Amazon: R&D as a Culture

Amazon’s most visible efforts are found in Lab126, an R&D subsidiary based in Sunnyvale, California. This is the branch responsible for all of the hardware products tied to Amazon — the Kindle, Echo, and the [ashamed] Fire Phone. The lab continues to push out devices, launching the Echo Show, Echo Spot, and redesigning the Fire TV and original Echo — all within a matter of months.

According to Dan Gallagher, reporter for The Wall Street Journal, Amazon spent over $5.5 billion in the second quarter of 2017, which was a year-over-year increase of over 40 percent, and the firm shows no signs of slowing its R&D expenditures.

Amazon’s profitability has benefited from its growing cloud business, which has provided some buffer against the R&D spending that does not provide breakthrough products. But Amazon persists in R&D even in markets where it is already beating its rivals.

Amazon’s Echo dominates the voice-activated speaker market yet the company has issued a newer version that is priced 45 percent less than its predecessor, a move expected to increase the market value of Echo five-fold by 2026. And, referring to the Alexa platform, Amazon CEO Jeff Bezos said: “expect us to double down.”

To better understand the origin of Amazon’s R&D strategy, we can look at Bezos’ 2010 Letter to Shareholders. Bezos explains why Amazon aggressively invests in technology, and it goes beyond Lab126:

“All the effort we put into technology might not matter that much if we kept technology off to the side in some sort of R&D department, but we don’t take that approach. Technology infuses all of our teams, all of our processes, our decision-making, and our approach to innovation in each of our businesses. It is deeply integrated into everything we do.” Bezos adds, “these techniques are not idly pursued – they lead directly to free cash flow.”

This quote explains the importance of R&D activities while at the same time highlighting aspects of Amazon’s culture: innovation and growth are built into every segment of the company, not just R&D.

Amazon’s R&D strategy suggests many best practices. Here are two:

Google X: Searching for Radical Breakthroughs

R&D, rather than siloed off to a separate corporate arm or department with a separate budget, is often tied throughout a company’s operations, as is evident in Bezos’ letter to shareholders. At the same time, R&D can also be responsible for many of these companies most ambitious (and sometimes bizarre) projects that are not part of a firm’s core competencies.

Google has a sandbox meant solely for these types of projects — “X.” X, aptly subtitled, “The Moonshot Factory,” focuses on radical projects that are 10 times better than any existing solution. Why? According to Astro Teller, director of Google X, a 10 times improvement is often easier than a 10 percent improvement.

Teller explains that when looking for a 10 percent improvement, a company is tied to existing technologies and solutions, but when shooting for 10 times, anything goes. Moonshots are at the intersection between a huge problem, a radical solution, and a breakthrough technology.

The most visible effort from Google X is its autonomous car project, spun out as an Alphabet Subsidiary called Waymo. The Waymo moonshot is framed as follows:

Problem: Close to 1.25 million people die on roads every year globally, and 94 percent of those accidents are caused by human error.

Radical Solution: What if cars could drive people safely from point A to point B at the push of a button — without needing a human to take over driving at any time?

Breakthrough Technology: Vehicles could have built-in sensors to detect pedestrians, cyclists, vehicles, road work and more from a distance of up to two football fields away in all directions. Smart software could predict the behavior of objects and road users to help the vehicle navigate the vehicle safely through everyday traffic.

— Source: Waymo, 2017

Today, Waymo is the frontrunner in the autonomous car space. Lucinda Shen, reporter for Fortune, writes that analysts from Morgan Stanley have estimated the firm’s value at or near $70 billion dollars. Waymo struck a partnership in 2017 with Lyft, which has the second biggest volume in U.S. ride service. Further, Shen reports that Waymo and Honda are close to finalizing a deal to build automated delivery vehicles, which is a significant commercialization push from a giant that generates around 90 percent of its revenue from advertising. But this success doesn’t come without struggles.

Google X kills projects — a lot of them. For every Waymo or Verily (Google X’s life science spinoff), there are countless projects that don’t make the cut. “We killed over 100 investigations last year alone,” Teller states in a 2017 Ted talk.

The organization moves quickly, fails a lot, and iterates based on feedback. Teller describes the thought process behind the killing of their buoyant cargo ship project:

“However cheap they would be in volume, though, we found out it was likely to cost close to $200 million for the R&D and materials to design and construct the first one. Since X is structured around tight feedback loops of making mistakes, learning, and new designs, $200 million is way too expensive for us to get the first data point on whether we’re on the right track.”

— Astro Teller

This cost-benefit analysis must run through every decision, from the mundane to the risky. Both organizations, Google and Amazon, exemplify a mastery over the R&D portfolio, allocating resources to both incremental and radical advancements.

The Innovation Ambition Matrix

The matrix helps managers to assess ongoing R&D initiatives, the number of initiatives and the level of investment. It provides a way to shape the overall ambition for the company’s innovation, whether it be short or long-term small projects that complement existing products, or high-risk efforts that might offer a breakthrough or a bigger payoff.

But there’s an additional problem in addition to the difficulties in valuing transformational R&D strategies.Nagji and Tuff found that, because chief executives often do not have a strong grasp of the myriad initiatives that might already be underway in their organizations, R&D can be haphazard and episodic.

According to Nagji and Tuff, “the companies that have a strong track record of creative endeavors articulate a clear R&D ambition, balance core and transformational ventures, and have the infrastructure, tools, and capabilities to manage and integrate initiatives.”

Determining R&D Focus

Beyond the core, adjacent, and transformational lens introduced above, firms must decide on where to focus their R&D efforts in relation to their product offerings. For the majority of companies, the focus has shifted towards software. From Strategy&’s, “Software as a Catalyst” report:

“Most of the world’s major innovators are in the midst of the same transformational journey. R&D is shifting more and more toward developing software and services. Software increasingly carries the burden of enabling product differentiation and adaptability, and enhancing customer experiences and outcomes.

This transition is for a good reason — it’s the same tactic that has skyrocketed tech giants ahead of the competition. According to Ben Thompson of Stratechery, “while digital infrastructure obviously needs to be maintained, by-and-large the investment reaps dividends far longer than the purchase of any physical good.”

— Strategy&

Thompson goes on to provide examples of the strides made due to tech giants’ R&D:

It was expensive to develop mainframes, but IBM could reuse the expertise to build them and most importantly the software needed to run them; every new mainframe was more profitable than the last.

It was expensive to develop Windows, but Microsoft could reuse the software on all computers; every new computer sold was pure profit.

It was expensive to build Google, but search can be extended to anyone with an Internet connection; every new user was an opportunity to show more ads.

It was expensive to develop iOS, but the software can be used on billions of iPhones, every one of which generates tremendous profit.

It was expensive to build Facebook, but the network can scale to two billion people and counting, all of which can be shown ads.

Aside from a general preference for software, metrics can help guide firms efforts and spending when creating a research and development portfolio.

— Ben Thompson, Amazon Go and the Future, 2018

Where Companies Fail in Selecting an Optimal R&D Portfolio

R&D, like any investment, should maximize returns over time. All too often, governance, stakeholder, and consumer pressures mean that short-term developments are prioritized over radical change that requires a longer time horizon. A study by Rachelle Sampson and Yuan Shi of the University of Maryland found a correlation between failure to invest in the long term and a decrease in a firm’s innovativeness.

But the issue isn’t as simple as telling executives to focus more on the future. There are many other factors that come into play when allocating funding to projects. A study by researchers Paola Criscuolo, Linus Dahlander, Thorsten Grohsjean, and Ammon Salter published in the Academy of Management journal found that many R&D funding decisions were undermined by simple human bias:

“The Biases That Keep Good R&D Projects from Getting Funded (…) R&D selection is not always based on objective estimates of a project’s costs and returns. We found that funding decisions in this organization were influenced by how novel a project seemed. Proposals that were deemed either too novel or not novel enough got little or no funding.”

— Criscuolo, Dahlander, Grohsjean and Salter

The illustration above developed by Criscuolo, Dahlander, Grohsjean, and Salter illustrates how R&D selection is not always conducted scientifically. How can this problem be overcome, and a best practice instituted for portfolio selection? Metrics can serve as the guide.

R&D Portfolios

A portfolio for R&D should reflect a company’s position regarding risk and reward. According to Nagji and Tuff, that means building a portfolio that produces the highest overall return and that meets its appetite for risk.

In Managing Your Innovation Portfolio, Bansi Nagji and Geoff Tuff state that a combination of noneconomic and internal metrics can best assess transformational efforts in their early stages. For example, one achievement of an R&D initiative is that a team learns by exploring and experimenting.

“For example, what if the only hurdle an initiative must clear to receive continued investment is that the company is likely to learn (not earn) from it? That is how Google has assessed transformational innovation from the start,” state Nagji and Tuff.

This concept of valuing organizational learning can be a tough sell, but it’s an important one. The most innovative companies are all known for their tolerance of failure and their belief in building a learning organization. This is central to their cultures that fuel growth and innovation.

Metrics to measure learning are, therefore, crucial because when ROI is not a persuasive metric where investors are concerned – and it is not where transformational R&D is concerned – much of a company’s value is realized in the learning that is achieved and the knowledge and talent amassed in employees based on their professional experience both inside and outside the organization.

A more quantifiable, traditional measure is the number of patents.

Measuring Patents

The number of patents is one measure of R&D efficacy that has persisted for years. On the surface, this makes sense. Many researchers within R&D departments come from academic backgrounds and published papers are signals of success. However, this measure is the equivalent of a blunt instrument when assessing a firm’s innovativeness.

Patents are subject to a number of confounding variables. For instance, if the firm partners with a university, there is a higher propensity to patent. If the firm conducts international business, there is a higher propensity to patent. Further, tech companies are moving towards open-source software development, resulting in a lower propensity to patent. These factors hinder patents’ value as a measure of communicating value to stakeholders or the greater market.

These limitations have led experts to creating novel metrics to better address the complexities of R&D success.

Modern Metrics for R&D

According to Anne Marie Knott, professor in Business at Washington University’s Olin Business School in St. Louis and author of the book “How Innovation Really Works,” “despite all the experts dedicated to helping companies innovate, the money companies spend on R&D is producing fewer and fewer results. In fact, my research shows the returns to companies’ R&D spending have declined 65 percent over the past three decades.”

To arrive at her conclusions, Knott used a metric that she developed to measure R&D productivity – the output that firms get in return for innovation inputs – which she calls research quotient, or RQ. This metric is a useful one for managers trying to secure R&D budgets from decision makers and investors. RQ calculates the percentage increase in revenue from a 1 percent increase in R&D spending across the organization (for the theory of RQ, see “Estimating Firms’ Research Quotient (RQ)”.)

The decline that Knott sees in the RQ of companies mirrors the overall decline in GDP growth in the United States for the past 30 years.

Knott suggests three theories that might explain the declining trend in returns from R&D spending. The first is that R&D work is more difficult in the current age because the best and most obvious ideas are quickly discovered and the remaining ideas are less novel or groundbreaking. The second theory is that there are diminishing returns to investment in research labor – more researchers decrease the number of innovations per worker due to duplicated efforts.

But Knott does have another explanation, which is that companies may have become worse at R&D.

Knott researched the financial data of all publicly traded U.S. firms for the past 40 years and found patterns related to industry technological advancements that reflect already evident disruptions – for example, the replacement of typewriters by computers. Knott’s conclusion from her in-depth research is that companies are experiencing less growth from R&D efforts because they are worse at innovating, not because innovation has gotten harder.

But it’s not all doom and gloom. What Knott identified using RQ is that companies aren’t necessarily fated when an industry declines or is disrupted; they often find opportunities elsewhere by diversifying.

Knott claims that “if the top 20 firms traded on U.S. exchanges had optimized their 2010 R&D spending using the RQ method, the collective increase in market cap would have been an astonishing $1 trillion.”

Part of that optimization uses RQ to leverage the potential longer-term benefits. According to Knott, “RQ also allows companies to link changes in R&D strategy, practices, and processes more closely to profitability and value.”