What Really Happened With Office and Windows 8

I was looking forward to reading the discussion of Office and Windows 8 in Steven Sinofsky’s “Hardcore Software” substack, but was pretty disappointed when it finally showed up.

There were a ton of tough discussions between Windows and Office at the time, all surrounded by a vast cloud of uncertainty. But 10+ years later, a lot of that cloud has dispersed. So I was surprised that his discussion seemed to be absent any additional insight achieved over this last decade.

This post is intended to provide a more Office-centric view of both what was going on at the time, our decision processes, our reasoning as well as learning from the past 10 years. I’ll try to be careful about “what we knew or thought at the time” from “what we learned as history played out”. Turns out hindsight is generally clearer that foresight.

Steven talks about the “platform shift” that Windows 8 represented, but in fact there was a much more significant industry-wide platform shift that was already well under way — from client-server to a cloud-mobile world.

I gave a lot of the technical and business background on this in Taking Office Cross-Platform from Inside the Windows Company as well as talking through the engineering challenges in redesigning our core engineering processes in Taking Office Agile.

To summarize, the shift to cloud was both profound and inevitable. It did not take great insight to see what the broad outlines were going to look like. The path to get there was much murkier.

In contrast to the perspective that we in Office were “overly conservative” and were hesitant to shake the boat, we saw the changes (and opportunities) happening in the industry as a once-in-a-generation shift, on a par with the shift to graphical user interfaces. It was clear that the Office business had a great opportunity to navigate this transition to cloud and mobile and emerge in an even stronger position. This would be anchored by an Office service with great native clients that were the projection of that service on to the devices our customers owned.

Although Microsoft was still trying to roll that rock up the hill, it was pretty clear at this point (and clearer and clearer every month) that we had lost the phone battle. With the launch of the iPad, Apple had defined the tablet market, but there was still a ton of uncertainty around how it would evolve and to what extent it would eat into the PC (mostly laptop) market as well as whether it would define a new market distinct from smartphones.

However it evolved, we knew the overall landscape would not look like the late 90’s and 2000’s, with Windows PCs representing 90+% of the market. There was no plausible path to a scenario where Windows dominated the market for mobile or tablet devices. We would be living in a multi-platform world, where individual customers would be accessing an Office service from multiple devices.

It made sense for Windows to try to address the tablet market by moving desktop-down (where they had a 300+m yearly run-rate of devices) vs. mobile up (where Microsoft was an also-ran). The challenge was that exciting new scenarios for tablets (which could attract app developers looking to explore new opportunities and niches) were much more about taking advantage of the characteristics of a tablet that overlap with the phone (easily mobile, light-weight, all-day battery life, mobile connectivity, touch interface, integrated sensors, camera and microphone) than they did with the desktop (large screen, mouse and keyboard). A new API for those 300m desktops or laptops did not excite app developers — new devices and new scenarios did.

Steven’s post actually captures pretty accurately how opaque and confused Windows guidance to Office was. “We want ARM versions of the desktop apps.” “But we’re not going to let anyone else build ARM desktop apps and we won’t even let your apps interact in even the most obvious ways with new ‘modern’ apps.” “We don’t want you to port the apps to the new platform.” “But it would be absolutely catastrophic if you ported the apps to other new platforms like the iPad.” “It is catastrophic that you haven’t put ‘Office’ on the new platform.” “But we want some completely new app that is Office but is not the old Office but still convinces customers and developers that old/new Office is on the new platform.” “We don’t know what that is. That’s your job. But think different.” “And it should be exclusive to the new platform even though we are just a small part of the market and there is really nothing in the new platform (if really used as a new platform and not linked to the Win32 desktop world) that actually differentiates it from the other tablets in the market which already have large market share and tons of apps.”

Queue head-banging.

This is actually a good sniff test for a strategy — if you write 10,000 words and nobody understands it, maybe it’s not really a strategy.

It was plausible that the Windows 8 strategy could help defend the low-end laptop/tablet market, e.g. in schools that were experimenting with tablets but clearly had a need for an integrated keyboard use model. (And eventually would end up adopting Chromebooks in large numbers.)

In fact, the tablet market reached a level of functional sufficiency faster than any other computing device. New iPads have faster processors, better screens, and better cameras but are functionally equivalent to that v1.

Tablets haven’t unleashed a new app world. There was no mystery magic tablet app from Office that would have saved Windows 8.

We ended up heavily investing in moving to Windows 8 APIs, but in a way that also positioned us to effectively target the broader mobile / tablet / desktop / web world. Those Win8 apps shipped where it made sense — on phones where that was the only API available (in very small numbers). On other devices where the older APIs were available and that typically were configured with keyboards and touchpad, it kept on making no sense to ship those apps. They were not a better projection of the service to those devices.

A the same time, we invested heavily in creating that Office service and moving customers there. Half a billion customers later, it looks like a pretty good strategy.

At the time, I could articulate our strategy pretty succinctly: “build an Office service and build the best projections (probably using native apps) of that service on to the devices your customers own”.

As we struggled with whether to release mobile apps and I brought these frustrating discussions home, my wife summarized it even more tightly: “give the people what they want”.

Looking back on this and other strategy discussions I was involved in, I continue to be (foolishly) surprised how often a strategy that you have trouble articulating is no strategy at all.