John Kramer was looking for a snack when his job changed. In his account of the episode, he was investigating the HoneyBook office refrigerator and turning up a Greek yogurt while the founders were deciding who should build the company’s sales team. They chose him. A snack is a modest ambition. Responsibility for customer acquisition is rather more filling.
He had joined as chief operating officer, and he admitted he knew little about sales. His initial instinct was to recruit someone who did. After talking to other operators, he concluded that the company first needed to learn how to sell its own product. An experienced manager would need a process to inherit.
Kramer hired two account executives, put his other responsibilities aside, and worked through the calls. There was a sleeping bag under his desk. About four months in, a morning dashboard finally let him anticipate the day’s results. He had something another person could run.
That episode offers a useful entrance to his career. The recurring question is who will do the necessary work, and what will make it possible to hand that work over.
What was on Oz’s plate
An earlier instruction had supplied the metaphor. During Kramer’s first week at HoneyBook, a board member told him to discover what was occupying CEO Oz Alon and take it away. The list included bank transfers, laptop purchases, furniture assembly, office electricity and temperamental Wi-Fi. Vision, apparently, still needed someone to keep the router working.
Kramer’s 2018 explanation of the COO role followed that advice through several years. A task could disappear because someone stopped doing it, because a process reduced it, or because a new hire took responsibility. The objective was to give people more time for the work they were particularly good at.
There is an appealing lack of ceremony in that description. It makes management legible through ordinary objects: a laptop, a bill, an overfull day. Seen through that lens, a company grows partly by getting better at deciding whose attention a problem deserves.
Salsa, which Kramer later co-founded, would apply a closely related idea to the software companies serving small businesses: take on an intricate job so somebody else can concentrate.
“When you remove unnecessary tasks from other people, it gives them a shot at doing their very best work.”
John Kramer · 2018
A job description, measured over dinner
Kramer also wanted to understand the job beyond his own desk. He started Performance COO, a San Francisco community of startup operations leaders. Over three years, the group tracked the responsibilities of more than 30 Bay Area COOs. Their average member oversaw 4.4 distinct functions.
His own early HoneyBook assignments had included an operating model, a board presentation, pricing and the first revenue collection. Later came teams, managers and the smaller responsibilities he called “orphan functions.” Some grew into departments. Others remained too intermittent to justify a dedicated employee.
He described interviewing roughly 100 people a year and identified recruiting as his favorite part of the work. His interest in people sat alongside an interest in diagrams: he mapped how his responsibilities changed as the company grew.
The dinner group and the tracking exercise belong together. When a title was confusing, he gathered people who shared it and compared the actual work. It was a practical response to ambiguity, with the considerable advantage of including dinner.
The things he thought were worth paying for
When HoneyBook reached a billion-dollar valuation in 2021, Kramer wrote a reflection on what had brought it there. He credited design, conviction, culture and recruiting. His opening principle was wonderfully brief: “Design wins.”
He argued that a company could not expect customers to supply its entire vision. He also defended cultural investments that could appear expensive or inconvenient: daily all-hands meetings, long retreats, frequent celebrations and decisions shaped by values even when a financial calculation pointed elsewhere.
His recruiting account was equally emphatic. He remembered continuing conversations with candidates after they had declined, trying to persuade people to join a company whose potential still needed explaining. These were the choices he chose to celebrate when a valuation gave everyone permission to look backward.
Taken together, the essays suggest how Kramer wants companies to be built: with deliberate systems, but also with a point of view. The spreadsheets establish what is happening. They cannot settle everything worth wanting. That distinction helps explain why an operations specialist might eventually want to start a company himself.
Two founders, two kinds of experience
Salsa began in 2021. Kramer brought experience operating a software business for independent businesses. His co-founder and CTO, Juan Barroso, brought payroll engineering experience: he had built a payroll engine covering ten countries, sold the business to Intuit and spent six more years engineering payroll systems there.
The pairing mattered because the work had two sides. Someone needed to understand what a software platform could offer its customers. Someone also needed to understand the machinery required to pay people correctly. The distance between those two tasks is where a promising product can acquire a very long to-do list.
Investor Sheel Mohnot had known Kramer for years. They had discussed problems from Kramer’s HoneyBook experience that might become the basis of a new company. When the Salsa idea emerged, Mohnot encouraged him to pursue it and Better Tomorrow Ventures invested.
The professional connection extended back to HoneyBook in another way: Oz Alon was among Salsa’s angel investors. A working relationship had become part of the support for a new venture.
and its working day
and operating support
familiar software
Payday belongs where the work is
Kramer’s argument for Salsa begins with a mundane inconvenience. A business already records information about workers in its operating software, then has to get that information into payroll. Hours, tips and commissions become another administrative assignment. Somewhere in that journey, a human being is asked to carry the data.
In his 2024 explanation, he argued that payroll should live inside the platforms businesses already use. Salsa would provide the underlying infrastructure, APIs and interface components. A software company could then offer a payroll experience suited to its own customers.
The distinction between industries matters to that argument. Compensation can combine different hourly rates, commissions and tips. A system needs to accommodate how a particular business works, and a developer needs building blocks that make that possible.
His longer-term ambition reaches into other HR capabilities and financial products for workers. It is an aspiration, rather than a claim that all those products already exist. For now, the immediate test is wonderfully concrete: can the information about the work get to the paycheck with less effort?
The engine they chose to buy
One early Salsa decision puts Kramer’s management logic under a sharper light. The team needed a tax calculation engine. Building one themselves would have required substantial time, followed by continuing maintenance as requirements changed. They chose to work with Symmetry.
Kramer explained that customers would not especially distinguish between calculations built internally and calculations supplied by a specialist. Symmetry’s customer account says that using its engine helped Salsa reach the market at least a year sooner than building the equivalent internally would have done.
There was still integration work. Configuration and unusual cases required attention; purchasing a component did not instantly produce a complete payroll system. Kramer also valued access to Symmetry’s leadership while Salsa was still without a product or customers.
This is the overflow problem again, now expressed through an engineering choice. Every task a company undertakes creates another obligation to maintain it. Choosing a partner can be a decision about which obligations deserve the team’s future attention. The result still has to work, whichever organization wrote the underlying code.
Back at HoneyBook, seventeen days later
In April 2025, Salsa announced a $20 million Series A led by Altos Ventures, bringing total funding to $30 million. The company reported more than tenfold growth in 2024 and another doubling early in 2025. Those were company-reported business growth figures; the announcement did not define them as a specific revenue measure.
The more personal milestone was HoneyBook becoming a customer. Salsa’s team worked alongside HoneyBook’s team in its office, helping shape a payroll experience for its members. Its first customers went live 17 days after the integration began.
For Kramer, that brings the story back to familiar territory. The company where he had learned to build and hand over operating processes was now using the infrastructure his new company supplied.
The appealing part of this loop is its scale. A founder can speak about markets and funding rounds, but an old colleague can ask a simpler question: did the thing make our work easier? HoneyBook’s launch gave that conversation a product to examine.
From the start of integration to first customers live, as reported by Salsa.
Getting everyone out of the little boxes
Kramer’s cultural preferences also traveled with him. In a 2025 post about Salsa’s distributed team, he described investing time and money in gatherings across countries. Earlier offsites had lasted two weeks, combining customer work and product building with time together.
He listed locations across the United States, Argentina, France, Portugal and Spain. The team’s summer gathering in Mexico City was its largest yet, with roughly half the team newly arrived following growth. The accompanying company account described more than 40 people gathering, many meeting in person for the first time.
The photographs supply a welcome change of wardrobe for a story about payroll: team T-shirts, outdoor gatherings, a little less business solemnity. They also fit an earlier conviction. Kramer had defended the cost of culture at HoneyBook well before he had to make those spending decisions at Salsa.
A distributed organization can share documents all day. Becoming familiar with the people writing them takes its own kind of effort. Kramer has repeatedly chosen to budget for that effort.
The work keeps finding another plate
By January 2026, Salsa’s product updates included a shorter onboarding path for businesses paying only contractors, verification for sensitive actions and ways for partners to place their own help content inside the experience. Each addresses a particular moment when a user might otherwise need another explanation or another step.
February brought access to raw payroll journal entries through exports or APIs, giving platforms more options for connecting accounting systems. It also added individual payment memos that travel onto pay statements and printed checks. Sometimes the useful improvement is simply letting a payment explain itself.
These are team releases, and they belong to the people doing the work across Salsa. They show the ongoing detail behind Kramer’s stated ambition. Launching payroll creates a continuing relationship with the businesses relying on it; the product keeps accumulating situations that need an answer.
The refrigerator story ends with a process another manager could inherit. The Salsa story remains in progress, with software platforms handing over work and asking for more. Kramer has changed titles since HoneyBook. His attention keeps returning to the crowded plate.