Last Wednesday, I started my day hearing a presentation from Ryan McKillan, Uber’s Head of Engineering in NYC and the person credited with creating the first modern Uber logo. That same night, I caught a quick drink after work with Ky Harlin, BuzzFeed’s Director of Data Science, one of the foremost experts on the science of digital creative.

Growth rates and technology roots aside, from an outsider’s perspective BuzzFeed and Uber look very different — one provides seamless access to on-demand rides, the other is a modern media company [in]famous for irresistible headlines like “27 Cats Who Really Nailed Being a Cat.” But underneath that, what’s striking listening to Ryan and Ky is how similarly — and effectively — both companies use data to fuel product adoption and deliver customer value. BuzzFeed and Uber — as well as other meteoric startups like AirBnB, Amazon, Github and Netflix — have created powerful closed-loop feedback systems. When customers use their products and take an action (for example: when you open the Uber app, tell it where you are and order a taxi), all of the data from that interaction is captured and fed back into the system to improve the next customer’s experience.

Uber's Car and Customer Location Data

To explain this more (before we get to the goods), it’s worth walking you through a quick refresher on control theory — the branch of engineering and math that looks at the behavior of feedback systems and loops.

In an open-loop system (no feedback), an input happens, and it’s processed into an output. No data is captured along the way and fed back into the system. Two practical examples are hailing a taxi (without the use of an app like Uber’s), or reading a paper newspaper.

Open Loop System Business

In the traditional taxi example, you hold up your hand signaling you want a ride (input), the taxi driver sees you — not because he knew you wanted a taxi at that place and that time but because he just happened to be driving there. You take the ride, pay your fare, and get out. Other than knowing a ride happened and you paid your fare, no data about who you are as a customer, where you were or what you were doing is saved by the taxi authority to make future taxi experiences better. The same is true when you buy a paper newspaper. You buy the paper and read it, but the publisher has no idea what articles you read, what content you enjoyed or disliked, or anything else about your interaction with their product.By comparison, a closed-loop (feedback) system like BuzzFeed or Uber takes your input action, measures your activity and behavior, tracks the output, and uses all that information to inform the rest of the system.

A Closed-Loop System Technology Startup Example

Uber knows the exact supply and demand for transportation in a given city in real-time, allowing it to optimize pricing for rides and match demand (ride requests) with supply (cars) to make the market more efficient. Similarly, unless you’re meticulous about blocking cookies, tracking pixels and your browser IP address, BuzzFeed knows exactly what you’re reading, when and how, and can use that not only to recommend you more relevant content, but also dynamically personalize your content experience and uncover larger-scale relationships between latent interest topics. Ky refers to these as “clusters,” which may show that a population’s interest in a specific celebrity actress also correlates with their interest in content about a specific cute animal.

As businesses and products become more digitally connected, the world’s best and fastest growing companies are the ones building these closed-loop systems, where software and data science are always optimizing for increasingly more customer value and organizational responsiveness. At Percolate, this theme is so important we’re not only building it throughout our product to deliver a closed-loop system to optimize your brand identity and marketing ROI, but we’ve also made it the theme of our upcoming Client Summit.

But even if you’re not a technology company (well, at least yet), and your current organization DNA isn’t well aligned to software, there are still many universal, actionable lessons on building a closed-loop growth system that can be learned from BuzzFeed and Uber. Here are 5 of them:

1. Centralize data science but decentralize data gathering and decision-making

Startups are increasingly making decisions based on expert data science, not management titles or seniority. At BuzzFeed, individual editorial teams — backed by real-time insights about things like optimal headline structure or what their audience is talking about online over different time intervals — act like autonomous product groups over their content areas.

How BuzzFeed Content Goes Viral
Mapping the global virality of a BuzzFeed article. Source: BuzzFeed. 

 

Uber, meanwhile, staffs over 40 data scientists in its San Francisco headquarters, then hires “ridiculously smart hustlers” to staff its operations teams in local markets. These ops teams get access to Uber’s central resources, technology and data but are pushed to operate autonomously, city by city, in a hub-and-spoke model to inspire creative approaches to customer acquisition and happiness.

2. Build a “closed-loop culture”

In a traditional marketing organization, when a manager wants to understand how the company blog is performing, they might pull a report from Google Analytics, or make a request to their analytics team, who responds by sending them data. At BuzzFeed, when an editor wants to understand how their stories are doing, the data science team starts by breaking down what the editor is really trying to understand, and what the right metrics and methodology are to measure the answer the business needs. Only after the root business challenge is diagnosed will an output be delivered. This process of (1) making the business’ root challenges and questions extremely transparent, (2) using lean and experimental teams to test hypotheses around those questions and (3) building the internal tools, processes and frameworks to support the system is culturally ingrained across BuzzFeed, and similarly, Uber.

3. Understand the accounting system for your industry

A lean, data-driven culture and business approach is only valuable if decision-makers ask the right questions and understand where economic value lies in their industry. For Uber, the right answer for pricing isn’t to maximize fares per trip (top line revenue), it’s to optimize for driver earnings per hour while efficiently matching supply and demand within their market.

Even Uber’s controversial dynamic surge pricing is closely underpinned by market economics with all the transaction’s participants in mind. For example, when a hotel surges its prices over a holiday weekend, it doesn’t increase the salaries of its desk staff, porters and room service providers (dynamic pricing has also allowed sites like Priceline and Hotel Tonight to flourish, since hotels stuck with extra room inventory will frequently drop prices rather than let an empty room go unused). By comparison, Uber’s dynamic pricing not only passes through more earnings to drivers during periods of demand imbalances, it also incentivizes more drivers to hit the road to balance supply and demand.

In other words, public complaints aside, Uber’s surge pricing doesn’t just make rides cost more, it makes the overall transportation market more efficient so drivers maximize their revenue and more people get rides. Uber knows exactly what operating metrics it needs to optimize for to deliver the most aggregate customer value in its market, and so all of its data science and product design efforts are focused on few transparent growth metrics.

But if your industry has a complex accounting system, you don’t always need to force simplicity where it doesn’t fit. And when you’ve truly mastered your craft or product, some of the accounting can be qualitative. As BuzzFeed puts it:

Unique visitors matter, shares matter, front page visits matter, app DAUs and MAUs matter, social media followers matter, traffic source diversity matters, time spent matters, editorial judgement matters, subjective UX, design, and brand perception matter, press pick-up and moving-the-conversation matter, scoops matter, diversity of content matters, and we are probably missing a few others. BuzzFeed is a combination of art, science, and good judgement. Understanding that balance is a competitive advantage.

4. Test and question everything, even your business model

In the past five years, technology and consumer behavior have changed radically. So does a business model from 15 years ago still make sense? Today, the Fortune 500 is turning over faster than ever.

Business Model Transition and Distruption

In 2014, KPMG surveyed 910 executives at U.S. multinationals and financial services firms, 90% of whom worked at companies with over $1 billion in annual revenue. Out of the 910 executives surveyed, 846 (93%) indicated they just completed, are planning or are in the middle of a business model transformation effort. Transition and disruption isn’t a temporary wave of enterprise turbulence: it’s the new normal.

In the personal automotive and on-demand rides market, ZipCar emerged in 2000 with a digital, short-term car-booking service that proves disruptive enough that it acquired over 750,000 users before being acquired by Avis Budget Group in 2013. Today, less than three years later, ZipCar also faces escalating competition from mobile-forward on-demand services like Uber and Lyft, as well as online travel exchanges like Expedia and P2P car rental services like RelayRides. In only 15 years the disruptor has become the disrupted.

And although Uber has gained massive traction as a car service, the company is already experimenting with entirely new lines of business, including UberRush, an on-demand courier service, UberFresh, a fast food delivery service and UberEssentials for consumer goods, partnering with companies ranging from Amazon to GE.

5. Build internal tools

Building internal tools to automate and optimize processes is near and dear to us here at Percolate. According to McKillan, Uber shares our passion. Uber’s operations teams “get way head of engineering in terms of what they’re trying to do, often with just really complex excel models,” says McKillan. As soon as operations teams find an efficiency or new way to growth the business, Uber’s engineers “fast follow” and build tools on top of their business team’s models that can be scaled across the company.

Similarly, BuzzFeed is constantly tinkering on internal tools. Everything from its CMS to its web analytics system was designed and developed in-house from the ground up to make its teams more efficient and their decision-making more accurate.

BuzzFeed Internal Tools Analytics Screenshot

One important caveat here however is internal tools need to built for the right reasons under the right circumstances. This is one of the biggest differences between outcomes of internal tools projects at technology startups vs. big companies. Startups typically launch internal tools efforts in short, agile and iterative sprints designed to give specific teams better information, or prevent duplicate or manual work. As a result, internal tool builds typically don’t tie up large swaths of budget or engineering resources, can be quickly spun up or wound down and are culturally embraced by their creators and the beneficiaries. By comparison, enterprise custom tool builds often fall victim to scope creep, slow project timelines, competing business priorities and inconsistent buy-in across executives, teams, departments and regions. If you work at — or lead — a big company, don’t fall victim to these mistakes: start small with the right team, or consider buying or partnering rather than building.

Overall, it’s easy to dismiss BuzzFeed or Uber as anomalies — rare examples of companies who unlocked or stumbled upon hyper-growth models in prime markets for venture capital funding (Silicon Valley, Silicon Alley) to fuel more breakneck expansion. Statistically, they are. But whether it’s BuzzFeed or Uber, or Amazon or AirBnB, it’s not a coincidence that the world’s most valuable and fastest growing new companies all share very similar values, cultures and corporate DNA, or are all built on top of similarly self-reinforcing, closed-loop software systems.

If you don’t work at a BuzzFeed or an Uber — again, just by the numbers, most of us don’t — but find yourself envious of or inspired by their success, remember: business model innovation isn’t just about radical industry reinvention — digital transformations can start with modest, pragmatic and incremental beginnings. In many ways, that’s exactly where startups like BuzzFeed and Uber first came from: they started out small, refined their respective closed-loop growth systems, then identified the inflection points, people and technology to make it scale.

We’ll be talking more about the growth potential of technology systems with partners like GE, Ogilvy and Pinterest at our upcoming Client Summit on February 5th in NYC at the Times Center. Click here to request an invite if you’d like to join the conversation.