Amazon has the distinction of being both one of the world’s biggest retailers and creating one of the fastest-growing tech companies in the world.

Leading figures from the company yesterday shared some of the business practices that helped fuel that success and led to experimental bets like Amazon Web Services, which went on to become the world’s largest cloud computing platform.

Fail often

First and foremost, Amazon has a culture that is built on a willingness to fail, said Doug Gurr, the firm’s UK country manager, echoing earlier statements by Amazon CEO and founder Jeff Bezos.

“Innovation and failure go hand in hand, we make a lot of mistakes and we fail a lot. We don’t believe you can separate failure from innovation,” he said, speaking at the AWS Summit in London.

“That’s absolutely fine, because innovations that succeed pay for an awful lot of ideas.”

Gurr gives the example of Amazon Marketplace, a service that allows third parties to sell goods via and that now accounts for nearly half of the items sold via the site each year.

However Marketplace wasn’t Amazon’s first attempt to create such a service, it was more like its fourth or fifth.

The first, known as Amazon Auctions, had just a handful of customers, said Gurr, as did the second.

“As Jeff has said publicly, we think Amazon auctions had seven customers, if he included himself, his siblings and his parents.”

It was only by not giving up on the idea and iterating on the service that Amazon figured out how to widen its appeal to small retailers and make it into the success it is today, he said.

Set impossible goals

Amazon’s R&D center in Edinburgh is one of three the UK that work on Amazon’s newest ventures, such as the drone delivery service Amazon Prime Air and its voice-controlled Alexa smart assistant, as well as the tech at the core of its retail platform, such as the AI powering its recommendation engine.

The center’s director Graeme Smith shared some of the practices the company uses to get the best out of its mix of software engineers, data scientists and machine-learning specialists.

A common challenge posed to staff is what Smith calls ‘constraint-based thinking’. Staff are asked to imagine how their approach to a particular problem or project would differ under vastly differing circumstances.

For example, how would you complete a project if, rather than having six months to complete it, you had two weeks? Or, how would you realise an idea if you had a team of 500 rather than five people? Or, how would you tackle this work if you had infinite computing power?

“One of the ways we get great ideas is we think about adding and removing constraints. You have to solve this problem in a totally different way.”

Allow personal creativity

Within the Edinburgh R&D hub, staff get one day every two weeks to work on a project of their choosing, a day that Amazon dubs ‘Anarchy Friday’.

These projects can be something aimed at self-improvement, a pet project or something to help the team that person works in.

“Most of them choose to work on something for their team and some of our best inventions have actually come out of Anarchy Friday,” said Smith.

The practice is supposedly commonplace among Silicon Valley tech firms, most famously at Google, where projects developed in this off time helped create Google News, Gmail and AdSense.

However, more recently doubt has been cast on how many staff take advantage of this famed personal time at Google.

Hold hackathons

Amazon regularly holds themed hackathons, programming contests where coders knock together applications in a short space of time, usually a couple of days or less.

These hackathons are staged both internally and externally, at universities.

“People come up with inventions, there are prizes, we have a People’s Choice Award, so folks get to vote for the ones that they like. Again some great ideas have come from the hackathons.”

It’s about the data stupid

From the roof garden of Amazon’s Edinburgh hub, you can see the Old Calton Burial Ground, final resting place of the philosopher David Hume.

Smith is keen to draw a link between the empiricism espoused by Hume in the 18th century and the data-led approach Amazon takes to designing it services today.

“Empiricism is the basis of the scientific method, that idea of rigorous measurement and experimentation and measurement again. That’s quite dear to us at Amazon, so I always enjoy pointing that out to people when they visit.”

Amazon is again treading a similar path to Google with this commitment to testing everything. Google uses data-driven processes to inform everything from its hiring practices to its design. When choosing which color to make links in search page ads, Google showed users many different shades of blue to measure which generated the most clicks.

However, this rosy account of Amazon’s working practices should be put in context of criticisms of how the firm operates. Last year, The New York Times ran a piece based on interviews with current and former workers who claimed employees were pushed to their limits in a drive for efficiency and productivity. Amazon CEO and founder Jeff Bezos later rebutted these claims.

Read more about successful companies…