In the company's early years, a swashbuckling president flew high and then crashed.
"Those Who Control the Present
Control the Past ..." - Nineteen Eighty-Four
It’s a chapter in the OpenText story that you won’t find anywhere in the company’s official corporate history book. Ten Years of Innovation, which covers the first decade of OpenText’s history from 1991 to 2001, is replete with stories of technological breakthroughs and strategic maneuvers that enabled the Waterloo firm to become Canada’s largest software company with annual revenues of $3.4 billion US and 14,000 employees on six continents. And yet you won’t find any mention of John Branch, OpenText’s first president, nor is there much space devoted to Gaston Gonnet, one of the company’s three founders. They have been virtually banished from the OpenText narrative.
OED Defines OpenText's Future
To understand the ostracism of these corporate protagonists, one needs to go back to 1984 when the seeds of OpenText were first planted. In that year, the University of Waterloo stunned the world by winning a contract to computerize the venerable Oxford English Dictionary. More precisely, UW was one of three partners chosen for the project.
IBM in Britain would provide computers and expertise so Oxford could work with the electronic data, and a British subsidiary in Florida called International Computaprint would manually key and punch the 60 million words in the mammoth 10-volume main dictionary and four-volume supplement into an electronic database. UW’s job was to combine the two dictionaries into one database that could be organized, displayed, edited and searched.
Waterloo had pulled off this miraculous stunt thanks to a key contact at Oxford University in England. A native of Britain, Mike Brookes, had immigrated to Canada in the late 1950s and landed a job as UW’s first chief building official. Eventually he was lured back to the UK for a similar job at Oxford. When he heard the university was considering an electronic version of its world-renowned dictionary, Brookes could think of no better candidate than Waterloo, which boasted a cutting-edge computer science program.
To lead the OED project, UW turned to two computer science professors—Frank Tompa and Gonnet. A native of New York City, Tompa had studied at Brown University in Rhode Island, then completed his PhD at the University of Toronto. Considered an expert in text manipulation and electronic databases, he joined the UW faculty in 1974.
Gonnet followed a circuitous route to Waterloo. Born in Uruguay, he studied computer science at university there. While still an undergrad, IBM paid for one of his professors to travel to UW to observe students using IBM computers. Intrigued by UW’s leading reputation in computer science, Gonnet enrolled in the PhD program. On completion he took a teaching position at a university in Rio de Janiero. When that experience did not go well, he returned to UW as a faculty member in 1978. While Tompa was an expert in databases, Gonnet was proficient at many aspects of computers and databases. “I have always been an algorithms person,” he said in a 2021 interview, referring to the instructions used to solve computational problems.
Tompa and Gonnet assembled a team of four full-time programmers, a handful of grad students and a revolving cohort of co-op students to organize the giant 600-megabyte OED database. At that time a typical computer hard-drive was 10 megabytes in size, so it was a mammoth task. A key recruit was Tim Bray, a Vancouver software developer who was brought in to manage the project to meet a 1989 deadline set by Oxford.
While much of the heavy-lifting consisted of tagging, coding and marking up the text to recognize the myriad fonts, brackets and punctuation in the printed OED, a critical element was a search engine that could quickly retrieve words and phrases. The job of designing that functionality fell to Gonnet. He was “generally very good at creating software for solving real problems,” Tompa said of his colleague in a 2021 interview.
Although, full-text search programs existed at this point, none was sophisticated enough to recognize the complex structures in the OED. Gonnet had already been tinkering with a program to search and manipulate math software, an application that would eventually lead him to co-found Waterloo Maple, a computer algebra company that still exists to this day under the name Maplesoft.
Math and algebra were full of symbols and variables. So was the OED. Why couldn’t the same program work for the dictionary? Gonnet asked himself. The breakthrough came when he began fiddling with a data retrieval program called “Patricia trees.”
Slowly Gonnet was able to sketch out the basics of a system that could search phrases and strings of text, one of the key stumbling blocks in the OED. The program would prove to be an invaluable tool, one that enabled researchers to find the most frequently used expressions or most obscure quotes from the vast dictionary.
Eventually he passed on the search application to Tim Snider, one of the programmers on the NOED (New Oxford English Dictionary) team, as the crew was dubbed. Snider worked on refining and expanding the search program to the point where it could retrieve data in less than a second. Another key addition to the team was Darrell Raymond who built a display program to render the complicated typography, columns, and fonts on the screen. “It was a precursor to what today we would call a web browser,” Bray said in a 2021 interview.
An important third contributor was Elizabeth Blake whose coding helped create a shorter, electronic version of the OED. Using Gonnet’s program, Blake wrote an application to extract quotes from Shakespeare, Shelley and post-1800 material that went into an abridged version of the dictionary. OED officials, who had been working on a shorter version by hand for years, were astonished. “It was one of the triumphs of the whole OED project,” said Bray.
One of the key takeaways from the OED project was the unique nature of the electronic dictionary itself. With its variety of textual structures nested within each other, it looked a lot like HTML, the yet-to-be invented internet markup language. Thus, the NOED team was able to get a preview of what the internet would eventually look like when it evolved into the Worldwide Web, noted Bray.
Even though work would continue for some years afterwards, the NOED crew delivered the project on time to meet the 1989 deadline for publication of a new combined print edition and the first electronic edition of the dictionary. A huge celebration was held in London to mark the occasion, but NOED leaders were already pondering their next move.
There was a sense they had created something special, a software application that could make some noise in the marketplace. “We realized that searching a large body of text was fantastic,” said Gonnet. “The amount of information you could get was incredibly good.”
With their work largely finished, the three leaders of the NOED project, Gonnet, Tompa and Bray, decided to form a company to commercialize their searching and sorting software and expertise. Other key members of the NOED team were invited to join as investors. About a dozen investors kicked in $10,000 to $20,000 each to form an original working pool of about $200,000. The startup’s first home was an austere, utilitarian office on the fourth floor in the old office tower above K-Mart in Waterloo Town Square.
As the only full-time employee among the three founders, Bray was appointed managing director. Tompa kept teaching at UW while Gonnet accepted an attractive faculty offer at the Federal Institute of Technology in Zurich, Switzerland in the fall of 1989. Even so, he would return for board meetings and continue to chair the board of directors.
After batting about some options, Bray came up with the name. They were searching for text and opening it on a screen. How about OpenText? he suggested. The official name was Open Text Systems.
The publicity garnered from working on the OED opened some doors. Trade shows and publishing companies came calling. By this time, they had merged the search system and browser into one platform. Potential clients “were floored” by how fast it could search a large body of text, said Tompa.
Growing Up and Branching Out
Although the fledgling company was holding its own it had yet to live up to its full potential. One problem was the nature of their customer base. OpenText got some traction with academic, government and publishing customers, including such prestigious names as Stanford, Columbia, and Princeton universities, but it was not a lucrative market. “If your clients are publishers and universities, you don’t have a very rich base to draw from,” said Tompa.
The other speed bump was shareholder resistance to Gonnet’s proposal of moving OpenText from the mainframe model to the burgeoning microcomputer market. He was even prepared to put up more of his own money to make this transition happen. In 1991 he got the okay to pursue the new sales strategy with a new company. The next step was to find a skipper to navigate the uncertain waters beyond the safe harbour of academe; a leader with the requisite business smarts to transform the company from a bunch of professors and computer geeks into a professional corporation. That someone would turn out to be a flamboyant but controversial manager from the Ottawa area named John Branch.
Branch’s decisive and dynamic approach to job interviewing would be a preview of his management style. After some deliberation, Gonnet narrowed the list of applicants to Branch and another candidate. As a final test, he gave each of them three days to compile a business plan for the new firm. Much to his surprise one of the candidates flew all the way to Zurich and presented his plan in person. It was John Branch. Gonnet was impressed, but he was also excited by the business plan. “It was masterfully done. It was impeccable.” Branch got the job.
Although Gonnet was confident of his selection, those back home weren’t sure what to make of the new hire. When he first walked into the OpenText office in Waterloo, Branch was a jarring presence. In his early 50s and overweight, he sported dyed-blond hair, wore a three-piece suit and smoked smelly cigars. The much younger staff was dressed in basic Grunge: T-shirts, blue jeans and plaid shirts. “I’m a computer scientist. You might have a hard time coming up with someone who was my complete polar opposite than John Branch,” recalled Snider, who by now had joined the OpenText staff. After watching Branch in action for a short time, he thought, “This guy is either really good or really crazy and in the end he was a little bit of both.” “He was a really high flyer,” Gonnet said of Branch. “He knew how to behave” in corporate settings. “I was a rural person” compared to him.
It wasn’t long before shareholders at OpenText Systems decided Gonnet had been right along. The company needed a more aggressive sales strategy, and Branch appeared to be the right person to lead the charge. A merger was agreed upon. The new enterprise would be called OpenText Corp. with Branch as president and Gonnet as the majority shareholder and chair of the board. Bray, who had previously been president, was shifted into the position of chief technology officer. One of the first things Branch did was go around and collect from customers owing money to OpenText. “No one had done that before,” said Snider. “We were losing money. Within a month we were making money.”
OpenText Opens a Swiss Bank Account
Another move Branch quickly made was to leverage a contact that Gonnet had with UBS, the largest bank in Switzerland. The vice-president of the bank had been on a committee seeking more entrepreneurial profs for the Federal Institute of Technology in Zurich. Gonnet was identified as a candidate. The VP had flown to Waterloo and interviewed Gonnet who was subsequently hired. When Branch learned of this, he arranged a meeting with the same VP to pitch OpenText’s software for searching financial data from media, stock markets and internal documents. A sale was consummated, and tiny OpenText with fewer than 15 employees suddenly had its first corporate customer, one of the largest banks in the world.
The UBS triumph led to an opening at Ringier, a major Swiss tabloid newspaper publisher. Ringier's hard copy library of over seven million images posed a serious, time consuming, problem for staff hunting for a specific graphic. Seizing the opportunity, Branch pitched OpenText's ability to create a machine-readable index for quick and accurate computer-aided search and retrieval. Naturally, Ringier execs wanted to get a look at this marvel in action. The fact that the hardware was sitting thousands of miles, and an ocean, away in Waterloo did not faze the intrepid OpenText president. Branch arranged for one of OpenText’s bulky Unix work stations to be flown to Zurich. Presumably, Ringier was as impressed with the improvisational skills of OpenText’s head as they were with its software—the deal was done. Gonnet couldn’t have been more pleased. “He (Branch) was flying at a different height.”
With Branch at the helm, OpenText’s velocity increased in the corporate space. Customers such as Universal Document Management Systems, Booz Allen & Hamilton, Mutual Life and Peugot inked contracts to use OpenText’s search and display software. A key selling point was email, which was starting to gain adoption in the corporate sector. Firms were discarding old email messages because they took up a lot of space. “We said, Oh no, don’t throw out your messages,” Gonnet related. “You will want to know information from this client two years ago, and it’s going to be in the mail message.”
But nothing could top OpenText’s reputation for speed. Not only could it search a large body of text in seconds, it also could find phrases and strings of words. This enabled it to win customers in Japan where the language has no spaces between words. The OpenText search program could even handle stop words. Words such as “the” and “to” slowed down search engines and they were configured to ignore them. Not OpenText. It could find the famous Shakespearean phrase “to be or not to be, ”—an expression that contained nothing but stop words—while excluding irrelevant text.
Word of OpenText's success soon spread to the investment market. IDEA Corp., an agency of the Ontario government, invested $600,000 in 1992, and Helix Investments, Canada's largest venture capital firm at the time, pumped in $1 million in 1993. Recruited by Branch, they both gained seats on the OpenText board. Helix's investment so excited Branch that he called the entire staff together and held up the $1 million cheque in triumph.
One of Branch’s greatest skills, according to Snider, was his ability to get the best out of the staff he had on hand. A lot of CEO’s get rid of people when they come in, but Branch never wanted to replace anybody. “He understood people better than anybody I’ve ever known. He could talk to you for five minutes and figure out, here’s how I’m going to get 120 per cent from this person,” said Snider. On one occasion Snider drifted into the coffee room at Allen Square, an office building in downtown Waterloo where OpenText moved in 1992. Branch was chatting with one of his salespeople. They were plotting how he and the sales team were to meet with a potential customer in Washington, DC. Branch's approach was like that of the director of a play or movie. The salespeople were his cast and he'd take great pains to instruct each of them in how and when to deliver their lines in order to mount a box office show. In Branch’s view the goal of the meeting was not to sign a contract, but to find out how much money the client had. “It blew me away that you could do that in sales, like it was just another technical aspect,” said Snider. “John Branch was phenomenal.” In his view, Branch was “the one who took us to being a successful company.”
Out on a Limb with Branch
Not everyone in the office was enamored of Branch. Andrew Van Mil, a software developer who joined the company early in 1992, had a hard time figuring out why Branch had been hired in the first place. To him, Branch was like a Donald Trump character. Showmanship took precedence over substance. For one thing Branch tried to get everyone to wear a dress shirt, tie and dress slacks. And Branch had trouble remembering employees’ names, noted the developer.
But what really caused Van Mil to question Branch’s leadership was his apparent lack of interest in the technology offered by OpenText. Branch was a pure salesman. As long as the product worked, he could sell it—and sometimes whether it worked or not was beside the point. On one occasion, Branch was preparing to visit a customer. “I am going to bring back a cheque,” he vowed to Gonnet. To do this he needed a disk with OpenText software on it. It didn’t matter what the software was. He was confident that as long as he could brandish a CD as a theatrical prop the customer would cough up the cash. Like any good impresario he knew his audience, and sure enough Branch returned to the office cheque in hand.
But Branch’s indifference to technology was not his only flaw. More seriously, he didn’t always play by the rules. In addition to his salary, OpenText agreed to pay him a five per cent commission on every sale he brought in. As time went on, he began collecting the five per cent when the sale was agreed upon, but before OpenText had received the client’s money. Branch had an expensive lifestyle, said Gonnet. He isn’t sure what Branch was spending his money on, but he seemed to have an “appetite” for cash on a monthly basis.
When OpenText found out, it came down hard on Branch, ordering him to stop collecting these premature commissions, said Gonnet. Not long after that, Branch announced a major sale “with big fanfare” at an OpenText board meeting. Bray, who had been talking with the customer in question, alerted Gonnet after the meeting that the prospects were good, but the contract was “far from signed.”
“I should have called a board meeting and made a big stink about this,” Gonnet said. Instead, he decided to put Branch “in the freezer” by demoting him and putting him in charge of the company’s Ottawa office where he lived. In the meantime, OpenText began looking for someone to replace Branch. They found their candidate in Tom Jenkins, a young and ambitious McMaster University grad who was vice-president of sales at Dalsa, another Waterloo tech company. Jenkins joined OpenText as chief operating officer in July 1994. “We hired Tom Jenkins to run the things we were not trusting John (Branch) to run,” said Gonnet.
Kicking the Parent Out of the Nest
With Branch “in the freezer” and Jenkins helping to steer the ship, Gonnet thought his problems were over. He was wrong. In the fall of 1994, Richard Black was troubled. As Helix's representative on the OpenText board, he was beginning to lose confidence in Gonnet’s ability to run the company. He had not cracked down hard enough on Branch, in Black's view, and his ownership of the majority of shares gave him too much control. The company was also teetering on the brink of bankruptcy, according to OpenText’s corporate history. Black began consulting other board members to see if they shared his concerns. It was a difficult decision. The charismatic and gregarious Gonnet had been one of the founders, and the architect of the company’s prized search engine. But with big bucks riding on the outcome, Black felt it was too risky to keep Gonnet and Branch in the picture. The Helix partner had seen the future and it involved Jenkins running the show.
“Tom was quite a bit younger [than Branch], seemed to understand what we were doing, could understand search engine stuff and was way more personable,” said Van Mil. Among the options considered was a plan to dilute Gonnet’s ownership by issuing more shares. At a meeting in December 1994, directors voted to fire Branch and elect a new board without Gonnet. Suffering from heart trouble at the time, Gonnet was unable to attend the meeting to defend himself. Jenkins was promoted to president and Ben Webster, founder and chair of Helix, was elected to chair the OpenText board. Stunned and hurt, Gonnet fought back with a lawsuit to recover his shares. It was settled in 1996 for $300,000, enough to cover his legal fees and the return of his shares, which he eventually sold.
Tompa likened Gonnet’s fate to a classic case of founder’s syndrome. “A company is like a child. The founder gives it birth, helps nurture it to grow and at some point the child is big enough to be on its own.” The kind of management shakeup that occurred at OpenText in 1994 happens from time to time at larger companies, Jenkins said in a 2018 interview. “There are haves and have-nots.”
Under the new helmsman, the Waterloo company would eventually take off to become the software goliath of today. Two important things enabled this to happen, both occurring shortly after Gonnet’s departure. Using OpenText technology, Bray built one of the first search engines on the internet, giving the company an important burst of publicity and fuelling its initial public offering on the stock market. Using its internet savvy, the company then formed a key alliance with a Chicago software company called Odesta. Together they built one of the best companies on the intranet, a key sub-category of the internet where corporations share and manage documents. The management shakeup also enabled Jenkins to consolidate his hold on the company. He would go on to serve as CEO until 2005, then became chief strategy officer until 2013. He continues to chair the OpenText board, a post he has held since 1998.
Reflecting on his days at OpenText and the OED after the passage of so many years, Gonnet is philosophical. “Maybe I have made some mistakes,” he admits, citing his failure to keep a closer eye on Branch. The fact he lived in Switzerland left him in a vulnerable position, he admits. Yet he kept an apartment in Waterloo and attended every board meeting. Unlike other local tech titans, no streets are named after him in Waterloo, but he takes pride in the fact that he played a pivotal role in the creation of two successful Waterloo tech companies, OpenText and Maplesoft.
In 2017, the University of Waterloo marked its 60th anniversary. Banners were erected celebrating the highlights of each decade. For the 1980s, the banner included the launch of OpenText and Waterloo Maple. When Gonnet saw this it salved some of the old wounds. “Those are my two babies.”
And Branch? No one interviewed for this blog seems to know what happened to the swashbuckling executive after his departure from OpenText. He seems to have vanished into the mists of corporate history, like a latter day King Arthur exiling himself to his redoubt of Avalon. Neither OpenText nor the tech sector is likely to ever see a president like him again.