Jeff Bezos and The Lean Startup

I just finished “The Everything Store” about Jeff Bezos and the fascinating rise of Amazon. Bill Gates, Steve Jobs, and Jeff Bezos… All known for their visionary yet autocratic rule that creates huge leaps forward but often with organizational strife, politics, and an inability to scale.

I’m also a fan of the book “The Lean Startup”. The term “Lean” in the book is unfortunate because it gives the impression that it’s about doing more with less. The actual idea is different and much more interesting.

In short… Start with the customer. Don’t just ask them what they want and then go build it. Produce the minimal viable product you can give them to see whether they actually like it. Not minimum as in “it works and 80% of the features are there but it’s beta quality”. But minimum like a web page that describes it with a “buy now” button. Put it out there and pay Google to get it in front of 10,000 people. If 1000 people click you have one problem to solve. If 3 do then you have another. Either way you learn.

These two approaches appear to be in conflict, but they actually work great together. Bezos provides plenty of leadership, but coupled with the philosophy that the data wins. And Lean Startup techniques are a great way to get the required data.

An important part of Amazon is writing product descriptions. In the beginning this was done by a large editorial team. In time, the algorithm team developed an automated alternative. And there was much debate and political wrangling. Bezos dictated not which to choose, but how to resolve the debate. It was resolved Lean Startup style with customer testing. Use both on the same products and see who buys more. The algorithm won and the decision was made.

The two extremes taken together seem like a winning combination. Leadership for things that really matter. Serious Lean product development for the rest.

Middle roads are problematic. Microsoft pushed the style of Gates and Ballmer down through the organization in a vast hierarchy of management. All of them working to emulate their leaders and be little visionaries. This grows rigid and political as the thousands of visionary leaders battle each other for dominance. Distributing and delegating the work of the leader by driving that style down through the organization ordains internal political battles.

In theory that leadership could be cleanly distributed. The sales leader says what customers want. Product management combines that with the overall market and owns the product plan. Engineering has the last word in how it’s done. In reality that’s challenging. Mobile devices always need to be lighter and have a longer battery life. Sales reiterates the obvious the need to do both. Product management tries to figure out the balance based on experience and abstract data, aka personal opinion. And engineering influences with technical reality mixed in with their own beliefs. Is it really not possible to do both at the same time? Or is engineering just not trying, interested, or competent? Meetings, politics, dysfunction.

Similarly, customer-focus without going all the way to Lean Startup ideas leads to product mediocrity and organizational bureaucracy. Lots of middle management and politics. With a little coaxing, customers will say they want all kinds of things. Especially if they are tortured by hours of powerpoint presentations and then asked “so, are you interested in that? Is that where we should be heading”. I don’t remember any customer ever saying to me “oh no, that’s a terrible idea”. That “customer input” is then taken back to the product management organization where it is mixed with executive vision, competing priorities, new product development, engineering capabilities, and much more. Add in countless meetings with all the managers for all the different parts of the organization to gain and keep consensus. And slowly a plan develops, which is then diligently executed and delivered a year later.

Asking customers in 2004 if they wanted a phone with a touch-screen and no buttons would not have worked. That required a visionary leader. Or Amazon choosing low prices vs targeted pricing based on analytics. Those decisions take years to fully play out.

Deciding on the shape of the bevel of the glass? A visionary leader can also answer that question, but customer trial is what scales. Delegating it to a middle management team of industrial design, engineering, and manufacturing needs to be done with guidance for how decisions are made. Management certainly needs to have expertise and lead – you can’t try out every alternative. But for the significant, hotly debated decisions, ask the customer. If you still can’t figure it out, ask your leader.

Services Causing Overstatement of GDP Growth

I know enough economics to get into things, but generally am far from complete understanding. So I’m writing this about something I’ve been thinking about and hope for comments from people more knowledgeable. In my Googling around I can’t find this brought up anywhere, so I’m assuming I’m wrong. That or it’s really interesting.

Say you and 999 of your friends live in a nice little country called Libertania. Everyone is gainfully employed and the 1000 people earn $5/hour, $10k per year producing a GDP of $10 million.

People in Libertania are naturally quite self-sufficient. I spend 40 hours a year fixing my own car, and 35 hours mowing my lawn – 75 hours total. You are more efficient at both, requiring only 30 hours to fix a car, and 32 hours for lawn mowing – 62 hours total.

One day we each read a book on the theory of comparative advantage. Following the recommendation, we decide to specialize. You fix both our cars. This takes you a total of 60 hours, saving you 2 hours or $10. I mow both our lawns, taking a total of 70 hours, saving me five hours or $25. Granted, it would probably get more equitable in time but it doesn’t really matter.

Win-win! Together we saved seven hours in time, worth $35! Even though you are better at both things, we can both benefit by specializing in what we do best.

With our savings, we could now either relax and enjoy more leisure time, or use that time to work more and earn an additional $35. But for now lets ignore that.

Other citizens of Libertania notice us relaxing with our extra time and want in on this deal. Bartering services becomes rather tedious to keep track of so we all switch to just paying each other. Keeping it simple, fixing a car for a year costs $200, and a year of lawn mowing costs $200. The system works so well that all citizens now specialize in one or the other.

The government of Libertania (a small one…) is a modern economy and interested in knowing GDP – roughly, the total goods and services produced by the economy. Before this specialization trend the GDP was $10 million. Now, each citizen pays another $200 a year for one of the two tasks, a total of $200,000 changing hands. The GDP increases to $10,200,000, and the government marvels at it’s success in raising GDP 2% over the prior year.

But remember, what we actually saved was 7 hours in time between us, for a national savings of 3500 hours annually. At our average wage of $5/hour, that’s worth $17,500. Still good, but it’s hardly the $200,000 increase that the GDP showed. $182,500 of that increase is due simply to a transaction taking place. It didn’t add any actual value to anything – nothing was produced. It appears to be 91% overstated.

The more we specialize and exchange services the more we grow GDP. That’s good. But the actual value to the citizens is overstated in the GDP growth. They didn’t get 2% better off. They got 0.175% better.

I don’t know what the actual gap between true comparative-advantage savings, and the transaction amount is. But I’m confident that it’s significant. To be the same would require the savings for each of us to equal the price of the service, which is absurd. We would have had to have saved 40 hours each in the deal to justify the full $200 in GDP. You can play with the numbers but I can’t see any way to make it close to accurate.

But you say, remember the 7 hours in savings. We’re hard-working Libertanians and would spend that time working and making more money to make our lives better. So in addition to the transaction, we work an additional 7 hours, or a total of 3500 hours for the entire country, for a total of $17,500 in true, new, value and income (depending on how you’re measuring GDP)

But, the true value did not go up. I traded my newly-freed-up leisure time for money, which is a net zero.

So now, that income is added to reported GDP growth to make it $217,500, with a true increase remaining $17,500. That actually makes the overstatement worse at 92%.

That seems like a big deal. Services are roughly 68% of the US GDP

Almost any conceivable number would indicate that GDP growth has been significantly overstated.


Next up, tax implications…

The tax rate at which the original transaction becomes a net loss for us is equal to 1 minus the overstatement – or 9%.