eventually you're going to have to learn to finish something”
Segment grew from $2.5 million ARR to $10 million ARR in one year because it sold into a growing cohort of startups, while penetration within that cohort also rose, creating three compounding growth forces.
“In one year, we went from $2.5M in ARR to $10M and so on.”
Advisor Mitch's advice to raise prices 1000x was the most impactful advice Peter received, and Peter says without it Segment would not have a company, while acknowledging pricing still has problems to fix.
“Mitch's advice to raise prices 1000x was pretty impactful.”
Segment's first product, ClassMetric, a classroom button that let students signal confusion, failed because it distracted students; in one BU anthropology class about 60% of students were not using it at the start and about 80% by the end.
“I think at the beginning of the class 60% of the students were not using ClassMetric at any given time. Then by the end, it was about 80% were not using ClassMetric.”
The hardest operating challenge after the Twilio acquisition is driving accountability and operating at the right altitude in an organization of about 600 people with a few hundred open rep positions, which is different from managing a small team of friends.
“how do you hold accountable and operate at the right level of altitude in that kind of organization”
Peter says he came to buy into Twilio's vision only after Jeff Lawson spent about 18 months meeting roughly once a quarter, and he now spends much of his time on the combined Twilio and Segment roadmap.
“He and I chatted once a quarter for 18 months and at the end, I was like, 'All right, Jeff, I buy the vision.'”
Customer data flowing through Segment is used mainly by five categories of tools: basic web analytics, email marketing, push and SMS notifications, advertising, and data warehousing, with use cases like support falling off sharply after those.
“There are basically five categories of tools that people really depend on.”
During the 18-month early struggle, Peter went to the hospital multiple times for panic attacks and lost 10 pounds in two weeks, reflecting how poorly the company was doing by any health measure.
“I ended that year and a half with going to the hospital multiple times for panic attacks. I lost 10 pounds in two weeks.”
After being wrong about multiple ideas, Peter says he took about a month and a half to recover emotionally from the market response, illustrating how founders can be shaken by repeated validation failures before a winning idea emerges.
“It took me a month and a half to recover from that”
After ClassMetric failed, the team spent about a year building an analytics tool to compete with Mixpanel and Google Analytics based on their conviction about what was right, and that effort also failed about a year and a half in, leaving them with four to six months of runway.
“they then spent a year trying to build an analytics tool to compete with Mixpanel and Google Analytics”
Segment helps companies collect all first-party customer data from websites, apps, help desks, and payment systems into one holistic customer record, then routes that data to marketing, sales, and support tools.
“we help companies collect and manage all of their first-party customer data”
Beyond Twilio and Segment, Peter spends time on climate tech and philanthropy, which he finds exciting after a dark decade.
“I also spend a reasonable amount of time and climate tech and both Charm and on the philanthropic side.”
The founders, including students from MIT and RISD, applied to Y Combinator in 2011 with a classroom lecture tool, ClassMetric, which let students push a button to signal confusion, and raised about $600,000 at demo day.
“we'd raised $600,000 at a demo day on the back of professors trying it out in the fall.”