Today, we’re going to talk about one of the most insightful quotes in the history of the Internet:

“There’s only two ways I know of to make money: bundling and unbundling.” –Jim Barksdale, CEO of Netscape

In the history of the internet, many (really, all) of the big leaps forward in how we use and interact with the internet have been instances of either bundling services at a particular layer of the stack, or unbundling them from that layer. For a primer on bundling and unbundling, and how important this is, please listen to this podcast as well as read everything written by Benedict Evans of a16z on the subject.

To give a few examples of why bundling matters, I’ll offer my own condensed version of the history of innovation and the Internet. In the beginning, the net was a scattered, disorganized place with no easy way to find or reach anything. Then companies like AOL came along, and bundled the internet into easily-accessible portals whereby topics and subjects of interest could be found via indexers like AOL Keywords (remember those?). Meanwhile, Netscape was busy unbundling the web from all the different dedicated portals, rebundling them into a different layer: the web browser. For a while, this worked ok- we found websites using lists like the Yahoo directory- but that didn’t scale very well. So Google came along and unbundled page discovery from the directory list, rebundling it into a smart search engine built on PageRank.

The web chugged along nicely for a while; at some point, the rise of social media (Facebook, Twitter et al) unbundled content discovery (or at least, parts of it) from the search engine and rebundled them into social graphs. Then smart phones came along, and turned everything upside down by unbundling the web browser into many distinct apps, and then bundling all of your various social graphs into a single contact list on your phone, accessible by all of your apps. Still following?

Part of why tech visionaries like Larry and Sergei are so misunderstood in the beginning yet hailed as geniuses later is that it takes a lot of vision (and a bit of craziness) to imagine a new way to bundle or unbundle something. If you want to really improve something, don’t make a new version of what already exists: find a new way unbundle what people want from that layer, and then bundle it into a different layer of the stack that makes more sense.

For example: when I want directions to a restaurant, I used to use Google Maps running on a web browser- not because the web browser is an especially good tool to do this, but because it was the only place you could get Google Maps. These days, I hardly use Google Maps in a web browser any more; it’s on a dedicated app in my phone, which goes with me everywhere. Truth be told, I don’t even ask directions through the Google Maps app much either; I simply ask Google Now, which is smart enough to know that information about directions happens to be found in the maps app, but skips the step of taking me directly to that app: it simply bundles that information together with everything else I need to know.

So, just to clarify, the function that I wanted- finding directions- got unbundled from the web browser into a distinct app, then bundled into the discovery function of Google Now. As a result, the process of getting directions became easier: not because someone improved Google Maps, but because it was moved into a layer of the stack that made more sense. Questions about information you don’t know should be directed towards layers of the stack that are good at helping you discover things. If you want to improve website discovery in the 1990s, don’t make a better version of the Yahoo Directory; build PageRank. If you want to improve the way news reached us in 2008, don’t make another news website; build Twitter. As Peter Thiel put it, “Every moment in business happens only once. The next Bill Gates will not build an operating system. The next Larry Page or Sergey Brin won’t make a search engine. And the next Mark Zuckerberg won’t make a social network. If you are copying these guys, you aren’t learning from them.” If you want to improve a service incrementally, build a competitor at the same level of the stack. If you want to improve a service disruptively, unbundle that service from its current layer of the stack and build it at a more intuitive level.

So what can health care innovators learn from this? I believe that the important lesson here is that the big breakthroughs in health care innovation will come as a result of unbundling and rebundling of specific products and services. And just like on the internet, the way to see this new innovation isn’t by looking at the old layer of the stack where that service used to be bundled- it’s by finding the brand new products and services that could not have existed before.

The best example I can give of this are electronic health records (EHRs). People have been clamoring for EHRs for decades, and after numerous starts and stops they are finally arriving in mainstream hospitals, clinics and care networks. They should be a no-brainer in the digital age, after all: how are paper records of patient health still a good idea, in 2014? And yet, if you look around a hospital, you won’t necessarily find that EHRs have improved day-to-day operations. In many cases, they make administrative and logistical problems worse. Doctors, staff, and hospital administration love to complain about them, and if you ask around you’ll get a broad sense that EHRs have been a complete failure relative to expectations.

But EHRs have not been a failure. They’ve been a great success, just not at the place where you were looking. What EHRs have done is to enable the creation of convenient, distributed, remote offices like Minute Clinic where patients with long-term, chronic illnesses can get quick checkups for a fraction of the cost of visiting a traditional doctor’s office or hospital. By unbundling patient health records from the point-of-care level of the stack (the doctor’s office, or hospital) and rebundling them at the provider level, in the cloud, accessible from anywhere, EHRs allow a whole new wave of cheap, convenient, and effective treatment options to open up for patients who don’t need a whole hospital at their disposal- they just need the kind of quick checkup that Minute Clinic can provide.

Bundling and unbundling innovation in health care isn’t limited to huge initiatives, either. Small changes in emergency room triage, such as sorting doctors to see patients, rather than the other way around- have made big differences in wait times. Why? By shifting initial triage (or in internet terms, ‘symptom discovery’) to a different level of the stack (from triage nurse level up to the ER doctor level), symptom discovery can happen just once, and more quickly. It’s a bit analogous to the jump from discovering a web site of interest using Yahoo Directory list, to using Google powered by PageRank. With Yahoo Directory, you had to sift through every web site out there to find the one you wanted. With PageRank, you simply tell it what you’re looking for, and that website comes to you.

There are two lessons here. The first is, to improve a health service don’t try to immediately build an incrementally better version of that service at the same stack- first try to brainstorm ways of how you could move the essential elements of that service up or down into a different layer of the stack that makes sense. The second is, when you’re evaluating whether or not a specific innovation has been helpful or not, don’t look at the old level of the stack and compare before/after, the way we look at hospital patient records before/after the arrival of EHRs and conclude that they haven’t helped. Humans love to A/B test, because it leads to easy comparison and evaluation. Resist that temptation: instead, look for what new service they’ve helped create at a different level of the stack. You may be surprised at how much innovation you’ll see through that lens.