Dave Stephens and the trouble with making work simple

In November 2007, Dave Stephens described a peculiar business achievement. His company had become easier to find on the internet and, in the process, harder to recognize. Coupa had followed advice about search engine optimization, assembled its keywords, and given those keywords plenty of room on its website. The rankings improved. Something else had slipped.

The remedy was pleasingly awkward for anyone paid to celebrate rising traffic. Coupa rewrote the site to respect visitors' time and explain what made the business useful. Traffic fell a little. Leads fell too. The leads that remained were better, and closed business increased. Stephens had acquired an expensive-looking lesson without needing an elaborate slogan: a company could improve its marketing indicators while damaging the thing they were supposed to measure.

It is an unusually good starting point for understanding him. Much of his career concerns the distance between a system that appears to work and an experience that actually helps somebody. A dashboard can be busy. A website can be findable. A worker can spend an afternoon dutifully clicking. None of those activities, on its own, establishes that anything worthwhile has happened.

“brand identity can’t be sacrificed for site traffic.”Dave Stephens, 2007

Before the startup, a purchasing department

Stephens studied at the University of California, Berkeley, from 1987 to 1991, earning a bachelor's degree in chemical engineering. His subsequent career took him into enterprise applications, where the raw materials were purchase requests, supplier records, contracts, and the procedures that make a business function. These are subjects that rarely interrupt a dinner party. They do, however, occupy a considerable part of working life.

At Oracle, he spent close to a decade developing its procurement applications. His own career account dates his work on iProcurement to 1998 and Oracle Exchange to 2000. From 2002, as a vice president, he directed procurement investment and oversaw product strategy, development, and quality assurance across seven products.

That background matters because simplicity meant something concrete in his work. A purchasing system has obligations. It must accommodate rules, people, approvals, and money. Making it pleasant cannot mean pretending those obligations have vanished. The interesting problem is how to carry the complexity without handing every piece of it to the employee. By November 2005, he had decided to leave Oracle and help start Coupa.

A founder learns to leave

Stephens and Noah Eisner founded Coupa, whose company history dates its founding to 2006. The aim was to make purchasing software accessible to more businesses, with an emphasis on ease of use. The early proposition brought together commercial open source software and a service delivered over the web. It offered Stephens another setting in which to pursue the practical problems he already knew.

Then came a less comfortable lesson in ownership. On October 3, 2009, he wrote about his separation from Coupa. The opening was direct. Leaving a company one had founded was difficult, he said, and it had been difficult for him. He still expressed confidence in its product, mission, management, and investors. Affection for a company and departure from it could occupy the same paragraph.

His proposed next step had a wonderfully limited business model: perhaps a very long bicycle ride, with a nod to Forrest Gump. He left the destination open. The account resists the polished founder story in which every exit immediately becomes evidence of a larger plan. At that moment, he allowed himself an unfinished sentence about the future.

The return ticket to a large company

His career continued through a second period at Oracle and a general manager role at ServiceNow. The alternation is revealing. Stephens has worked inside both young companies and organizations large enough to require layers of coordination. That gives his interest in easier software a particular setting: he has also been responsible for running complicated operations.

When ContractRoom announced his board appointment in August 2018, its account described two Oracle stints totaling 13 years, including leadership of a 1,200-person workforce in 24 countries. It also listed integration work involving Sun, RightNow, Taleo, and Eloqua. Those names locate him in the less photogenic part of corporate expansion, where acquiring a business is followed by the difficult task of making its systems live together.

A merger announcement takes a day. The work implied by those integrations takes rather longer. His experience places him close to the consequences of decisions about applications, processes, and organizational habits. It helps explain why his next company would concern itself with the daily mechanics of work. He had seen plenty of mechanics.

The second startup remembers the first

FreeAgent was founded in 2016. When it announced $5.6 million in seed financing in February 2018, Stephens was its CEO. The initial pitch focused on small businesses with useful products that needed help selling them. He described the software as the system he would have liked to possess when starting his first company.

That is a more personal origin for a CRM than a market size calculation. It suggests a founder remembering the work surrounding the product: keeping track of prospects, following conversations, and turning interest into business. For a small team, administration competes directly with the hours available to make and sell anything.

There was continuity on the financing side too. BlueRun Ventures investor Jonathan Ebinger had known Stephens through Coupa. In March 2018, he described Stephens approaching him more than two years earlier with an early idea for improving CRM for smaller businesses. BlueRun supplied seed money and later followed with another investment. The connection crossed companies; the new proposition had to earn its own place.

Montana enters the story

Stephens's public professional profile places him in Bozeman, Montana. A regional development newsletter offers a useful glimpse of what building a software company there meant in practice. In July 2018, the Northern Rocky Mountain Economic Development District helped FreeAgent obtain a Big Sky Trust Fund grant for hiring.

Stephens described the grant as a way to accelerate recruitment of people in Montana. For many of those hires, he said, the job was their introduction to the technology industry. It is a smaller detail than a financing round, and a more tangible one. A company's expansion can be counted in dollars; it can also be counted in people acquiring a kind of work they have never done before.

There is no need to turn that episode into a theory about where every startup ought to live. Its interest lies in its specificity. The founder associated with Oracle and Coupa was also building a team with help from a regional organization. Software can travel widely while a first job remains a very local event.

Two rounds, three years apart

By April 2021, FreeAgent was making a broader case. Its $12.6 million Series A was led by Pelion Venture Partners, with Altara Ventures and existing investor BlueRun participating. The company said subscription revenue had grown 2.5 times and reported six consecutive quarters of growth.

Those figures describe that period, rather than a permanent condition. They nevertheless mark a change in the scale of the undertaking. The early pitch had concentrated on helping smaller companies sell. In 2021, Stephens was presenting FreeAgent as a platform alternative in the CRM market, a field in which businesses already had powerful, familiar options.

The distinction between a convenient application and a platform brings its own obligations. A platform must support different people, records, and ways of working without making each user negotiate the entire apparatus. Growth gives a software company more resources to address that problem. It also gives it more opportunities to complicate the answer. Stephens's old interest in the employee's experience had acquired a larger stage.

A conversation replaces a few clicks

FreeAgent later became servis.ai. In December 2024, Stephens introduced a voice interface, demonstrating its use with a ticketing application. The proposal was straightforward: people should be able to operate business software by speaking to it. A task that requires navigating through screens might instead begin with a spoken request.

Five days after the first demo post, he explained the reasoning in a separate essay. Voice could help people retrieve information, update records, assign tasks, and initiate workflows. His enthusiasm was plain. The useful question underneath it was familiar: how much of the interaction belongs to the user's work, and how much belongs to the machinery required to record that work?

By May 2025, he was introducing himself in a product demonstration as Chief Product Officer. The company's team page uses the same title. In the demonstration, he walks through different views for an administrator, an HR user, and an employee. The ordinary employee gets a narrower view suited to their responsibilities. There is a certain satisfaction in that detail. After the promises about conversational computing, somebody still has to decide which screen a person needs to see.

The guitar beyond the screen

There are lighter details around the edges of this career. His author biography lists watching ice and roller hockey, supporting the San Francisco 49ers, and playing steel string guitar. The guitar is a useful reminder that a software executive has interests that do not need a product roadmap. Sometimes a set of strings is quite enough.

His public writing also stretches beyond product announcements. He has written about freelance work and corporate technology, and in December 2024 said he was working on a book. His plan was to publish chapters before the full release and learn what business leaders found useful along the way. It is a suitably provisional method for someone whose earlier writing includes a frank account of discovering that better numbers could conceal a worse website.

Reading these episodes together suggests a persistent concern with the people on the receiving end of a system. That is an interpretation of his work, rather than a personality test. It rests on specific choices: revising the website, remembering the missing tools of a first startup, hiring people into their first technology jobs, and giving an employee a smaller slice of a complicated application.

The bicycle ride imagined in 2009 offered a temporary escape from business software. His career brought him back to it. The problem waiting there was stubborn enough to survive several companies and a change in the way computers listen: making the machinery useful without requiring everyone else to become its mechanic.