Part IThe Story
There is a number Jim Goodnight likes to say out loud, and it is thirty. "When you present a credit card to a merchant," he told a WRAL interviewer in the summer of 2026, the year his company turned fifty and he turned eighty-three, "we have 30 milliseconds to say yes or no whether or not we accept it." In that interval the software checks whether this purchase matches this person's past. It looks at frequency. A stolen card, he explained, gets carried "to store after store as quickly as they can," and the hurry itself is the signal. Thieves are impatient. The software is not. The man who built it is less patient than either.
The walls of his office are lined with rocks. The Wall Street Journal noticed this in 2003, and by now nearly everyone who writes about him notices it: petrified dinosaur eggs, mosaic chunks of native granite, large glassy meteorites, fossilized wood, arrowheads made in North Carolina during the Archaic period, which ran from roughly 8,000 BC to 1,000 BC. The collection holds hundreds of pieces from more than forty countries. He started it as a boy in Greensboro, picking quartz and arrowheads out of the ground, and returned to it as an adult with the resources to buy meteorites. He has called rocks and minerals a genre of art. A statistician might also call them a sample, a record of slow processes running for so long that the result looks like it was always there.
So the company has two clocks. One counts in milliseconds and approves or declines a transaction before the cashier looks up. The other counts in millennia and sits in glass cases in the hallways of a corporate campus in Cary, North Carolina, where, according to a 1998 Fast Company profile, the front gate clanged shut at six every evening. Jim Goodnight has occupied the same chair, chief executive of SAS Institute, since the company was incorporated on July 1, 1976. Through that half-century he has run software that decides faster than any person could, and he has decided the largest questions about the company himself, slowly. He took a quarter-century not to take SAS public, then spent five years and tens of millions of dollars preparing to take it public, and then, in April 2026, chose not to again.
The gate is the obvious symbol, and the M&Ms are the famous one. The rocks may tell you more. They did not get where they are quickly, and they have not moved since.
By the Numbers
Fifty Years in One Chair
50Years as CEO, from July 1, 1976 to the company's half-century in 2026
$138,000SAS revenue in its first year of business
$3.2BSAS revenue in 2022
Two-thirdsGoodnight's ownership stake; co-founder John Sall holds the rest
34%Share of revenue spent on R&D in 1993, an industry record
3–5%Annual employee turnover, against a software-industry norm of 20–25%
22.5 tonsChocolate candy distributed to employees every year, on Wednesdays
A Hardware Store, a Satellite, a Time Clock
He was born in Salisbury, lived in Greensboro until he was twelve, and then moved with his family to Wilmington, where his parents, Albert Goodnight and Dorothy Patterson Goodnight, ran a hardware store. From age twelve he worked there after school and on weekends. A hardware store teaches you inventory and patience, and it teaches you that customers come in with a problem rather than a request for a product. They want the faucet to stop dripping and do not much care which washer does it. Decades later, describing what SAS sells, Goodnight would say almost exactly that about software.
He grew to six feet four inches and played basketball, and he later named his high school coach as an early influence. Math and chemistry were his best subjects, thanks in part, he said, to "a wonderful chemistry teacher" at New Hanover High School. On October 4, 1957, he was in ninth grade in Wilmington when the Soviet Union put Sputnik into orbit. Fifty years later, accepting the International Affairs Council of Raleigh's Citizen of the World award, he described the mood: "We thought we were the greatest nation on earth, but suddenly realized that if the Soviets could put a satellite in orbit, they could rain down their nuclear weapons on us at the press of a button." Congress passed the National Defense Education Act. Money poured into science and math. "Kids were going off to college like they were going off to war," he said. He went off in 1961.
He applied to both North Carolina State and the University of North Carolina at Chapel Hill and picked Raleigh, by one account largely because of his friendship with Roman Gabriel, a Wilmington football star bound for NC State. Few college decisions have had larger consequences for a state's economy, and he has enjoyed saying so. A 2013 headline captured him at full rivalrous volume: "Thank God I went to NCSU, not UNC."
He was an applied mathematics major. In his sophomore year he took the only computing course available, which the statistics department taught, and it became the hinge of his life. That summer he worked two programming jobs at once: mornings for NC State's Department of Agricultural Economics, then from one in the afternoon until nine at night for a tobacco company in Durham. "After that summer," he later wrote in the New York Times, "I was absolutely sold that I wanted to be in software." In the fall the statistics department hired him as a part-time programmer, and he stayed on through his junior and senior years. He found it interesting enough to keep going, and there was nowhere else to go anyway. "At the time," he said, "there was no computer science degree."
A 1974 newsletter profile caught what the work gave him: "You can see results, see programs grow and develop." The pleasure he describes is a builder's pleasure, closer to watching a garden grow than winning a contest.
Then came Florida. Partway through his graduate studies, Goodnight went to work for General Electric's Apollo Support Department, building engineering programs and later information-retrieval systems for the ground stations that talked to the capsules. It was the most glamorous engineering project on earth, and he hated the workplace. Guards checked employees in each morning. The coffee came from machines in the hallway, stale, and you paid for it. Turnover was high. Discipline was tight in a way that had nothing to do with the work.
— Jim Goodnight, in the New York Times, October 20, 2002You could be questioned about being five minutes late arriving in the morning. I resented that, since I often came back at night to work. They didn't treat programming like a creative activity.
Most of SAS's culture follows from that sentence. He had been treated as a punch-clock worker by an institution that needed his mind, and he had watched the people around him leave. He drew a conclusion that would cost him nothing to act on for a decade and billions of dollars to defend later.
After about a year, a family illness brought Goodnight and his wife, Ann, back to North Carolina. They had married in 1966. They met while he was a senior at NC State and she was at Meredith College, and she also became an NC State graduate in 1968. He returned to graduate school, took a master's in experimental statistics in 1968 and a doctorate in 1972, and joined the faculty. The story could have ended there, with a tenured professor running regressions on soybean yields. Instead it continued on the fifth floor of Cox Hall.
Two Jims on the Fifth Floor
Growing up in 1950s New Jersey, Anthony James Barr thought his first name was too ethnic. To set himself apart from his first-generation Italian-American peers, he went by his middle name, James, Jim for short. He arrived at NC State as Jim, worked at IBM, and returned to Raleigh in 1966 as Jim to lead a regional effort to analyze crop data. When a graduate student named Jim Goodnight joined him, the two became, inevitably, the two Jims. Barr eventually reclaimed his birth name, but by then he had helped build something that would make his collaborator one of the richest people in America, and he had walked away from it.
The project they were building came out of the land-grant university system. Goodnight described its logic to CIO magazine with an insider's matter-of-factness: one university in every state receives federal money to establish colleges of agriculture and engineering, and "every student they graduate, every PhD, has to come up with something new." New crop varieties, new feeds, new fertilizers. Each experiment produced data, and each dataset needed analysis. In the late 1960s eight Southern universities joined to develop a general-purpose statistical package. NC State housed it because Cox Hall had a mainframe that could handle the volume. The funding came first from the National Institutes of Health and later from a coalition called the University Statisticians of the Southern Experiment Stations. Few institutions sound less like a software startup.
The division of labor was clean. Barr built the framework. He wrote an analysis-of-variance program on thousands of punched cards without a computer to test it on. Goodnight wrote procedures, the statistical routines, including key predictive models. In Goodnight's telling to a public-television interviewer years later, around 1966 or early 1967 "Jim Barr came up with the idea of merging all of these together and put the framework around him and so he contributed some of the systems work and I contributed some of the procedures." Barr was shorter and a few years older. They released the first version in 1968.
What made SAS different was that it was more versatile and more comprehensive than writing analyses from scratch in PL/I or Fortran. Howard Plemmons was an NC State student in the early 1970s whose research involved measuring the food scraps children left on cafeteria trays, about as unglamorous a dataset as exists. He remembered the early language with a craftsman's admiration: "There were just a few procedures in the data set language, and it worked perfectly." Later he became a senior manager at SAS. Many early careers there followed that path, beginning with someone who needed to count something and ending at the company that helped them count it.
NC State first gave the program away. As demand grew, beyond agriculture, beyond academia, beyond the United States, the team began charging to cover service and international shipping. Then it began charging more.
The texture of those years survives in an odd document. In May 1974 the SAS Project Group published the second issue of a newsletter called SAS Communications. The lead item describes Herbert J. Kirk, who built a SAS-based system for North Carolina's new pesticide licensing law, passed after injuries and deaths from misused pesticides. His subsystem graded the licensing exams, kept the licensee records as SAS data sets, and printed the licenses. Below that is a letter from a reader, Harvey J. Gold, reporting a bug in the user's guide. When tied observations were ranked with the FRACTION option, he explained, the procedure returned the fraction for the highest tied rank, not the average: 3/8, or 0.38, instead of 2.5/8, or 0.31. The editors replied without defensiveness: "We are grateful to Dr. Gold for pointing out both our mistake and an alternative use for the RANK procedure. Please make the appropriate changes to Page 73 of your copy of the User's Guide."
That reply contains the whole customer philosophy in small form: a user finds the error, the makers thank him in print, admit the mistake, credit him with the insight, and ask every other user to fix the manual by hand. The company would later formalize this as the SASware Ballot and claim that 80 percent of suggested improvements made it into the software. In 1974 it was a polite note and a request to get out a pen.
John Sall joined in 1973 (more on him shortly). By 1976 the software had about 100 customers. That year 300 people came to the first SAS user conference, in Kissimmee, Florida, a turnout large enough to tell the four people running the project that they had more than a research tool. "The enthusiasm of our users," Goodnight said, "was a big encouragement for us to go ahead and strike out on our own."
In March 1976 four SAS researchers filed to make SAS Institute a for-profit business. Barr owned 40 percent and was chairman. Goodnight, the next-largest shareholder, was chief executive. Sall and Jane Helwig, who had worked on documentation, split the rest.
Three years later, in 1979, after a falling-out with Goodnight, Barr sold his 40 percent for $340,000, about $1.6 million in today's money. In the mid-1990s Forbes estimated the stake would have been worth more than $1 billion. Today it would be worth many times that. Barr founded his own company, Barr Systems. He told The News & Observer he has not spoken to Goodnight since.
— Tony Barr, to The News & Observer, 2026I think it was a stupid decision. I should have just stayed in my little corner and enjoyed the wealth.
He blames himself, not his partner, and what he regrets is leaving. Helwig left a few years later. After that, as Forbes put it, "it was pretty much Jim Goodnight's company." Of the people who started SAS, the ones who stayed became billionaires and the ones who left did not. SAS would spend the next fifty years building a culture on that observation, though nobody says it so bluntly.
Across the Street
The university wanted them nearby, so close that the arrangement amounted to a condition. "As part of the deal with the university, we had to be across the street," Goodnight told CIO. The statistics department still needed help with its consulting questions. So the four moved directly across Hillsborough Street from Nelson Hall. They opened in July 1976 with four employees and a product, Base SAS, of about 300,000 lines of code. In the first year the company took in $138,000.
"We were hoping to make it through the first year," Goodnight said fifty years later. In interviews he is plain-spoken, with a slight gift for understatement. All four had left good-paying university jobs with health insurance. "Most of us were in our 30s by the time we left and had kids."
That detail explains a practice that later became a philosophy. The founders had children who needed to be dropped off at school and picked up afterward. "That's one of the reasons we established sort of a nine-to-five working time," Goodnight said. "So we were one of the first nine-to-five companies around." The 35-hour week, later praised by management scholars as enlightened human-capital strategy, began as four parents arranging work around their families.
Goodnight is honest about the technical side too. "In the early days, we certainly did work more than that," he told PBS. Back then the mainframe turned a program around in 24 hours, and "we had to come back at night because that's the first time we got a turn around from the big main frame out in the distance somewhere." Once desktop machines arrived, he could get "20, 30, 40 turnarounds a day," and the reason to stay late disappeared. The nine-to-five culture was partly kindness and partly an engineer noticing that the bottleneck had moved.
John Sall's path to Cary was even less direct. He was born in Rockford, Illinois, in 1948. He studied history at Beloit College and graduated, as he put it, "into a weak job market," so he went to Northern Illinois University for a master's in economics. There he found statistics and computer science "very captivating." He came to NC State for graduate-level statistics, never finished a doctorate there (the university later gave him an honorary one), and joined the SAS project in 1973. He wrote many of the earliest analytic procedures: time series, econometrics, categorical data, matrix manipulation. In the late 1980s he started work on a Macintosh program that his colleagues called John's Macintosh Project, which became JMP and shipped in October 1989. Two decades later, a billionaire who owned a third of one of the largest software companies on earth, he was still writing code and leading a development team. Asked about his favorite software, he named a Macintosh development environment called Metrowerks CodeWarrior, abandoned during one of Apple's platform transitions, and lamented that he was "stuck with" the free tools that replaced it: "It's free, but it's not as good." Sall is a man who grieves for a discontinued tool, which tells you something about him.
He also gave the best account of the company's early triumphs, which he says did not feel like triumphs. Asked for a "Eureka" moment, Sall answered that "the moments of discovery were more like 'Oh, we were idiots,'" because the team kept having to rewrite the product for each new computing environment: minicomputers, workstations, personal computers, Windows, client-server, the cloud. "But making the commitment to evolve led to an amazing sequence of growth."
By 1979 the company had outgrown Hillsborough Street. Goodnight wanted somewhere more "serene," and the group chose Cary, then a small town west of Raleigh. They moved in 1980. "There was a two-lane road called Harrison that went into Cary, which is three or four miles away," Goodnight recalled. "And there were no places to eat out here." So SAS built a cafe. The amenities, which visitors later read as a deliberate utopian program, began this way: a need came up, and the company met it with a building.
Childcare began the same way. Goodnight remembers that Jane Helwig had a baby "and she was talking about quitting work to raise her baby. And I said, 'Well, let's do a daycare here for you.'" Company histories record that SAS opened on-site daycare in 1980 to keep an employee who had planned to stay home with her child. At the time SAS had roughly a hundred employees. By the time Fortune wrote about it, the daycare served about 850 children at around a third of the usual cost.
The fresh fruit, the reasonable hours, and the M&Ms every Wednesday all began in the first year. Europe came soon after. About four years after leaving campus, Goodnight flew to England to recruit someone to open an office there, because orders were arriving from Europe and the company had been shipping them from Raleigh. A couple of years later the head of European operations moved its headquarters to Germany. Japan came next, then Australia and New Zealand. The company spread around the world while its headquarters stayed on the same road.
The Ledger Behind the Candy Jar
Fast Company's 1998 profile, titled "Sanity Inc.," was written as a fairy tale and admitted it. It described a kingdom, "a secluded realm west of Raleigh," whose inhabitants were "happy, productive, well rounded," loyal to "the kingdom and to its king, who in turn is the model of a benevolent leader. The king — almost unbelievably — goes by the name Goodnight." The gate shut at six. On Wednesdays, by order of the king, hundreds of pounds of M&Ms, plain and peanut in equal amounts, went out to every floor of every building, 22.5 tons a year, ten times what the company bought fifteen years earlier.
Underneath the fairy tale was a spreadsheet, which matters if you want to understand Goodnight and not just admire him. SAS spokespeople have said again and again that the benefits exist for business reasons rather than altruistic ones. The company judged each proposed perk on three questions: would it strengthen the culture, would it serve a significant number of employees, and would it save more than it cost. The on-site medical center is the clearest example. Employees and their families made about 40,000 free visits a year. Fortune reported that the clinic cost about $4.5 million a year and saved about $5 million in productivity that would otherwise have been lost in outside waiting rooms. Sick days were unlimited, and the average employee took two.
Turnover is the larger figure. SAS ran annual employee turnover of three to five percent while the software industry ran twenty to twenty-five. Stanford's Jeffrey Pfeffer, who later co-wrote Hidden Value and used SAS as a central case, estimated that the difference saved the company $60 million to $80 million a year in recruiting, training, and lost knowledge. In a business where the product is code and the code lives in employees' heads, retention is closer to inventory control than to a human-resources metric. Goodnight's hardware-store upbringing shows here. Skilled people who leave take their knowledge with them, and a company that keeps losing them has to rebuild that knowledge again and again.
— Jim Goodnight, to North Carolina magazine, December 2000We are a knowledge company. Ninety-five percent of our assets drive in and out the front gate every day and my job is to make sure they come back.
He used versions of that line for decades, and he meant it literally, the way an accountant would. The amenities were the cost of keeping the assets coming back.
The structure served the same purpose. As SAS grew it added divisions instead of management layers. Pfeffer counted three levels in the whole organization. At one point 26 or 27 people reported directly to Goodnight, until a reorganization around 2000 cut the number to five. "I'm here to sort of nudge the ship along," he said then. He explained the logic to PBS more bluntly than any consultant would have dared: "We only have about three layers of management so there's no chance of advancement. Your salary will continue to grow and in fact some people make more than their managers do for being here so long." He also encouraged people to move to new internal roles every six months or a year. In effect he had removed the corporate ladder and replaced it with lateral moves and a rising salary.
The design has a cost, and his critics have named it, mostly in the words they chose. "More than one observer," Forbes reported in the mid-1990s, "calls James Goodnight's SAS Institute 'the Stepford software company.'" Fortune called it "famously paternalistic." Even Fast Company's kingdom carried an edge: a gate is a boundary, and a king is a king. People who wanted the big moves of a 1990s software career, stock options, a jump to a startup, a steep climb, had no reason to come to Cary, and many of them didn't. The people who came tended to stay. Gale Adcock, a nurse practitioner at SAS who would become its chief health officer, wrote in 2020, in her 29th year there: "We are strongest when we depend on each other to weather tough times." Inside SAS that sentiment is sincere. From outside it can sound like loyalty to a company town.
Google eventually sent people to study SAS. Stanford taught it. Fortune put SAS on every Best Companies to Work For list from the list's inception in 1997 through 2021, and ranked it first in 2010 and 2011. The candy that made business-school audiences smirk had been, from the start, a line item that paid for itself.
Users at the Top of the Org Chart
In 1985 Goodnight sent a memo to all employees redrawing the company's organizational chart. SAS users sat at the top.
It looked like a gimmick and turned out to be an operating principle. In SAS's first year of operation the company began polling users through the SASware Ballot, and it would claim to fold in four out of five suggested improvements. "The one thing we have done over the years is listen to our users," Goodnight told PBS. "We try to please them and if we can please them, that's really all we care about." He said this while describing direct competition with "at least 200 companies," because the system had grown so broad that it competed with someone in nearly every corner.
Users repaid the attention with devotion that outsiders found almost alarming. In March 1996, for the 21st annual user conference, 3,000 customers came to Chicago from around the world for three days of speeches, awards, and talk. Forbes compared them to Deadheads. Veterans wore up to twenty conference badges on their jackets. Others walked the hotel corridors selling self-published books on fine points of the software. There was a quiz bowl, and some contestants studied for days. Howard Dresner, a Gartner research director who often spoke at SAS events, summed it up: "It's a love-in. I was afraid that if I said anything negative, they [the users] would lynch me."
The devotion had a price tag. A typical license ran about $50,000 a year for fifty users. By the late 1990s, all but two of the hundred largest U.S. public companies used SAS. The U.S. Census Bureau counted the population with it, and the government calculated the Consumer Price Index with it. CoreStates Financial employed about 200 people in its retail banking group whose main job was working in SAS. Users had built careers on the language, and that is why they wore the badges.
Keeping them meant constant rebuilding, which Sall's "Oh, we were idiots" makes sound more accidental than it was. In 1993 SAS spent a record 34 percent of revenue on research and development while it built a new menu-driven interface. In 1994 Computerworld found that among the world's fifty largest software companies, SAS spent 2.5 times the industry average on R&D. The code base grew from 300,000 lines to about 6.5 million by the mid-1990s and to about eight million by the time Goodnight told a public-television host it had been "rewritten three or four times." The biggest rewrite came in the mid-1980s, when SAS rebuilt the entire system in C, replacing PL/I, which ran only on IBM. Goodnight later called this the greatest technological disruption of his tenure: "the realisation that we needed to support everything from the new PCs through to mainframes."
Mary Musacchia, who worked in government relations for Goodnight, remembered that his judgment also cut the other way. "Jim can see around corners," she said. "In the early '90s, people were saying mainframes were dead. Jim said mainframes aren't dead, and time proved him right." She quoted a Barbara Mandrell lyric, "I was country before country was cool," and applied it to SAS: "We were making intelligence from data in the 1960s and using computers to do the work. That was Jim. Now the industry talks the talk of data mining, but Jim walked the walk long before it had a name."
What he refused to do was ship before the code was ready, and on this subject the laconic man gets profane. Asked by ZDNet about vendors with different release philosophies, he said: "Well, Microsoft is a monopoly, they can do what the hell they want to." He grumbled that Oracle was "getting to that perch in the database business." At Elon University in 2009 he explained the alternative: "When we develop our software, it's tested for months and months. We believe in getting the bugs out first. And people have learned to trust our software." He described his customers' fear of new releases as an injury other vendors had done to his whole industry. "It's the other vendors, like Oracle and Microsoft, that have given software a bad name."
The users' attachment to the SAS language eventually turned against the company. In 2002 a British firm, World Programming, released software that could run most SAS scripts. In September 2009 SAS sued, alleging that World Programming had used licenses of the SAS Learning Edition to reverse-engineer the product and had copied its manuals. In the English High Court, Justice Richard Arnold cited the Wikipedia entry on the SAS language to conclude that it was a programming language, and therefore lacked copyright protection. He referred the question to Europe. On May 2, 2012, the Grand Chamber of the European Court of Justice ruled that neither a program's functionality, nor its programming language, nor its data-file formats were protected by copyright. Judge William Alsup asked the parties in Oracle v. Google to comment on the ruling.
SAS did better at home. After a three-week trial ending October 9, 2015, a federal jury in North Carolina awarded the company $79.1 million after trebling, finding that World Programming had misrepresented its intentions to obtain the license and had copied portions of the manual. In December 2018, Judge Sara Cockerill of England's Commercial Court refused to enforce the American judgment and, under the 1980 Protection of Trading Interests Act, allowed World Programming to recover non-compensatory damages it had paid.
The case is a strange footnote in an otherwise orderly history. Goodnight's company had built a language so widely spoken that a rival could live inside it and a court could treat it as common property. The loyalty of users like Harvey Gold had turned the language into a commons, and the courts decided that a commons cannot be fenced.
The Bubble He Didn't Board
Pressure to go public peaked in 2000. Here the story of SAS is usually told as a parable, and Goodnight's only real gift to the parable is his stubbornness.
By then SAS had more than $1 billion in revenue, a famous campus, and analysts and bankers who could not understand why it was still private. The company seriously considered selling about 25 percent of itself on the stock market and giving employees stock options. As USA Today's Kevin Maney later calculated, a public listing of Goodnight's two-thirds stake would have placed him "a notch below the likes of Microsoft's Bill Gates and Oracle's Larry Ellison." In July 2001 the Financial Times reported that SAS had delayed its plans for a partial flotation. It never revived them.
Goodnight later gave ZDNet a revealing detail from the preparation. The consultants, Ernst & Young, told SAS it needed to stop running its internal enterprise systems on its own software, because SAS "is just not a recognised brand" for that purpose, and suggested Oracle. "We spent three or four years and $10m and it was just outrageous," he said. "And in the release [of Oracle] that we were working on there were over 10,000 bugs." When the interviewer asked whether that was a reason not to go public, he backed off slightly: "Well it works now." The irritation was plain all the same. Going public had meant first agreeing to look like everyone else, starting with the plumbing.
Maney's 2004 account of the decision used an analogy: It's a Wonderful Life. In the version where Goodnight stays private, SAS remains "a manicured, homey place of on-site health care, hefty benefits and piano players in the cafeteria." In the version where he lists, Wall Street forces cost cuts through the 2001–2003 downturn, and the campus becomes "a much drearier place after waves of layoffs and lost perks." Maney called it "one of the best — or at least luckiest — strategic decisions in the technology industry in the past decade," and the hedge in "or at least luckiest" is the most honest phrase in the piece. Goodnight did not foresee the crash. He just declined to take part in the mania that preceded it.
SAS did well in the bust. It hired aggressively while competitors laid people off, and it picked up talent the public companies had shed. That was the payoff of staying private: no quarterly report to defend, no stock price to prop up, no debt.
The critics were not quieted. Mike Campbell, chief executive of TempoSoft and a former senior executive at SAP, told Maney that Goodnight should go public precisely because the market would make him hungrier. "True greatness requires a grand plan fueled with equal amounts of paranoia and greed," Campbell said. "He can and should set his sights much higher." Campbell meant this as advice, but it also describes everything Goodnight had organized his life against. The Apollo workplace ran on a version of paranoia, and the 2000 offering would have run on a version of greed. Goodnight had seen both and had built a company with a gate meant to keep both out.
The decision mattered again in the autumn of 2008. A Briton named Will McAllister had become a father a month before he joined SAS that October. More than a decade later he remembered a video message from Goodnight assuring employees that no one would lose a job during the financial crisis. "It had a huge impact on me at the time," he wrote, "and has influenced how I have dealt with our teams at this difficult time." The company kept that promise through the recession. Speaking at Wharton afterward, Goodnight reported that 2009 had been SAS's third most profitable year on record, with revenue of $2.3 billion. He spent most of the talk on something else: hippos in Africa, child-welfare fraud in Los Angeles County, and his worry that American children were "just not that into" science and math.
Citizen of a Company Town
In July 1996 The News & Observer ran a profile of him under the headline "Citizen Goodnight." It was apt, though by then the citizenship was hard to separate from ownership.
Cary grew around SAS the way towns once grew around mills, and two of its mayors, Koka Booth and Harold Weinbrecht, worked at the company before going into politics. Goodnight invested heavily in local real estate. He owns the Prestonwood Country Club and the Umstead Hotel and Spa on the edge of the SAS campus. He also financed Chatham Park, a large planned development in neighboring Chatham County. When residents there grew uneasy, he declined to own the project. "Quite frankly, I'm basically a banker for them and I am not involved at all in their daily plans," he told an interviewer in 2014. He then admitted that he had opposed development himself when it threatened his own interests: "We've had to object to some things that have gone on in Cary because it was not good, we didn't think, for SAS." A local critic posted the transcript under the headline "I'm just the banker." Both readings are fair. He is a banker and a neighbor, and on any given day he is one or the other.
His most consistent cause, pursued for nearly three decades with the persistence he brings to code, has been education. Here the Sputnik memory returns. Accepting his Citizen of the World award on October 25, 2007, he described Mount Airy, about 138 miles from where he stood, once the undisputed hosiery capital of the world. "Today, the hosiery capital of the world has moved more than eight thousand miles from here to Tirupur, India." He listed textiles and furniture after it. A son of the hardware store watched the state's industries leave one by one and decided the only lasting defense was a workforce that could do work that could not leave.
In 1996 he and Ann, with John and Ginger Sall, founded Cary Academy, which opened the following year a short distance from SAS as a model for putting technology into every part of schooling. Each student got a laptop. In a Cary Academy boardroom in 2008, having just come from an informal employee gathering called "Java with Jim," the man Forbes described as "both towering and taciturn" found plenty to say. "Today's kids are born with a cell phone in one hand and an iPod in the other," he said. "Yet when they come to school, they enter a classroom with a teacher and a blackboard. Education hasn't changed in 100 years. Kids are bored."
SAS created an educational software division that became Curriculum Pathways and offered it free. By 2014 Goodnight said the company had put more than $100 million into it since 1999, and in 2013 alone more than 42,000 schools and nearly 120,000 teachers used it. At one point 99 percent of North Carolina high schools had registered.
This work put him at odds with his own party. In 2014 he sat in Fortune editor-at-large Stephen Miller's office looking, in the magazine's phrase, "positively glum," and "a little hangdog on even his best days." The cause was not SAS. It was the North Carolina General Assembly. "They keep cutting taxes, which lowers revenues, and then they have to cut the state budget and freeze teachers' salaries," he said. "The average starting salary for teachers in North Carolina is $30,800." He had raised the issue with Washington, and Washington had told him it was a state matter. "The only problem is that states are doing a bad job." He also said: "You can talk all you want about the income gap, but if you don't get education right, it won't go away."
His politics have always been those of a business Republican, which meant they came under strain. On January 6, 2021, Goodnight turned 78. That afternoon a mob broke into the U.S. Capitol, and that night two members of Congress he had supported voted to object to certifying the presidential election: Richard Hudson of North Carolina, to whom he and affiliated committees had given more than $50,000, and Virginia Foxx, to whom he had given $2,800. He had given more than $300,000 to Republican candidates and committees in the 2020 cycle. SAS spokesman Trent Smith issued the statement: Goodnight "does not plan to continue support for elected officials who did not support the integrity of the election process, which is so vital to our democracy." For a man who rarely says anything publicly, it was a pointed statement. He let a spokesman deliver it.
His quieter campaigns have gone further. On January 17, 2019, he stood at Millbrook Elementary School in Raleigh with other North Carolina chief executives to argue for expanding the state's pre-kindergarten program to 75 percent of eligible children. He helped organize a Business Roundtable report on early learning and third-grade reading. The Goodnight Educational Foundation endowed a professorship in early literacy at Western Carolina University in 2023 for $1.5 million.
The giving that bears the family name most clearly is at NC State. In 2008 Jim and Ann Goodnight founded the Goodnight Scholars Program for students majoring in STEM or STEM education. Its eligibility rule shows what kind of donor he is: it is limited to North Carolina students from low- and middle-income families, with family incomes that the program described in 2015 as ranging from $30,000 to $99,000. That is roughly the income of families who run hardware stores. The award was $17,500 a year in 2015, $20,500 in 2019, $23,000 in 2024, and now $25,000. Each year fifty first-year students and fifty transfer students are chosen, and the class entering in 2024 came from 34 counties. The program's motto is "pay it forward." The Goodnights also endowed 28 named faculty positions at NC State, and the building that houses mathematics and statistics, opened in 2009, is called SAS Hall. Ann Goodnight served sixteen years on the NC State Board of Trustees and eight on the University of North Carolina Board of Governors. She also became a serious art collector, and the museum park at the North Carolina Museum of Art bears both their names. In that marriage both partners collect.
There is a pattern under all this that he never states. The software was built to help the agricultural colleges produce new knowledge every year. The scholarships, the professorships, the master's program in analytics that SAS helped start at NC State in 2007, the free classroom software, and the pre-K campaign are all attempts to produce people the same way, steadily and close to home.
No Procedure for Succession
On July 12, 2021, the Wall Street Journal reported that Broadcom was in talks to buy SAS. The next day Goodnight sent an email saying the company was not for sale. The talks ended. That same month SAS made its own announcement, which few people would have predicted a decade earlier: it was preparing for an initial public offering, which it hoped to complete by 2024.
The reasoning, as company executives later explained it, was not mainly financial. SAS had no debt and, as chief operating officer Gavin Day put it, no pressure to go public by any particular date. The offering was a succession plan. Goodnight was 78 and owned two-thirds of the company, and Sall, the only other owner, was in his seventies. Listing the company would turn a personally held company into an institution that could survive both men. It would also let SAS offer stock options, the incentive the company had famously done without for forty-five years.
Preparation was slow and expensive. By September 2023 SAS had spent between $50 million and $60 million getting ready, and it pushed its timeline back because the work was taking "longer than anticipated." It consolidated its financial systems into a single new system, finishing in January 2025. In 2023 the company that had kept its no-layoff promise through the 2008 crisis carried out a staff reduction plan. It dropped off Fortune's Best Companies list after 2021. The New York Times had already run a 2009 headline foreshadowing this period: "At a Software Powerhouse, the Good Life Is Under Siege." The siege came from open-source tools that students learned for free, from cloud platforms, and from a generation of data scientists who wrote in Python and had never needed to attend a quiz bowl.
The company's response was to invest heavily. In 2019 SAS announced $1 billion for artificial intelligence research. In 2020 it partnered with Microsoft to run SAS on Azure. In May 2023 it pledged another $1 billion over three years for AI applications in banking, health care, and insurance. Bryan Harris, the chief technology officer, framed it in the Goodnight style, measured in half-centuries: "We think this is where the second leg of growth of SAS over the next 50 years is going to be." In November 2024 SAS bought Hazy, a British company that generates synthetic data, data that looks real enough to train models on without belonging to anyone.
The press began calling the founder something that made him visibly uncomfortable. In 2019 CNBC introduced him as the "Godfather of AI" and noted that when he started, computers ran 300 instructions a second with 8K of memory. Goodnight is more careful about the technology than the title suggests. In 2026, asked by WRAL about generative AI, he drew a line that a statistician raised on balanced experiments would draw: "We don't use AI to produce answers, but we use it to try to explain the answers." When SAS's software flags a money-laundering pattern, the language model's job is to write the explanation, not to make the call. "Generative AI is not going to replace humans," he said. "But humans need to be part of the loop." He described a small group working on quantum computing, where the hardware has improved and the algorithms have not, the same position his team was in when Cox Hall had a mainframe and almost nothing to run on it.
In May 2025, SAS named Gavin Day, a longtime company executive, chief operating officer and put him in charge of the offering. Day said SAS would go public "when the market is ready." The market was apparently ready. The company was not, or its founder wasn't. On April 22, 2026, The News & Observer reported that SAS had walked back its pledge to go public, a move the paper said "creates space for Goodnight succession options." Which options, nobody announced. After fifty years and two near-misses, the company remains private, owned by two men, one of whom has run it since the day it opened.
There is a LinkedIn post from 2020, written in Goodnight's plain register, that reads differently now. "My advice has always been to recognize when to stop digging a hole," he wrote. "Once that hole is deep enough you might as well just get out of there and stop digging." He wrote it about failed products. He may have had something larger in mind.
That July, WRAL asked him the question everyone asks: why did he stay in North Carolina? His answer was about as short as an answer can be. "Well, I was born in North Carolina, and I was raised here in North Carolina. So North Carolina's home."
In Raleigh, the North Carolina Museum of Natural Sciences is joined to its Nature Research Center by a walkway called the Betsy M. Bennett Bridge to Discovery. Visitors cross it from the older museum to the newer building, from the exhibits to the labs. Pieces of Jim Goodnight's collection are on loan there, set in cases along the crossing: stones that were old before anyone thought to count anything. They sit in the one part of the building that everyone passes through and nobody lingers in.
Part IIThe Playbook
What follows distills fifty years of SAS into twelve operating principles. Some will travel well to any company. Others work only with Goodnight's specific advantages: majority ownership, an enormously profitable niche, and the patience to wait decades. The last principle is less a recommendation than a warning about the problem he has never solved. Read them as a whole. His methods depend on one another, and copying one without the others tends to fail.
The Goodnight Playbook
- 1.Build the company around the lives its founders actually lead.
- 2.Treat every perk as a line item that must pay for itself.
- 3.Put the user at the top of the org chart, literally.
- 4.Grow by adding divisions, not layers.
- 5.Pay for R&D the way you pay rent.
- 6.Ship late and let trust compound.
- 7.Refuse the liquidity event that would rewrite your incentives.
- 8.Rewrite the foundation before the platform moves.
- 9.Hire into the downturn.
- 10.Invest upstream of your own talent pipeline.
- 11.Know which hole to stop digging.
- 12.Solve succession before the founder becomes the moat.
Principle 1
Build the company around the lives its founders actually lead
SAS's nine-to-five culture did not come from a management theory. It came from four people in their thirties who had given up university health insurance and had children to take to school. Goodnight says so without embellishment. The daycare had an equally concrete origin: a co-founder with a newborn who was about to quit.
The lesson is to build the institution around the real constraints of the people founding it and then make those constraints permanent, before growth tempts you to drop them. Most startups do the opposite. They design for founders who are temporarily single, sleepless, and willing to give up everything for eighteen months, and then wonder why the culture breaks when those founders get older. SAS's early hours worked because they did not depend on its founders being young. A company built around midlife can keep people through midlife, which is exactly when software engineers become most valuable and most likely to leave.
Tactic: Write down the three non-negotiable constraints of your own life today, and build them into company policy before you hire employee number ten.
Principle 2
Treat every perk as a line item that must pay for itself
Outsiders saw the M&Ms. SAS saw the arithmetic. Every proposed benefit faced three questions: would it strengthen the culture, would it serve a significant number of employees, and would it save more than it cost. The medical center cost about $4.5 million a year and saved about $5 million in lost productivity. Turnover of three to five percent, against an industry norm four to five times higher, saved by Pfeffer's estimate $60 million to $80 million a year.
That discipline is what kept the perks alive. Generosity justified as kindness gets cut at the first bad quarter. Generosity justified on a balance sheet is protected by the balance sheet. When Goodnight said that ninety-five percent of his assets drove out the gate every night, he meant that the benefits were maintenance spending on those assets. The framing also made the culture defensible to skeptics who would never have bought a values argument.
Tactic: For your next proposed benefit, estimate its annual cost against the retention and productivity it protects, and kill it if you cannot make that case on paper.
Principle 3
Put the user at the top of the org chart, literally
In 1985 Goodnight redrew SAS's organizational chart with users at the top. The gesture formalized something that went back to the 1974 newsletter, in which the editors thanked a reader named Harvey Gold for finding an error in the RANK procedure and asked every user to correct page 73 by hand. The SASware Ballot began in the company's first year, and SAS claimed it incorporated 80 percent of the suggestions it received.
The payoff was a user base that behaved more like a guild than a market: customers studying for a quiz bowl, wearing twenty badges, selling their own books about the software, and building careers on it. That kind of loyalty raises switching costs far beyond anything a contract can enforce. It cannot be manufactured with marketing, and it cannot be faked. It comes from users seeing their own suggestions show up in the next release, year after year.
Tactic: Publish a running list of shipped features that came from customer requests, with credit to the requester, and review it yourself each quarter.
Principle 4
Grow by adding divisions, not layers
SAS grew to more than 12,000 employees with about three levels of management. Goodnight did not hide the trade-off: "there's no chance of advancement," he said, but "your salary will continue to grow and in fact some people make more than their managers do." Internal moves every six months to a year provided variety, and a rising salary provided progress, without the promotion ladder.
The design solves a problem most companies never notice. Managerial hierarchies turn good engineers into mediocre managers because promotion is the only way to get a raise. Separating pay from title keeps experts doing expert work. It requires real comfort with ambiguity: no titles to fight over, and a chief executive who at times had more than two dozen direct reports. It also screens out people who need visible status, and that screening is part of what the design is for.
S
SAS against the industry default
How the company's operating choices diverged from software-industry norms, and what each choice primarily protected.
| Practice | Industry default | SAS approach | Protects |
|---|---|---|---|
| Working hours | 60-hour weeks, all-nighters | 35-hour weeks; nine-to-five since 1976 | Retention |
| Career growth | Promotion into management | Three layers; pay rises without title | Retention |
| Compensation upside | Stock options | Profit sharing; no public equity | Control |
| Release cadence | Ship, then patch | Test "for months and months" | Trust |
| R&D spend | Roughly industry average | 2.5x average (1994); 34% peak (1993) | Product |
| Capital structure | Venture, then IPO | Private, debt-free, founder-owned | Control |
| Downturn response | Layoffs | Hire aggressively; no-layoff pledge in 2008 | Countercyclical |
Tactic: Build a salary band that lets your best individual contributors out-earn their managers, and announce it before your first engineering manager is hired.
Principle 5
Pay for R&D the way you pay rent
In the 1990s SAS led the industry in R&D intensity for eight straight years and peaked at 34 percent of revenue in 1993. Two decades later it still put roughly a quarter of revenue back into development, nearly double what other large software companies spent. Analysts credited this spending, more than the culture, for the company's long run of growth.
The point is that R&D was not cut when times were bad. A public company treats research as the easiest place to find savings when earnings disappoint. Goodnight treated it as fixed overhead, like rent, which he could do partly because no outside shareholder could stop him. The result was that SAS kept rebuilding itself through every platform change, and each rebuild deepened its moat. That consistency matters more than any single big bet, because the bet is only as good as the years of steady spending behind it.
Tactic: Set an R&D floor as a fixed percentage of revenue and require board-level approval to go below it in any single year.
Principle 6
Ship late and let trust compound
Goodnight's contempt for buggy software is personal and long-standing: Oracle's 10,000 bugs, Microsoft's monopoly that let it "do what the hell they want to." His alternative was to test for months and accept slower release cycles. "People have learned to trust our software," he said, which was precisely the point.
In software that approves credit-card transactions within 30 milliseconds, prepares drug data for regulators, or calculates the Consumer Price Index, reliability is the product. A bug is not a nuisance there. It is a breach of the reason customers bought the software. Customers pay $50,000 a year for fifty seats because they believe the numbers. Each release that works reinforces that belief, and each failure damages it far more than one success builds it. That asymmetry makes conservatism the rational strategy for anyone selling certainty.
Tactic: Identify the single failure mode your customers fear most, and make preventing it an explicit release gate that no deadline can override.
Principle 7
Refuse the liquidity event that would rewrite your incentives
Twice, in 2000 and again after the 2021 Broadcom talks, Goodnight came close to selling or listing SAS and pulled back. He understood what an outside owner would require. Campbell said it plainly: greatness requires "equal amounts of paranoia and greed." A public SAS would have cut costs in the 2001 downturn. A private SAS hired.
Barr's story shows the founder's side of this. He sold 40 percent for $340,000 in 1979 and later called it "a stupid decision." Liquidity offered early is usually priced against the founder. Liquidity offered at the peak, as in 2000, is usually priced against the culture. Saying no requires a business that throws off enough cash to need no outside money, and SAS had that, which is the condition this principle rests on.
G
Fifty years of saying no (and almost saying yes)
The decisions that kept SAS private, in sequence.
1979
Tony Barr sells his 40 percent stake for $340,000; ownership consolidates around Goodnight and Sall.
2000
At the peak of the tech bubble, SAS weighs listing roughly 25 percent of the company and issuing options; it spends years and about $10 million moving its internal systems to Oracle in preparation.
2001
The Financial Times reports SAS has delayed its partial flotation; the company hires aggressively through the downturn.
2008
Goodnight tells employees in a video message that no one will lose a job in the financial crisis.
2021
Broadcom talks are reported July 12; Goodnight emails July 13 that SAS is not for sale; the company announces IPO preparations targeting 2024.
2023
SAS has spent $50 million to $60 million on IPO readiness and pushes back its timeline.
2025
Gavin Day becomes COO to lead the offering "when the market is ready."
2026
On April 22, SAS walks back its pledge to go public.
Tactic: Before any financing or sale conversation, write out the three practices you would refuse to change under new ownership, and walk away if the buyer cannot commit to them.
Principle 8
Rewrite the foundation before the platform moves
Sall says the company's discoveries felt like "Oh, we were idiots." That sounds modest but describes a discipline. SAS rewrote its core again and again: from IBM-only PL/I to C in the mid-1980s, then for minicomputers, workstations, PCs, Windows, client-server, and the cloud. The code base grew from 300,000 lines to about eight million and was rebuilt three or four times along the way.
The judgment ran both ways. Goodnight ported to PCs before most of his competitors, and in the early 1990s he also refused to abandon the mainframe when the industry declared it dead. Musacchia's phrase, "see around corners," describes a habit more than a gift. Ask what your customers will run in two years (Goodnight's planning window) and be willing to throw away working code to meet them there. Being an incumbent is not the danger. Being an incumbent who protects old code is.
Tactic: Once a year, ask your lead engineers which part of the core they would rewrite from scratch if they could, and fund the top answer.
Principle 9
Hire into the downturn
From 2001 to 2003 SAS hired aggressively while the technology industry shed workers. In 2008 it promised no layoffs and posted its third most profitable year in 2009. Counter-cyclical hiring works because talent is cheapest and most available exactly when public competitors are forced to let it go.
This depends on two conditions most companies do not meet: a balance sheet with no debt, and an owner who answers to no one about quarterly margins. It also has a second effect that compounds. Employees who watched their company hire while others laid off become, like Will McAllister, people who repeat the story to their own teams a decade later. Loyalty earned during a crisis lasts longer than loyalty bought with perks.
Tactic: Keep a standing list of the ten people you would most like to hire, and set aside cash so you can move on them the quarter a recession arrives.
Principle 10
Invest upstream of your own talent pipeline
Goodnight's educational giving looks like philanthropy, and it is, but it also works as supply-chain management. SAS helped NC State start the country's first master's in analytics in 2007, partly because, he said, "there's a little bit of a shortage in the world of analytics experts." The Goodnight Scholars, 28 endowed faculty positions, SAS Hall, Cary Academy, more than $100 million in free Curriculum Pathways software, and a pre-K campaign all reach further and further back toward the start of the pipeline.
Fortune noted the self-interest, since young users learn tools that include some of SAS's own, and then noted that it was "hardly Goodnight's aim." Both things are true. The insight is that an owner who expects to stay in one place for fifty years can justify investments that pay off over twenty. A company that plans to be acquired in five years cannot.
Tactic: Pick one university program near your headquarters and fund a curriculum in the exact skill you find hardest to hire for.
Principle 11
Know which hole to stop digging
"Recognize when to stop digging a hole," Goodnight wrote. "Let go of your pride or investment if a product or venture isn't succeeding." He added that he did not punish himself or his employees for mistakes, because they were "part of the process of innovating and growing."
The IPO record tests this principle. The company spent years and about $10 million on the Oracle migration for the first offering, then abandoned it. It spent $50 million to $60 million preparing for the second, then walked away again. Either decision can be read as wise or as indecisive. What Goodnight models is a willingness to treat sunk costs as truly sunk, even when they are large and public. Most executives keep spending to justify what they have already spent. He stops.
Tactic: For every major initiative, write down at launch the specific evidence that would make you abandon it, and review that list on a fixed date.
Principle 12
Solve succession before the founder becomes the moat
This principle is a caution drawn from an unresolved problem. The qualities that made SAS durable, a single owner, a long horizon, and resistance to outside capital, also mean the company has no obvious path beyond the man who has run it since 1976. The 2021 IPO plan was described as a succession mechanism, and in April 2026 the company withdrew it, "creating space," in the newspaper's words, for other options that have not been announced.
Every founder-led company eventually faces this. The longer the founder runs things well, the more of the company's logic exists only in that person's judgment. SAS kept that logic in a culture, and cultures can outlive founders. It also kept it in a two-thirds ownership stake, and ownership stakes pass through estates. That is the remaining open problem in Goodnight's otherwise long-sighted approach: how to make a company outlive the person who designed it.
Tactic: Name a credible successor and a mechanism for transferring ownership while you are still fully in charge, then test the plan by stepping away for a defined period.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— Jim Goodnight, to WRAL, July 2026I say if you treat people like they make a difference, they will make a difference.
— Jim Goodnight, to PBS's LivelyhoodI have seen programmers that have been working 12, 15-hour days and the next day they come back in and have to correct all the bugs that they put in their code because they were too tired to think about what they were doing. So it really doesn't pay off.
— Jim Goodnight, on SAS's leadership philosophyInnovation is the key to success in this business, and creativity fuels innovation. Creativity is especially important to SAS because software is a product of the mind.
— Jim Goodnight, at Elon University, April 9, 2009I've never been a really hands on, top down manager. I don't really believe in it. It works for some CEOs, but what I like to do is find really good people, give them the ball and let them run with it.
— Jim Goodnight, to WRAL, July 2026We don't use AI to produce answers, but we use it to try to explain the answers.
Maxims
- Your assets leave every night. In a knowledge business, the job of the chief executive is to make sure they come back in the morning.
- Every perk needs a ledger. Generosity that pays for itself survives a bad year, and generosity that doesn't gets cut first.
- Put the user above yourself on the chart. The customer who finds your bug and gets thanked in print becomes a lifelong advocate.
- Pay without promotion. Raising salaries without raising titles keeps your best experts doing expert work.
- Fatigue writes bugs. Long hours borrow against tomorrow's productivity at a punishing rate.
- Test until the trust compounds. When you sell certainty, one bad release costs more than ten good ones earn.
- Rewrite before you must. The core you are proud of is the one most likely to hold you back on the next platform.
- Be greedy in the downturn. Talent is cheapest when your competitors are forced to let it go.
- Fund the pipeline twenty years out. The scholarship you endow today becomes the engineer you hire later.
- Stop digging when the hole is deep enough. Sunk costs, even large public ones, should never decide the next dollar.
