The Early Days of Visio Corporation

Recollections by Ted Johnson, Visio Co-founder, 27 March 2014

Summer 1990

When Visio’s initial two co-founders, Jeremy Jaech and Ted Johnson, met in the summer of 1990 to discuss starting a company together, only two things were nailed down: they wanted to build an application for Microsoft Windows and they wanted to build a drawing application. They believed that Windows was poised to become the standard operating environment on business PCs and, consequently, the best platform for which to build a business application. Their interest in building a drawing application was based on their personal experiences with hard-to-use CAD applications, hard-to-use graphic arts illustration software, and general-purpose Macintosh-based drawing applications, notably MacDraw.

One more experience significantly informed their interest in building a business drawing application. During this time, Johnson’s wife worked as an internal auditor at Seattle First National Bank. As part of a departmental audit, internal auditors created flowcharts of the business processes and controls of the department’s business. They used MacDraw to create these flowcharts and found the experience less than ideal. But the bank felt the value of process visualization outweighed the difficulty of creating the diagrams.

From this, Jaech and Johnson concluded three things:

  1. People in business who are not artists sometimes had to draw (or diagram) as part of their job;
  2. Even on the graphical Macintosh, no great tools existed for doing this; and
  3. No tools at all yet existed for doing this on Microsoft Windows.

This led the founders to further explore the idea of “business drawing and diagramming” and quickly discovered that most people in business who had to draw (this was 1990), were most likely to draw by hand on paper, perhaps graph paper, using rulers and, often, plastic stencils (or templates).

Plastic Flowchart StencilPlastic Electrical Schematic Stencil

The “discovery” of these plastic drawing stencils helped open the founders’ eyes to the wide variety of common shape libraries that existed and, therefore, the wide variety of types of drawings and types of users creating those drawings. (This exploration continued over the years and more and more drawing types and categories of users were uncovered leading to product line extensions and the Visio Shapes products.)

These elements merged to form the original idea for Visio: Visio would make drawing creation easier by enabling users to assemble drawings and diagrams from standard shapes organized into recognizable collections (stencils). Moreover, it would enable the creation of a wide variety of drawings by using a common drawing engine customized by these shape libraries.

Fall 1990

With this initial product definition, Jaech and Johnson recruited their third founder, Dave Walter, and, over the next several months, convinced the original four founding engineers to leave their jobs and join the new start-up. Those founding engineers were Mitch Boss, Mark Davison, Richard Miyauchi, and Peter Mullen.

As the team refined the original stencil idea and began to design the architecture of the product, three things emerged that would ultimately be extremely important to the long-term success of Visio.

  1. Many of the drawing types we intended to support were “connected diagrams” such as flowcharts and electronic schematics. To enable the creation and subsequent editing of these diagrams, connectors needed to “stick” to the shapes they connect. Also, straight line connectors were insufficient. Connectors needed to support right-angle bends and arrowheads.

    Simple flowchart showing connector lines with bends and arrowheads

    Because Visio retained the relationship between shape and connector and, therefore, between shape and shape, it would be possible to extract these relationships and build an abstract graph from the diagram. This was believed to be an enabler of richer add-on solutions.
  2. Visio’s shapes could not be simple “dumb geometry.” As we experimented with the shapes we found in the many plastic stencils we examined, it was clear that simple geometric scaling would result in ugly diagrams and be unacceptable to our users. A classic example of this behavior is the flowchart’s document shape.

    Illustration of smart vs. dumb sizing using a flowchart document shape

    A more extreme example is the block diagram’s arrow.

    Illustration of smart vs. dumb sizing using a block diagram arrow

    In addition to smart geometry, we represented every graphical and non-graphical property of a shape as a programmable “cell” in what we dubbed a ShapeSheet®—a spreadsheet-like representation of the data underlying a shape. Like cells in spreadsheet programs, Visio ShapeSheet cells could hold values or formulas. Formulas could contain references to cells in the ShapeSheet of any shape on the page. The combination of formulas and references everywhere enabled such features as changing the color of one object and changing the color of all objects that reference it. Cross-ShapeSheet references were also the foundation of Visio’s “glue” feature.
  3. Visio’s stencils (collections of shapes) had to be end-user extensible. We believed this was important because we (Visio Corporation) could not possibly create every possible shape needed by our customers. Also, by opening up the shape creation process, including the SmartShape® technology which was used for representing non-linear scaling and smart properties, Visio could potentially create a third-party market in so-called “shapeware.” (Shapeware was, in fact, the company name when it launched Visio 1.0.) This openness, along with a mechanism for creating programmatic “add-ons,” allowed third-party solution developers to build products using Visio as a diagramming tool.

Looking toward that future when Visio would be used as a platform in third-party tools, Visio 1.0 contained an editable name and three generic data fields on each shape. This information could be used in conjunction with the connection data, shape type, and master shape to extract models from drawings. Subsequent versions expanded this idea of non-graphical data stored on a shape or connector.

Visio 1.0 Format Special... dialog
Visio 1.0 Format Special... dialog

While the development team was refining the design and beginning early implementation, CEO Jaech met with venture capitalists. In early 1991, Jaech secured first-round funding of $800,000 from Technology Venture Investors (TVI) of Menlo Park, CA.

1991

1991 was a year of hard-core coding while the Visio idea and design took shape. Visio was coded in the C Programming Language targeting Windows 3.0, which had been released in early 1990. The Visio development team had all previously worked at Aldus on PageMaker which ran on Windows 1.0 and 2.0. Compared to those earlier versions of Windows, version 3.0 was much improved, though still a 16-bit architecture. Windows’ real breakthrough version—Windows 95—was still many years in the future.

Toward the end of 1991 we brought on Shivonne Byrne as Visio’s VP of Marketing. Byrne was instrumental in crafting Visio’s positioning and marketing messaging as well as guiding the efforts that would result in Visio’s name and distinctive branding.

Toward the end of 1991, the executive team was preparing to go on the road to raise a second round of venture capital. As part of that preparation, we wrote a business plan and put together a working demo of the software. That demo ended up looking very similar to the way Visio 1.0 looked when it shipped. Most of the critical features were in place including the ability to assemble drawings by dragging and dropping shapes from the stencil onto the page—though the only shape stencil we had was the flowchart stencil.

We had a close working relationship with Microsoft. During the latter part of 1991, Microsoft disclosed features of its upcoming Windows 3.1 upgrade. A feature of keen interest to Visio was Object Linking and Embedding (OLE). OLE would allow Visio drawings to be embedded into other OLE-compliant applications’ documents, notably Microsoft Word documents—a feature highly desired by our target customers.

OLE also defined a common automation (programming) interface between programs. Visio adopted this and, as a result, add-ons could be developed in Visual Basic that automated the Visio 1.0 engine. Drawings could be automatically created from data and information about drawings could be extracted.

The idea of an external API had been in the original Visio specification but may have been cut from 1.0 were it not for OLE’s standard definition of such APIs and its support in Windows 3.1, Microsoft Word, and Visual Basic.

1992

In early 1992 (March), we secured $3.5 million in second-round funding. Joining TVI from the first round was Kleiner Perkins Caufield and Byers (KPCB). With funding secured, we began to grow the team in all disciplines—marketing, sales, finance, development, documentation, and test.

While the development team was adding final features, improving performance, and, of course, fixing bugs, the product’s first program manager, Morgan Brown, recruited a collection of college interns. Brown assigned the interns to researching drawing types and the shape sets that would be required to support them. This research resulted in a deeper understanding of the needs of various engineering disciplines and discovered new uses such as accident scene mapping.

Not all the shape sets defined during the summer of 1992 shipped in the Visio 1.0 box. Instead, we packaged some of the shapes in separate Visio Shapes products we charged separately for or gave away as purchase incentives.

This is also when we formalized the job title of “ShapeMaster”—a development role focused on creating the shapes and shape sets to address a specific diagramming task. Olav Kvern was Visio’s first ShapeMaster and led the intern team that he and Brown had recruited.

November 1992

Visio 1.0 first came off the production line on November 2, 1992. Due to the excellent work of our sales team, customers could purchase products in software stores shortly thereafter. Egghead Software elected to buy-in a large quantity and pay their invoice on time, which was a huge boost to our young business.

Visio enjoyed excellent PR from the beginning and was fortunate to be featured by Microsoft on-stage at events due to our support of Object Linking and Embedding (OLE). Visio’s early support of this standard made it the “poster child” for OLE. Visio was often demonstrated in presentations to other third-party developers to which Microsoft was evangelizing OLE.

In just over two years, we had started a company, built a team, raised two rounds of venture funding, designed, built, tested, documented, and taken-to-market the last highly-successful commercial desktop application.

Visio 1.0 running on Windows XP (1024 x 768 screen)
Visio 1.0 running on Windows XP (1024 x 768 screen)