Apps and online services NOV 20, 2026 7 MIN READ
Microsoft Bob
My post on failed products examples covers twelve famous cases from tech and why each one failed.
The timeline as text
- January 7, 1995Bill Gates unveils Bob at the Consumer Electronics Show
- March 10, 1995General availability date in Microsoft's support lifecycle list
- March 31, 1995Bob formally reaches stores on Microsoft Bob Day
- August 1995Windows 95 ships, five months after Bob
- 1996Microsoft pulls the plug in early 1996; Bob 2.0 never ships
- January 1997Gates blames hardware demands and too small a market
- July 2006Ballmer names Bob as a project Microsoft gave up on
- October 7, 2017Melinda French Gates writes that the team killed the project
- Company
- Microsoft
- Category
- Apps and online services
- Launched
- March 1995
- Ended
- 1996
- Stage reached
- Introduction
- AI product
- No
- Status
- Microsoft dropped Bob in early 1996, Technologizer reported; Bob 2.0 never shipped
Microsoft Bob failed because it needed a more powerful PC than most people owned in 1995, and reviewers found its cartoon house too childish for adults. Microsoft dropped it in early 1996, about a year after it reached stores.
What it was
Microsoft Bob was a program for people new to computers that ran on top of Windows 3.1 and turned the screen into the rooms of a house.
Melinda French Gates, who led the project, later described the idea in a LinkedIn post: "To write, you clicked on a pen and paper; to check the date, you clicked the calendar on the wall."
It bundled eight programs, among them a word processor, a checkbook writer, a calendar and a geography quiz, Harry McCracken wrote in his history of Bob for Technologizer. Cartoon "personal guides" walked users through them, with a dog named Rover as the default.
Karen Fries, who had co-led Microsoft Publisher, was its program manager. Fries's proposal drew in part on research by Stanford professors Clifford Nass and Byron Reeves, who had shown that "humans respond to social stimuli even if they are presented from a non-human entity," wrote Tandy Trower, a Microsoft user interface veteran.
Bob needed a 486 processor, 30MB of free disk space and 8MB of memory. McCracken wrote that 8MB was twice the typical amount in a home PC of the time.
What happened
Bill Gates unveiled Bob at the Consumer Electronics Show on January 7, 1995. He predicted its design principles would become "the next major evolutionary step in interface design," the newsletter Soft-Letter wrote, as McCracken quotes it.
The unveiling got "amazingly broad and favorable press coverage," Steven Sinofsky, who was on Microsoft's Office team, wrote years later.
Microsoft had sent copies to journalists in December 1994 and apparently embargoed reviews only until the CES announcement, McCracken wrote, so they began appearing in January, before Bob was in stores. Most weren't glowing.
Microsoft's support lifecycle list, in a 2005 copy, gave Bob's general availability date as March 10, 1995. McCracken wrote that it formally reached stores on March 31, which Microsoft declared "Microsoft Bob Day."
Windows 95 shipped in August 1995, five months after Bob.
Before launch, Microsoft had predicted Bob would sell like its hits Works and Encarta, McCracken reported. According to retail research firm PC Data, only around 58,000 copies were ever sold.
In early 1996, Microsoft "pulled the plug," McCracken wrote. A Bob 2.0 that had been in the works never appeared, and neither did a Mac version Microsoft had talked about.
The Office team adopted Bob's code for the Office Assistant in Office 97, Trower wrote. Its characters included the paper clip better known as Clippy.
Why it failed
Gates later blamed hardware that couldn't keep up and too small a market. Reviewers at the time faulted the design, and later accounts add Windows 95 and a shell too thin to stay inside.
What Microsoft said
I found no Microsoft statement from when Bob was dropped. Its reasons came later.
Gates wrote in a January 1997 column that "the software demanded more performance than typical computer hardware could deliver at the time and there wasn't an adequately large market," McCracken reported.
The same month, McCracken wrote, David Thacher, Bob's former group product manager, told the Orange County Register that the team "didn't spend enough time talking about what Bob did."
In July 2006, Steve Ballmer told financial analysts that Microsoft had almost never decided a project had failed and stopped it, "with a couple of exceptions like Microsoft Bob that I'll cop to in advance," CRN reported.
What reporting showed
Journalists who tried Bob at length were generally "far less impressed" than analysts who had only seen demos, McCracken wrote. His history quotes their reviews.
Stephen Manes wrote in The New York Times that "Bob is a poor neighbor," storing its data in formats that "better programs cannot easily import."
John Dickinson wrote in Computer Shopper that its rooms and characters were drawn "as if the program's target audience were the under-12 set."
The Boston Globe's Michael Putzel wrote that if anyone but the largest software maker had introduced it, "you would never hear of this program."
Bob's 8MB requirement was "a huge amount of memory," the Puget Sound Business Journal wrote.
Not everyone panned it. The Wall Street Journal's Walt Mossberg recommended Bob to anyone who found Windows frustrating, while noting that the computer press had called it "too simple, too corny, too condescending."
What later analysis found
McCracken, writing in 2010, gave a simple explanation for the failure, "It was unappealing," and called its characters "grating and infantile."
Houston Chronicle columnist Dwight Silverman told McCracken there had been "a lot of replacement shells" in the Windows 3.1 era, and that when Windows 95 came out, "such shells were rendered irrelevant."
PC World ranked Bob seventh on its 2006 list of the 25 worst tech products of all time, saying it "was soon buried in the avalanche of hype surrounding Windows 95."
Trower, who reviewed Bob at Gates's request, wrote that it was "perceived as too childlike." Bob was also "a thin veneer," he wrote: once users needed a feature only Windows had, the training wheels disappeared, and not everyone could find their way back. He had warned Fries and Gates at the time that "the implementation was a problem."
Sinofsky wrote of "the abysmal failure of Microsoft Bob" in his Hardcore Software history in 2021. The product "ran into the buzzsaw of technologists who simply didn't buy into the shell or veneer Bob created around Windows," he wrote.
In 2017, Melinda French Gates wrote that the program "needed a more powerful computer than most people had back then, and a lot of critics found the visuals too cute." She also wrote, "we killed the project."
The stage it reached
Microsoft Bob died in the introduction stage, which my product life cycle guide describes as the launch period, when awareness is low and sales are small.
PC Data's 58,000 copies were a long way from the hit Microsoft had forecast, and Bob 2.0 never shipped. McCracken noted that this cut against the company's "famous willingness to stick with new products through multiple versions until they caught on."
Bob never found its market, so I count it as a failure, not a decline.
The lesson
I think Bob's first problem was a mismatch between its users and its requirements.
Melinda French Gates wrote that the team started the project because "only about a third of households had a personal computer." Bob was for people still learning their way around those machines, and by McCracken's account it needed twice the memory a typical home PC had.
If your product is for beginners, make sure it runs on what beginners own.
Bob's simplicity was a layer on top of Windows, and the layer ran out as soon as you needed something only Windows did. That's the reverse of the KISS principle, which is about removing complexity, not dumbing things down.
Microsoft stopped Bob but kept its cartoon helper, and McCracken wrote that far more people met the Office Assistants than ever came across Bob. When you kill a product, write down which idea failed, so the next team doesn't ship it again.
Causes at a glance
Company statementTiming
Gates said it demanded more performance than typical PCs of the time could deliver
Source1
Company statementWeak demand
Gates said there wasn't an adequately large market for it
Source1
Reporting at the timeUnclear value
Reviewers found its cartoon house childish and condescending to adults
Later analysisCompetition
Windows 95 came five months later and made replacement shells like Bob irrelevant
Later analysisQuality and reliability
Insiders called it a thin veneer that dropped users back into Windows
Sources
- 1.Technologizer (Harry McCracken), The Bob Chronicles
March 2010 · Later analysis
- 2.Technologizer (Tandy Trower), Bob and Beyond: A Microsoft Insider Remembers
March 2010 · Later analysis
- 3.Microsoft, Obsolete Products
Company statement
- 4.CRN, Ballmer Analyzes Microsoft's 'One Big' Vista Mistake
July 2006 · Reporting at the time
- 5.PC World, The 25 Worst Tech Products of All Time
May 2006 · Later analysis
- 6.Melinda French Gates (LinkedIn), This failure taught me a lesson I'll never forget
October 2017 · Later analysis
- 7.Hardcore Software (Steven Sinofsky), 042. Clippy, The F*cking Clown
August 2021 · Later analysis
More from the graveyard
Jan 2007
Clippy (Office Assistant)
Microsoft Assistants and chatbots AI
Off by default from Office XP in 2001; left out of Office 2007, released to consumers on January 30, 2007
Apr 2017
Windows Vista
Microsoft Apps and online services
Microsoft ended support, including security updates, on April 11, 2017
Mar 2016
Tay
Microsoft Assistants and chatbots AI
Taken offline after about 16 hours; a brief return ended on March 30, 2016