This is part two of a series of stories by Robert Gaskins who helped invent PowerPoint at Forethought Inc. in 1984 (read part one here). It was the first significant acquisition made by Microsoft. We spoke to Robert about building a startup in the 1980s and what life was like negotiating with, and working for Microsoft. After the sale Robert reported directly to Bill Gates, heading up Microsoft’s business unit in Silicon Valley. He managed the growth of PowerPoint to $100 million in annual sales before his retirement in 1993.
This post looks at what it was like running a startup in the 1980’s and the differences Gaskins and his team experienced back then compared to running a startup today.
Here’s Gaskins in his own words …
A different age
Developing PowerPoint in the 1980s was so incredibly different that a description of how we worked in the mid-1980s doesn’t seem to make any sense, unless one remembers that the Internet was completely unknown outside a small research community in 1984.
I had access to an ARPAnet connection when I worked at the Bell-Northern Research laboratory in Palo Alto in the late 1970s, but I had no real experience with it—it wasn’t something we used every day. Lots of things about the way both startups and larger software companies worked prior to the web don’t make any sense in the current age.
A software startup in the 1980s was a more difficult problem. You had to plan ahead a great distance and “call the shot” accurately, since a considerable investment was required before much in the way of feedback, let alone sales, could be available.
Today it’s the fashion to make web services in a very “lightweight” way, to offer minimally viable web products free, and to iterate rapidly as experience is gained with initial web users. Without the web, none of that was possible.
As a startup you can get crushed by a 100 pound gorilla before you even realise it sat on you. You have to be nimble enough to avoid that whatever era you are working in, whether that’s the 1980s, or 2016. Twenty years after we started developing PowerPoint Paul Graham famously said:
The best way to solve a problem is often to redefine it
This was a path we followed at the time. We certainly had a lot of competitors who were making presentation software for MSDOS PCs and for Apple IIs—that is, for the old command-line operating systems. I drew up a table of competitors with such products, and there were more than 30 of them, mostly selling for between $300 and $700 a copy (in 1983 !) But PowerPoint wouldn’t run on those machines.
The real competitors would be whichever successful, big software companies from that world would first make the jump to Mac and Windows. Mac had early enthusiasts, but it remained underpowered for a long time and wasn’t so attractive to businesses using MS-DOS PCs. Windows—or, at least, something like Mac’s graphical platform delivered for machines that were open and less expensive—was what customers were waiting for.
We were “writing off the entire installed base of PCs,” betting that Windows and Macintosh would grow to dominance. So these existing products were not the problem.
Windows was announced at Comdex in Fall 1983, before Mac shipped. It was much slower in shipping than anyone expected, so we redirected our first version to Mac, but when PowerPoint for Mac shipped in 1987 I thought we could ship a Windows version in 1988. I turned out to be wrong: we didn’t ship the first PowerPoint for Windows until 1990, two years later than planned! By then we had very large competitors from the MS-DOS world, with massive numbers of users, who had only to ship a Windows version to take their users with them. Largest of all (over half the total market) was Harvard Graphics from Software Publishing Corporation (SPC).
There was also Freelance from Lotus Development, popular because it interworked with the Lotus 1–2–3 spreadsheet. WordPerfect was developing a presentation product to go with its market-leading MS-DOS word processor. People in the publishing industry liked Persuasion from Aldus which was a Mac product, but was moving to Windows as rapidly as possible.
But, as it turned out, PowerPoint was the first major presentation application for Windows, and it shipped in 1990 on the same day as the first adequate version of Windows—Bill Gates could and did use PowerPoint to announce Windows 3.0. Freelance and Persuasion shipped on Windows about one year later; Harvard Graphics shipped on Windows about two years later; and the WordPerfect product never really appeared on Windows in timely fashion. A year or two is an eternity in a rapidly exploding market; PowerPoint on Windows immediately became the international standard, and has defended that position for 25 years.
A rocky road
From mid-1984 to mid-1987, we faced the very real prospect of liquidation at least every six months. For three Christmases in a row, we took off a week for the holidays only after carefully calculating that we had enough money in the bank to pay our payroll taxes, accrued wages and vacations, and minimal severances, since it seemed almost certain that we would have to liquidate early in the New Year. This made the holidays considerably less festive. But we went through a dazzling series of improvisations and survived to ship a product. No one considered leaving.
What was very surprising was how long it took Windows to arrive. When Windows had been introduced in the fall of 1983, everyone (including Microsoft) expected it to ship fairly promptly; as it turned out, Windows 1.0 didn’t ship until the very end of 1985 and even so was still very sketchy, so that led us to adjust our plan and ship PowerPoint on Mac first. (This was a tough decision, because we were shipping other Mac software and we knew all too well that even Mac itself wasn’t selling very well). By the time we were acquired by Microsoft, in mid-1987, we expected (and Microsoft expected, in the negotiations) that PowerPoint for Windows could ship by mid-1988, a year later.
There’s no mystery why PowerPoint for Windows took three times longer to develop than we expected – it took so much more work to ship a Windows application than a Mac application. The version we called PowerPoint 2.0 shipped on Mac in May 1998. It had taken 34 person-months of development to get to PowerPoint 1.0 from a blank sheet of paper, and then it took an additional 26 person-months of development to design the second version with color and to ship it on Mac. Total: 60 person-months.
The initial Windows version was PowerPoint 2.0, with the same specs as on Mac, but the additional work just for the Windows version took us 185 person-months of development! Three times as much as all the work to design and ship on Mac! (And this does not include the work on an additional “graphing” module, which was developed separately.)
That was a surprise to us, and to everyone up to Bill Gates; we didn’t realize how much more mature and functional the Mac platform was. The result was that everyone (all of Microsoft’s applications, and all of our competitors) were surprised by how much longer it took to ship adequate Windows applications than anyone had imagined.
It all worked out. We had no Windows revenue for two more years, but because Windows was delayed, sales of our Macintosh version increased in the meantime. We shipped
PowerPoint 2.0 for Windows on the same day as Microsoft shipped Windows 3.0, a very major upgrade, and that got a lot of attention. Our competitors were at least one year later even than we were, up to two years or more, so we were not facing any Windows competitors. Sales of our new Windows version immediately went up to producing more than 90% of our revenue, dwarfing all our Mac revenue, and it stayed that way.
Sales – 1980’s style
We couldn’t reach potential customers directly (there was no web), so we had to groom editors of computer magazines and feed information to them, hoping they would print it in their magazines. We had to travel to the editors to demonstrate our software and leave copies for them. To get in touch with those editors, we had to maintain our own files of contact information, addresses and phone numbers, and employ a PR agency for whom that was their stock in trade. We traveled to meet industry consultants who were frequently quoted in the computer magazines, and we hired them at big prices to “advise” us, expecting them to say nice things to editors in the future.
Traveling imposed large information costs; people didn’t have mobile phones, so had to call in periodically for updates, maybe getting a short document faxed to a hotel and delivered by a bellboy. Traveling abroad could mean being completely out of touch for several days.
We advertised in the computer magazines, hoping to interest potential customers. We paid for 800-number phone lines so that potential customers who wanted information or the names of dealers could call us; some people did call, a few called every day just to chat with our people at our expense. We spent a lot of money to try to collect registration cards, so we would even know the names and addresses of our actual customers who had already spent money with us, since they bought through retailers who bought through distributors.
Beyond the poor information flow, lack of the web imposed all kinds of costs and overheads and delays. The necessity to write and print physical books about the software led to long lead times, and complicated dependencies, such as that the number of pages in the manual had to be known before the dimensions of the box could be fixed for the box design, to be sure that the manual would exactly fit without rattling. The same sorts of problems arose with all kinds of printed material that are no longer necessary because the information is now delivered over the net.
Since there was no way to deliver software over the Internet, we had to manufacture physical diskettes in physical boxes and place them in other companies’ inventory stocks (and finance that inventory). Having to sell those boxes through distributors and then dealers imposed the high cost of a field sales force, which was hard to track and evaluate. Since every product in the hands of a customer had to pay for its manufacturing and overheads at each level of distribution, there was no possibility of free or minimal-cost versions early in a product’s life, which implied some minimum level of function.
This system also imposed very high costs on making upgrades, with the result that the quality of the first-shipped product had to be high:
The cost of an update forced by a single important bug was significant even for Microsoft, and could put a startup out of business.
The software couldn’t just run on our own servers with a known configuration and the possibility of easy updates, but had to run in the unpredictable environments of the users’ own machines, a much harder problem for both development and testing. If a bug corrupted data on a user’s machine, the update had to diagnose and correct such problems all by itself, working in an unknown disconnected environment.
Lack of focus almost killed us
Now, the usual advice is that a startup should focus on one goal and focus sharply. We consciously planned to do the opposite: we built up both a publishing division and a development division, with the two working on different projects at different time depths (and, since a startup is always understaffed, almost everyone had to share their attention between the two projects).
While we developed PowerPoint, over three years, we published three other Mac products developed by others – including FileMaker. “Published” means that we consulted with the developers on design, we did the testing and bug-tracking, we wrote the manuals, we typeset and manufactured the manuals and boxes and duplicated the disks, we designed and ran advertising, we handled all PR and press tours, and we handled all sales (to wholesalers and dealers, as well as direct) and customer support, including technical support.
Immediately after PowerPoint shipped, we got acquired by Microsoft and cancelled our contracts for our published products, abandoning all that and leaving PowerPoint alone to be our sole focus of attention.
When just beginning the development of PowerPoint, at first I very much agreed with the dual-track plan to publish software developed by others as practice to build the delivery vehicle for PowerPoint, and thought it very clever. During the development, I came to feel that our actual experience showed that to be mostly wrong; the published products did not pay for themselves, but ran huge deficits that repeatedly brought us close to liquidation and very nearly killed PowerPoint.
Over three years, while we spent less than $500,000 developing PowerPoint, we spent about $2,500,000 on the losses of our published products!
Partly this was because of the poor general market for all Mac software in 1985 and 1986, and I thought we were stubborn not to recognize that fact and cut our losses on what was after all a sideshow. Because we lost so much in the publishing operation, we had to raise money about every six months, which was a great effort and repeatedly brought us very nearly to bankruptcy.
Doing it differently today
I really envy people doing software with the web. One could do much better these days. But we didn’t have it, and so we were forced to do many things that wouldn’t be sensible today.
The situation reminds me of the oft-repeated story about a high school class studying Romeo and Juliet, who are mystified by all the tragic confusion at the end. “Why didn’t Juliet just use her mobile phone to call Romeo when Friar Laurence gave her the sleeping potion, and get it straightened out with him?”
If you think about it, you realize that much of literature is built on plots that wouldn’t make sense if mobile phones existed. I’ve listened to BBC Radio 4 interviews with detective-story writers who say that they purposely write stories set in the early twentieth century, so that they can take advantage of an age before mobile phones; but that ploy is only going to work for readers who can remember what that far away time was like.
The same is probably true for descriptions of mid-1980s software startups; they can really only be understood by readers who were there.
We even convert PowerPoint files!
This is the second in a series of stories from our interview with Robert Gaskins. Our first post looked at how Microsoft beat Apple to acquire PowerPoint for $14 million.
Stay tuned for our next post in the series – a firsthand account of what it was like working directly with Bill Gates.
Image credits: The “Byte Magazine” cover is courtesy archive.org and their treasure trove of old computer magazines. Other images are sourced from Robert Gaskin’s excellent website covering the history of PowerPoint. His book “Sweating Bullets” remains the definitive read on that topic.