Microsoft Office: Existential Choices In the Face of Platform Shifts

While building a business and also trying to understand when something is a platform shift or just system noise, one has existential choices to make. This is a story of how often Microsoft Office faced such decisions.

The Kobayashi Maru screen in ST:WoK detailing a “no win” competitive situation faced by Savik.

1/ It’s a toy. It’s the next big thing. Tech is magical. Tech is trivial. When talking about “blockchain” it is easy to debate in abstract.

But what about when you’re living a potential technology transition/disruption.

A story about Microsoft Office.

It seems as though everyone has an absolute strong opinion on blockchain, based on doing any sort of analogy to the past. If one suggests blockchain is as big as the internet itself then clearly, according to many, that person understands neither blockchain nor the internet.

By sharing this story about the platform shifts that have been going on for 25 years, constantly a potential “threat” to Microsoft Office, I hope you can see that one could make strong cases for how technologies like blockchain (though perhaps not literally) can be present during platform shifts and then the actual disruption takes on a very different “shape”.

That’s because with established markets, especially big ones, the “next big thing” is never a new version of the old thing built using new technology. You have to consider how all the plates are spinning in the air and anything can change.

By 2000 Office enjoyed a solid 8 years of explosive growth reaching $10B in revenue that year. Product “pivoted” from individual to enterprise purchasing and IT deployment. It seemed extremely secure.

Many don’t remember but Microsoft Office was a consumer product. The whole GTM was about retail, mass market, individuals since that is what the software market was. In the mid-late 1990s Office (and Microsoft with Windows NT and Exchange) made what could be called a massive pivot to the Enterprise and building out a manageable account sales model.

Competition from prior gen (Lotus/IBM, Borland, WordPerfect) still “existed”. New competition was “Open” Office which was an open source copy. We were confident in execution abilities that lesser/free product wouldn’t pose material threat.

Office had grown up as individual applications each taking on a laundry list of category or “head-on” competitors. In the battle of suites none of the major competitors had a full competitive suite (word processor, spreadsheet, graphics, database, and then mail). Given that history, the prevailing view on the team was that a head-on competitor even an open-source project posed a minimal competitive risk. Nevertheless we were totally obsessed with OpenOffice as they were with us. I recall one trade show where I was watching a demo at a booth and noticed that their “PowerPoint” menu rendering was different than Office (and than their Excel and Word). When I pointed this out they were shocked and promised a “fix” by later in the afternoon. I left worried even less given their view of cloning Office.

It was still there though. Really new were the first of two different technology approaches to compete.

(1) Office written in Java (also potentially open source, WORE) [write once, run everywhere or WORA run anywhere]

(2) Office for the browser

Try not to use hindsight in thinking about what to do.

It is very difficult to learn these lessons from the past when you know the outcome. This is the challenge of learning from case studies as they team in Business School. But the case method is not about the facts of specific technology, but how managers and leaders go about deciding what to do. So we should focus on that.

Speaking of business school, during the timeframe of this thread I was on sabbatical teaching at HBS (I did not go there, but studied CS in grad school). The students were literally a direct reflection of “Office is dead”. It was stunning. I was constantly asked why I was still working at Microsoft.

Also talk about Business School timing, in 1998 when I was on sabbatical “Innovator’s Dilemma” was brand new and so all the talk in the HBS corridor was about disruption. It is why the students and all the faculty were so convinced Office was the first and biggest potential roadkill. Being a few doors down from Clay made for an exciting albeit defensive semester.

Java Office wasn’t one product as much as a movement. Many were “applets” (Office was apps, applets were the lean/mean on demand) “replacing” Office. Componentization was movement. Many in Microsoft were certain a component architecture was future, especially for enterprise apps.

The technology appeal/enthusiasm of composing workflows and apps and delivering this code in bits and pieces (across any OS) was very high because the dislike of Office was its monolithic nature and “bloatware”. Just running SETUP was massively painful.

BUT we had tried to build components with “OLE” and tried the whole “edit a document made up of parts” approach it was a total failure. Magically flipping menus, editing in little rectangles…ugh.

This whole “components” movement was very much a technology first. It was almost an outgrowth of “object-oriented” which I had just spent the first 5 years of my career “dealing with” working on C++ (which was new to the world at the time). So many IT professionals though the “cure” for the bloat and management difficulties of Office would be to use components…because components are by definition easier or something like that.

It is important to recognize that the IT world was new to Microsoft and the big things going on at the time was “reduce total cost of ownership” which was basically caused by Windows + Office and also the notion that IT should be building solutions. Solutions were built on components as reusable pieces and enabled much more rapid and flexible applications [sic].

So components had a two-pronged attack on status quo. And it just so happened that Java was the cool new object-oriented programming language. It didn’t matter that many people had already come to realize object-oriented didn’t really change much of what got done, other than make things slower and introduce complexity in coding.

This is a screen shot of an Excel component inside a Word document. The basic idea is you would not need to switch apps/windows to edit since the goal was a fancy printed document anyway. Just click on the table and the menus and toolbars would change to “Excel” and you can edit spreadsheety things.

SO should we be scared that the next time would be right and the other folks would figure this out and destroy us. Or would this be yet another false signal for the future of everything?

The stakes were pretty high!

Components like the above example were a total failure. They stressed the PC, were fragile, and generally clunky for end-users. They made great demos of a “Suite” but that was about it. Also they were enormously complex to build and test, which was super frustrating for teams still focused on the legacy competitors in categories.

BTW, these Java Applets were usually like 10X the resource usage that Office was and offered 1/100th the capability. These things hardly worked. It was extremely difficult to make these out to be credible alternative/threats.

You can easily see the technical buzzsaw. (Familiar?)

OMG these Java applets were awful. Java was slow, HUGE in memory, and totally flakey. There’s some rich history here for Office in particular in that the first Windows apps were written using an interpreted language (C-like code) and the team had basically spent a couple of product cycles getting rid of all the interpreted code. Now Java comes along and tells the world “no this is the way”. It seemed nuts to us. @jondevaan was especially articulate about this since he had spent so much time on the interpreters and trying to make Excel fast. For me personally, the fact that Java was garbage collected also gave ample evidence that this was nuts, having spent years defending the reality that C++ did not have GC even though real® object-oriented programmers used GC.

Thus you can see some analogy to both WWW at the time and Blockchain today how these new solutions often perform terribly for things that you know how to make fast. WWW was a slow to render document format that didn’t do very much over a protocol that used ASCII for G_D’s sake.

That hardly stopped the constant stream of IT, developers, internal champs (.net!) saying components/rewriting Office this new way was the future. It was very stressful to battle this constantly.

Reminder, this had been tried before BY US so clearly we must have been early!

It is hard to overstate how many of the strategic thinkers and industry pundits were pushing hard on an Office rewrite. It was clear that Office was over and they had seen the future.

But components, an interpreter, and garbage collection was a trifecta of stuff that seemed like such a crazy place to start.

CONCURRENTLY, the Exchange mail team had pioneeed DHTML/AJAX/XMLHttpRequest and had taken browser based email to a new level (later used in gmail, google maps) opening another front in reinventing Office aimed at Outlook. (Note the basic/premium mode — that’s opt-in DHTML)

Then along came Internet Explorer and Dynamic HTML (DHTML) in combination with XMLHttpRequest. This seemed like magic. It also had the bonus that for the time being it only worked in Internet Explorer which the IE team loved.

This was a whole other competitive challenge and an architectural approach that was aimed at Office.

So imagine that was a whole other set of meetings. For Microsoft watchers, the .net people were focused on Java and the IE people were focused on HTML and both wanted Office to be rewritten in these new platforms. Imagine all the meetings I went to.

In the internet era, Outlook was the new anchor of Office and so risk there was bigly risk.

But again faced w/subset of features it seemed remote and Outlook was only 3 yrs old!

Though this time the effort was one team versus another. Not org, but tech strategy/architecture.

Outlook was brand new. It almost didn’t work. The reviews in 97 were lukewarm at best (please don’t make me Google Walt Mossberg’s review!). Outlook 98 quickly added internet protocols thus introducing the dual product schism that haunted Microsoft email clients for a decade. Then Outlook 2000 brought the two “modes” together. And so on. In other words, we had work to do just to make *Windows* email work.

But the Exchange team having pioneered/invented AJAX had developed the perfect API for solving problems Outlook had on Windows which was just doing requests to the Exchange server and showing lots of rows of email. It made Outlook look slow! At the same time Outlook had implemented a bunch of features on the client and so we had client/web asymmetry which was exactly what customers didn’t want. That was a lot of ammo for why the architecture for building mail was wrong.

Outside, there were many efforts at building browser based authoring tools using Ajax.

It was really frustrating to the IE team that the Office team was not doing AJAX Office apps. So naturally IE was evangelizing all our “competitors”. That’s how platforms work! NBD.

OWA as outlook web access was called was great for lightweight email but not for powering through email and calendaring like most did. It was also a great IE demo. So Yahoo, Hotmail, AOL and more were heavily evangelized to use DHTML and many cool demos started showing up including people building Office-like things. PowerPoint (or drawing) was a target in particular.

Office supported HTML as a file format but not as the editing runtime. Tons of work went into collaborative servers working with Office. See edition.cnn.com/TECH/computing…, example.

Office was facing 3 competitive disruptions:

free

client-side Java

browser/HTML

For each one was it:

a) next major platform shift

b) totally wrong idea

c) potentially right idea too soon or wrong implementation

🤔🤔🤔

Meanwhile we‘re a $10 billion biz.

Startups springing up around Java and AJAX. Mostly taking same approach of copying Office directly. Certainly that is how we saw them (bias) and so not risky.

The company-wide concern over Office “franchise” was intense. Lots of “help”. That’s how big companies work.

Imagine you’re working on a business that just went from $1B to $10B in 10 years having trounced a dozen competitors in each category and won on both Mac and Windows (and OS/2). Now seemingly out of nowhere the internet hits and simultaneously you’re facing deep architectural competition from Java as well as HTML and also structural competition from Open Source. And all of them were aimed directly at your weak spot of cost of ownership, complexity, and bloat. It was stressful.

I kept lists of all the new products that did “Office” in those technologies. There were giant tables in all of our planning docs and product planning was researching dozens at any given time. It seemed like every trade show (that’s where we learned back then) report was yet another browser-based or Java Office thingy.

Meanwhile the internal demand for a response was very high, from technologists. The sales people wanted none of that unless it was what they were already selling but with the objections (bloat, TCO) removed.

Definitely most difficult time for Office: we had become established and business and now faced tons of new/potential risk.

BUT none of the technologies were being used for anything “real” except Outlook. Google Maps/Gmail were years away.

Oracle was also out there w/“Network Computer”. Office viruses/malware undermine quality. Bloatware. WWW was HUGE.

Everywhere was the next fundamental architecture.

Can you get a picture of how fragile Office could seem?

How today blockchain might look to institutions?

Meanwhile none of these new things worked. I mean literally they didn’t work. Putting aside the big one OpenOffice complained about (file format compatibility) the basics of editing was super clunky. Performance was awful. PLUS most people had poor connectivity and every one of these browser-centric or online tools had to figure out offline. There was no there there.

EXCEPT every pundit and customer was focused on the future demanding answers.

BET: We had already bet big on server-side collaboration (SharePoint).

Needed a bet on core value of Office. Quietly started our own HTML clients — “companions” to Office.

Basically classic idea: if something was to replace Office we would build it ourselves.

Our answer was to focus on what we believed we could make work.

We had made a bet on HTML as a rendering format (editing would come later) in Word and PowerPoint very early. Trivia: Word’s Internet Assistant for HTML was released before Netscape 1.0. Saving a slide deck as a series of JPEGs with navigation buttons was a hallmark of early DIY and corporate web sites. PowerPoint had built this add-in super early and it was quite popular.

We had acquired Vermeer FrontPage in 1998 and were working super hard to bring the notion of a “living web” of Office documents to the workplace. The first iteration of this was called “Office Web Server” and that morphed into FrontPage broadly. OWS as we called it is the “team site” portion of SharePoint or as I like to call it “a place to put files”.

But the big new bets was to build web version of Word, Excel, PowerPoint — native HTML applications that ran in a browser and enabled editing of real Office docs within OWS/SharePoint. We knew we could not build all of Office because we had just been watching everyone fail and we also had to balance the existing business.

It was super difficult because most HTML apps were “free” and so even making these was viewed as risky. Was this “free Office”? Sales didn’t want these at all!

But from a product perspective the investment was being made and the timing right. But that’s not enough.

Marketing and the sales force were literally terrified of “web Office”. So many customers have been “won” over in competitive battles with all these new fangled competitors the last thing the field wanted to was to go back and revisit all those deals [these deals were the start of multi-year enterprise agreements, today we call this SaaS but back then it was “maintenance”]. They assumed that if Microsoft did web office it would also be free. It would certainly cost less than “full” desktop Office, simply because it did less (you know because code is priced by the line).

There’s a whole backstory on how you can build a product and then not have a way to bring it to market. That’s the other part of innovator’s dilemma. The first part is a product capability and the second part is pricing/distributing it.

Gmail came out. Google acquired Writely, 2web. Built successors of those into gmail. Basically squeezing Office docs out of Google webmail flow. It took 6+ years to see the scenario come together.

Still not losing any Office sales to these.

Then gmail came out and subsequently had integrated the early versions of what became Google Apps. We’re now in 2006-ish.

Keep in mind Google Apps was in beta until late 2009 and not even revenue generating (directly) and it would be until present before significant enterprise effort.

For those wondering if we had any make v buy decisions during this time…of course we were constantly bombarded with ideas. But from a buy perspective none of these worked and to be honest the technical buzzsaw precluded that. When Google acquired the web tools we were puzzled more than worried, but also know the history that the CEO long had a business strategy of just doing things to distract and/or disable Microsoft (Open Office!). So we were on our toes. As a practical matter there was no way to acquire anything Java and the browser based tools would not have integrated with the Office Web Server. So classic integration challenges for the incumbent.

Next step was to build the flow of the Office Web Apps into browser based mail.

But still deep concerns over customers expecting “parity” with existing Office and also the economics (pricing).

In times of platform shifts, product needs to move even if business can’t/won’t.

Many of us had moved from places like IE and Office and were now occupying jobs in Windows (and “Windows Live”) land. So naturally an early thing we did was take the Office Web Apps and connect them to Hotmail and OneDrive (Skydrive) to compete with gmail (at the time Hotmail was battling to overcome/deal with the huge/free mailbox).

The fears of free Office were immense. Even the existence of the web apps was viewed as potentially nasty pricing pressure on Office. Clearly the product team had not through through the sales & marketing challenges was conventional wisdom.

So today Office is probably a $35B business. Google Docs is big but not even close to 10%. Office Web Apps are great but still vast majority of people experience productivity through desktop Office.

What’s the lesson from this? Pretty tricky. Easy to see two sides.

Also during this time the other bet was being made in Office which was to do “hosted Exchange”. That’s a whole other part of this story with a whole similar sets of dynamics and history of “must have” and “can’t have” along with “won’t work”.

All the while Google Docs (now G-Suite) took things in a different direction, whether by design or out of competitive necessity because head-on wasn’t working. Focusing on real-time collaboration, de-emphasizing direct Office compete, and building different features combined with massive “free” distribution brought a lot of people to the product. But just not a lot of money. Google does not report G-Suite revenue but it is part of the $4B or so other revenue in Google proper which includes all the Cloud, Hardware, and more. So I just guessed.

The platform shift has happened. It wasn’t an Office clone, java, or even browser based clones.

It happened differently with the shift to mobile, apps, messaging, and real-time collaboration.

The nature of work is changing and tools are both leading and following.

Ultimately the platform shift did not come from Java or even editing/authoring tools in the browser (and not really DHTML/Ajax). The disruption came about because of many moving parts.

In particular the rise of mobile form factors along with messaging apps altered the landscape of what productivity means. This is happening now but has been going on for some time.

Here’s a post from me about the changes in the nature of work — Continuous Productivity: new tools and a new way of working for a new era.

The key is relative to what started all this, blockchain, that the disruption happened at a different level of abstraction that the conventional wisdom would have said. It happened at a higher level “app” not at the pure technology level. That’s because the new generation of tool-makers comes along and assumes a whole bunch about the tech stack and just builds, but does not view the stack as the differentiator. It is what can be done with the stack.

When major shifts happen, everything moves not just specific technologies.

WWW did not just change protocols used for client/server or formats for documents. *Everything* changed, just not at once.

The future is here, just not evenly distributed.

— Gibson // END

Note: Updated post with some minor edits.