Should Your Startup Go Full Stack?

I’ve written about the problems associated with selling into value chains where there are disparate P&Ls. In this post, I’ll dive into the solution to this problem: going full stack. I’ll also cover when going full stack makes sense.

Chris Dixon of a16z has noted the rise of a number of high-profile, full stack startups across many industry verticals. What is a full stack startup? Full stack startups are those that, rather than selling a novel piece of technology to incumbents, use the technology they’ve built to serve the incumbent’s customers directly, bypassing the incumbent in the process. There are many of examples of modern full stack startups and their respective antitheses. A few examples:

Uber vs Hailo - Everyone knows that Uber bypassed taxis and instead when straight to the end consumer. On the other hand, Hailo responded to Uber by selling an Uber-like platform to local taxi companies. Hailo has struggled while Uber has run away with the market.

Buzzfeed vs ThoughtLeadr - BuzzFeed is changing the media business because their entire strategy is based on sourcing and producing content that’s designed to be easily consumed through the best discovery channels: social media. Everything they do stems from a deep understanding of how users share content. As a result of that, they have devised an incredible formula that seamlessly intertwines native advertising with socially shared digital content. ThoughtLeadr and many companies like it are trying to empower media organizations to rethink content strategy and distribution around native advertising and social sharing. BuzzFeed does the tech and content production in house. To be clear, I am close with the team at ThoughtLeadr, and I actually love what they’re doing. BuzzFeed can’t be the only media property on the planet. The world needs companies that can help old-school media properties thrive in the digital age. ThoughtLeadr is doing just that.

Oscar vs traditional healthcare insurers - Oscar recognized that health insurance should be completely rethought around smartphones. Smartphones enable new models where insurers consume vastly more consumer data (think movement, etc), and present a potent communication medium. Rather than trying to help traditional health insurers transform their entire businesses - customer acquisition, service, preventative care, etc - Oscar decided to build a brand new health insurance company that didn’t have any legacy baggage.

HealthSpot vs Teladoc - Teladoc owns and operates its own technology to enable virtual physician consults, whereas HealthSpot built/assembled technology and sold it to provider organizations. Healthspot recently ceased operations, while Teladoc recently IPO’d and is growing hand-over-fist as healthcare continues to consumerize.

One Medical Group (OMG) and Kaiser Permanente (KP) vs traditional healthcare providers - KP and OMG are two prominent examples of “full stack” healthcare providers. KP is enormous and has been around for a while and operates as an health plan and fully-featured provider organization. OMG is the leading pioneer of the Direct Primary Care model, which borrows many health-plan like features and functions. These organizations both invest heavily in their own technology development, and use that to offer differentiated healthcare services to their customers. KP and OMG stand in stark contrast to more traditional healthcare providers, who simply make money for doing stuff to patients.

I don’t mean for these examples to paint the picture that being full stack is always the right answer, and that being partial stack is the wrong answer. There are many successful companies that are explicitly not full stack: Workday, SalesForce, Slack, Zenefits, etc. These startups have been successful partial stack companies because they do not break the daily workflow of their users.

For example, let's look at SalesForce. Although sales professionals like to complain that SalesForce is clunky, slow and a not helpful in their daily workflow, SalesForce is just a tool that’s designed to reinforce a process that the sales people should have been adhering to in a paper world: contact and pipeline management. Zenefits users are still buying health insurance, and still filling out forms. The process itself really hasn’t changed; all Zenefits has done is unemploy health insurance brokers. Slack users are still designing, coding, etc; with Slack, they can collaborate with their colleagues more effectively. The same can be said of the other examples above.

On the other hand, tech companies should consider going full stack if they break the operational processes of the primary users of their solution. Uber could never sell into the taxi industry because Uber commoditized taxis, and taxi company owners didn’t want to subject themselves to that. Oscar could never sell its novel technology to traditional health insurers because it would have required rethinking every function in the insurance company. HealthSpot couldn’t usher traditional healthcare providers into a retail care delivery model.

Why can’t incumbents adopt technology solutions that completely break the daily workflow of their primary revenue-generating employees? In short, it’s too operationally disruptive. The organizations simply cannot absorb the change. When an employee’s daily workflow is completely rethought, the vast majority of her existing operational knowledge is not only worthless, but actually destructive in the new paradigm. Processes that made sense in the old paradigm no longer make sense in the new paradigm. Convincing thousands of employees and managers across disparate geographies and changing incentive structures to unlearn bad habits (to be fair, they were good habits in the old paradigm) is nearly impossible. In the case of Oscar, this is even more problematic as the changes would have spanned employees across every function in the company. Old and new paradigms simply cannot coexist in the same organization. Note: this is the fundamental problem that Clayton Christensen details in The Innovator's Dilemma.

As we see more verticals SaaS-ify, I suspect most startups will rightfully decide not to go full stack. For example, TalkDesk, ServiceMax, and Veeva rightfully didn't go full stack.  Although the new wave of vertical-SaaS vendors strive to be the all important system of record, they are generally seeking to re-inforce processes that should have already existed. In most cases, SaaS solutions are not radically changing the cost structure associated with delivering the services in question and are not re-inventing the business.

In summary, startups should go full stack if they are going to genuinely break the workflow of their target user. Otherwise, they should go partial stack.

PS, Given my background in electronic medical records (EMR), I'll add some commentary to that industry. EMRs break the daily workflow of the primary revenue generating employees of healthcare provider organizations. So why haven't EMR vendors gone full stack? Not a single traditional EMR vendor has even tried to offer healthcare services. There are a few reasons: during the golden era of EMRs, telemedicine wasn't reimbursable, so it wasn't practical to consolidate providers centrally given the intrinsically local nature of healthcare delivery. On the other hand, every novel tech-enabled healthcare service builds and manages its own technology in house: One Medical Group, Doctor on Demand, Teladoc, etc. The providers that are pioneering new delivery models recognize the importance of technology in their new models and are rightfully insourcing it to go full stack. This tech investment also serves as their barrier to entry and differentiator.