Only one father of the dot-com boom is still innovating today--Amazon CEO, Jeff Bezos.

Let's look at one of Bezos’ innovation initiatives; its grocery delivery service, AmazonFresh. E-groceries is a small, unprofitable market, but Bezos views that as a challenge to his innovative spirit.

After all, the $603 billion grocery industry sports 1.1 percent profit margin. And with consumers buying 1.2 percent of that amount online, it remains to be seen whether anyone can make a decent return on investment in e-groceries.

Take a look at the landscape: Webvan is the best example of a company that tried and failed in that quest. It raised $375 million in its November 1999 initial public offering, achieving a peak stock market value of $1.2 billion. The company bragged about its 26-city expansion plan, signing a $1 billion Bechtel contract to build high-tech warehouses worth $30 million each. Then, it filed for bankruptcy in July 2001, after losing money every year.

A better model of online grocery delivery is FreshDirect. As I learned in a May 2010 interview with its CEO, Richard Braddock, a former Citigroup executive, FreshDirect was founded in 2004 on the idea of making a profit.

And FreshDirect has done that. “In 2009, FreshDirect generated more than $250 million in revenue and it has been profitable since 2008 and was substantially profitable in 2009.”Braddock made the business profitable by operating in the densely populated New York region before expanding to Philadelphia in 2012,

Bezos clearly took a lesson from FreshDirect’s success. AmazonFresh started in 2007 in the Medina and Mercer Island neighborhoods of Seattle and by 2012 had spread to other Seattle communities. On June 10, the service rolled out to a limited number of Los Angeles zipcodes.

Bezos, as I see it, has also taken four lessons from Webvan’s failure:

1. Hire those in charge of a failed strategy. 

If you wanted to get into a new business, you could not be faulted for hiring someone who had been spectacularly successful in that new business to run your operation. But Bezos did the opposite-- hiring executives responsible for Webvan’s failed strategy.

AmazonFresh is led by Doug Herrington, Peter Ham, Mick Mountz and Mark Mastandrea, former Webvan officials who had spent years analyzing and fixing the problems that led to its demise.

Hiring people who have failed is hardly a guarantee of future success. But it seems to be working for Bezos because these four Webvan vets know the business intimately, have thought about what they did wrong, and are eager to prove that they can apply the lessons learned to revive their reputations.

2. Acquire technology invented by the failed executive.

A few years ago, a colleague who studied at MIT with Mountz invited me to meet him at his robotics company, Kiva Systems. Mountz built Kiva on ideas and technologies originally developed at Webvan and that are a key part of AmazonFresh’s strategy.

When I saw Kiva’s product in action, I was blown away. It consisted of huge frisbee-shaped robots that could be steered via a grid of floor magnets. The robots picked up the warehouse rack containing the item in the customer order and delivered it to the operator. He picked the item from the rack; scanned its barcode to verify that it was correct; and put the item in a carboard box with that customer’s shipping label.

In 2012, Amazon acquired Kiva for $775 million in one of its largest-ever acquisitions.

3. Focus on attractive micro-segments.

While the grocery industry is huge, only a tiny fraction of that market is attractive as a place to operate an online grocery business. Finding those micro-segments is a crucial element of AmazonDirect’s strategy.

Webvan executives later realized that it makes sense to segment markets based on mean travel time between delivery stops. For example, a single block in San Francisco may house 200 people, while one block in Moraga only has six.

AmazonFresh and FreshDirect only deliver to densely populated areas. Certain neighborhoods in Seattle have many front doors with AmazonFresh totes. That’s where Amazon saw the lowest mean travel time between deliveries.

4. Fix the business model before expanding.

One of the most interesting points of the Webvan case that I taught in my Competitive Environment and Strategy class was the company’s obsession with getting big fast-- its then-CEO, George Shaheen, likened running Webvan to building a rocket to Mars.

Mike Moritz, a Webvan board member and partner at Sequoia Capital, said that Webvan “committed the cardinal sin of retail, which is to expand into a new territory before we had demonstrated success in the first market.” According to Moritz, Webvan was failing in the Bay Area market even as it expanded into other regions.

Through AmazonFresh’s very gradual expansion, Bezos demonstrated that he learned from that Webvan mistake.

The four lessons he applied from Webvan’s flop highlight Bezos’ broader lesson for start-ups: winning flows from focusing the best brains and technology on making life better for consumers.