Amazon buildings

The Amazon culture and its effect on Seattle area startups

There is a lengthy piece on the New York times on Amazon’s employee culture. It is largely negative on their push-to-the-max, dog-eat-dog world and stack rank nature of the company.

Having worked with, know of and learned from over a dozen employees at Amazon, I can tell that it attracts a certain kind of employee. It certainly is not as bad as the article indicates, but many over achievers love to work there.

While many ex-employees have talked about the “challenging” work environment, I have personally talked to ex Amazon employees who wish more companies were like Amazon after they left.

What are the parts that they like the best. Relentless push to do better when you believe you have hit the peak. There seems to be no real “peak”. It is a state of mind. Doing better is what we all strive for and Amazon’s culture pushes you beyond that.

The intense focus on numbers and customer metrics over going by the highest ranking person’s opinion. This happens in many companies, including the best run large technology firms. When the highest-ranking person enters the room, the conversations is about “their experiences”, “their interactions with customers” and their prioritization. Amazon’s not like that, is what people tell me. Data and analysis trumps opinions.

Disciplined experimentation is another area that most folks believe Amazon excels at. The willingness for people to be given the latitude to try multiple experiments over time prevents boredom.

Amazon’s culture may not be good for 80 or 90% of the folks who are seeking employment, and that’s okay since they dont need all the employees of the world to work for them. Just the ones that value work over everything else.

Now that I have explained the other perspective, I have an observation about the impact of Amazon’s inwards looking focus on the company and its customers on the local Seattle ecosystem for startups.

Amazon buildings

Amazon buildings

The impact is pretty significant. The number of ex-Amazon employees looking to start companies is far and few between.

There are a few, for sure, but most employees at Amazon work at a startup themselves (Amazon) and are either a) happy working there, b) are burnt out at the end and want an easier job at a larger company or c) dont ever have any time to engage with startups or have time for side projects.

What this means is that many ex Amazon employees make good employees at companies who are trying to reboot their culture, but not a good fit for starting companies.

Uber Platform for Hotels

Becoming Ubiquitous – How Uber “apps” are making it a platform – #UaaS

Most startups would believe that have “made it” when they become a platform. That’s usually when other startups depend on their underlying technology as an infrastructure and include it in their product. Given that most startups are now developing API’s so other startups can leverage this, it is not unusual to see many new startups developing on another startups’ technology.

In the last few months that’s the pattern I am seeing with Uber. Emirates already offers its business and first class fliers the pickup and drop off service from their home or office to the airport and now there’s a company that is partnering with other airlines to provide an app that users Uber as the car service for the airline’s top customers.

I read yesterday about an Uber based escort service as well.

There are many other apps that have been in the works to create a service using Uber. Since they have opened their API, many use cases have emerged.

Open Table has been considering integrating its app with Uber so before and after your table reservation you can book an Uber to pick you and drop you back.

Same for Museum apps – they are looking to offer a full service to provide an ability to use Uber to give the patron a seamless experience.

There are many bar apps that also have integrated with Uber. After a drink or two, or if you checkin to a bar at a given time (using Swarm or Four Square for example), you will be automatically prompted to use an Uber to head back.

In a few cases, I have heard of Ticketmaster and other ticketing apps integrating with Uber for event based taxi rides.

There are many other use cases as well, but all this means more data for Uber.

Where are people taking rides to and from, when are they going, and why (which can be inferred in many, but not all cases).

To become a platform though, you first need a reliable service. Then the ability to evangelize the platform to developers, which means someone that can be at hackathons, build initial prototypes, troubleshoot for other developers and also help them find a way to monetize.

That’s the path Uber is going towards.

Uber as a Platform

Uber as a Platform

I had a chance to meet a team of developers who are building a bachelorette party app last week. The 3 cofounders (all women) were building an experience platform for the entire party. From inviting the friends of the bride to selecting the venue and onwards, they can help plan the entire party.

A key part of their platform is to bring all the “friends of the bride” together to the same venue. That’s where they are hoping to use the Uber API to integrate into their app.

The incentives offered by Uber are pretty good. They claimed they had met with the Uber team and they were told they could get 10% to 12% of the fare as a commission. Which I thought was pretty high, but it makes sense to Uber only if it believes the folks coming to the party wont already take an Uber to the party.

Welcome to Uber as a platform.

JP Morgan Server Market Analysis

New Market analysis – The data center “server” market #napkinStage

Another weekly feature I want to start is an analysis of a new market, so I can understand trends and learn about opportunities.

The hardware market for servers is a fairly large market overall. These are expensive high end systems that have a 7-10 year life-cycles and tend to be shelved and end-of-life-ed after that.

According to IDC and Gartner over $50 Billion dollars is spent on server hardware each year. That constitutes about 8-9 million units each year. So, the average unit price of a server is about $5000.

There has been a dramatic shift on both sides of this market – the buyer (service providers and data center owners) and the sellers (OEM’s, ODM’s and Contract Manufacturers).

First the buyers. The biggest tend has been the rise of the cloud. From 70K to 85K buyers and service providers in 2000, there are now only 25K buyers. The rest of the companies have “given up and gone to the cloud”.

Data Center Server Market Distribution of Buyer

Data Center Server Market Distribution of Buyer

The second trend, related and a resulting effect of the first trend, on the buyer side we have gone from a even spread at the top (long head) to a sharp head (consolidated top buyer profile).

What that means is that in 2000, the top 100 buyers accounted for 25% of the server purchases. Now in 2015, the top 10 buyers are accounting for 25% of the purchases. That indicates consolidation in a significant level. Companies like Google, Facebook, Twitter, Baidu, Tencent, Alibaba, Sina, Microsoft, IBM (with acquisitions) and Amazon (AWS) now account for a quarter of all server purchases annually. The next set of buyers – traditional ISP’s such as AT&T, Verizon, GoDaddy and about 1000 others account for 10% of the market.

Finally buyers are now also purchasing more “commodity” servers with cheap hardware components and systems and going away from expensive OEM servers. While HP,Dell, IBM and Lenovo continue to be the market leaders, more ODM’s such as Quanta, Wistron and Invetec.

What this means is that the major OEM vendors are losing market share to ODM’s and contract manufacturers or Electronics Manufacturing services (EMS) such as Foxconn and others.


Operating Costs

How accelerators make money to manage operating costs

There are over 500 startup accelerators in the US and over 1000 worldwide. Most accelerators are aligned with Universities (at over 35%), some are government funded (local government mostly) at 29% and some (15%) get grants from rich individuals and institutions such as Kauffman Fund. The remainder (21%) are privately funded accelerators such as 500 startups, Angel Pad, etc.

First, the definition of a seed accelerator, so we can understand the scope of the program:

A fixed-term, cohort-based program, including mentorship and educational components, that culminates in a public pitch event or demo-day.

While there is no reliable data on how many of these accelerators are doing well, graduating great companies and surviving, there is some data on how they are managing to stay afloat and “keep the lights on”.

Most accelerators, raise some money to invest in the startups they fund. Many (over 61%) offer some form of space to their startups to operate in during the cohort. Accelerators also have a staff of 1-5 people (some even more, but the average is 1.8) to manage the program, support the startups and recruit, select and engage the local community and ecosystem of entrepreneurs.

All this costs money. In the US, that’s usually upwards of $400K (that’s the low bar) and in other countries, more than $250K per year.

Typically the cost of the space and maintenance is about 30% to 40% of the budget, the cost of people about 40% – 45% and finally the cost of programs, marketing, etc. tends to be about 20%. This excludes the investment in the startups.

While investors in the accelerator are willing to fund the startups (and take a % stake in them), most are unwilling to pay a “management fee” for running the program.

Having interviewed and talked to many accelerator programs, over the last year, I have a list of 9 different ways programs have tried to raise the operating costs of the accelerator. I thought I’d document these so it would be useful.

  1. Sponsorship: The most frequently used means to raise operating funds, is to have large corporate sponsor. Some local government organizations also sponsor these accelerator as a means to be connected to the community. Many accelerators also raise sponsorship from local legal, accounting and real estate firms who benefit from the startup community or wish to target entrepreneurs and startup talent with their products and services. Nearly 60% of companies and 30% of all operating budget funding is sponsor driven for the 15 accelerators I know.
  2. Events: Many accelerators run events that aid future entrepreneurs, community participants and local businesses. These events are typically networking opportunities and charge attendees a nominal amount of money to cover the costs, enable marketing for the accelerator and pay for the “marketing resource” at the accelerator. Some accelerator programs also put together hackathons and still others run large industry events to generate operating cash. Typically the problems with running these events is that they take up resources and time, but if you can generate enough cash from these events, you can support 1-2 resources who can help with other activities at the accelerator during the non-event days.
  3. Entrepreneur-in-residence programs: A relatively newer program is the EIR, where employees at large companies or those at smaller ones who want to learn how to be more entrepreneurial, end up spending time at the accelerator in exchange for a fee. Typical fees are between $25K to $50K in the US. These EIR programs are full immersion programs and last 6-12 months or 1-2 cohorts. During the program, the EIR is going through the entire process from start to finish and “learning on the job”. Many of the participants end up becoming investors or entrepreneurs at the end of the program and return to their companies, learning about lean methodologies, innovation approaches and how to build on an idea and bring it to market.
  4. Grants: Both government and private donors typically give grants (no strings attached usually) to accelerators to support entrepreneurship, which promotes local jobs, makes a city more attractive to larger companies and also helps the local economy.
  5. Rentals: Many accelerators charge a portion of their investment as a fee for the space during the program per seat. So, if the accelerator invests $100,000, and the startup has 3 founders and employees, then $5000 might be charged per month of the startup for the 3-4 months they are in the accelerator space. This is more of the domain of co-working spaces, but many accelerators are starting to do this as well.
  6. Research Reports: Few accelerators I know write research reports based on their startup data for larger companies. These companies pay for the syndicated research reports so they can use them in their internal presentations. These research reports tend to be focused on a particular area of expertise and also a market domain. It is not unusual to see companies pay $50K for a syndicated report for the year about the startups within a specific area of their interest.
  7. Code Academies and Hacker schools: Many accelerators have also joined with coding schools, which teach programming to new and interested talent. This serves two purposes. First, the accelerator can raise cash by conducting training and second the graduates become good source of talent for the accelerator startups, who pay a fee to recruit the talent.
  8. Innovation scouting for larger companies: Many larger companies are also looking to recruit talent, acquire companies and learn about new disruptions and innovations. These companies are willing to pay a little money to scouts who can help track, recruit and manage a startup pool of entrepreneurial talent. Many accelerators provide this as a service to larger companies.
  9. Distribution, Sales, Design and Marketing consulting: A few early stage accelerator are also providing the equivalent of the “coding” school for non developers by running marketing and sales training programs. The difference is that the graduates are employed by the accelerator program and they end up being consultants to the startups who charge a fee for their services.

These are the various programs I have seen, and I’d love your input on if I have missed any that you have seen.

Elements of UX design

Why some apps and websites have never changed their user interface #DontFixWhatsBroken

The Google search “user interface” has been the same for over 17 years now. The simple text search box with a button for Google search and “I’m feeling lucky”. That’s it. Nothing has changed at all.

Feeling Lucky

Feeling Lucky

Same for iOS and similarly for Instagram, etc.

Media properties go through several changes every 12 to 18 months. Some even undergo changes more frequently than that.

Why do some websites and apps – Facebook, etc. change so often and have the users go through the pain of learning the new experience?

And why do some services NEVER change at all even after user feedback about their experience.

The basic user interface theory suggests that once your user / customer knows how to make something work, they like it and get used to it. After that it is hard for them to change. Many of your users may not even like the change, since it forces them to learn new things and be productive at the same time.

I have a theory of User interfaces, which is just a theory, but I’d love contrasting opinions on this. I believe that most users dont care about the User interface. They care about the experience and want it to be seamless, easy to understand and simple.

Which means, they expect the complexity to be hidden.

As an example the search input for Google has not changed, but the response pages have dramatically changed over the years.

From a simple list of blue links, now, Google provides contextual and relevant information.

That means the search engine has changed a lot (in the back end), but the complexity is largely hidden from the user.

Which is the reason why most apps in the future wont have a User Interface is my belief.

The user interfaces we know of are mostly there or will be there soon. Learning new interfaces will take us a long time.

A combination of micro services and service based apps, will result in the death of mobile apps and pretty much most apps.


The input elements for apps will likely be questions (business apps) or statements (communications) via voice, gestures, etc.

The output elements will have multiple levels of detail (overview, specifics and detail) and while I think they will evolve, they will start to coalesce around the known.

I’d love to know if you think I am wrong.

Granular Pricing and Event based Pricing as a Service – (Praas) #NapkinStage #WillFund

One of the things I’d like to do every week from now is to talk about problems that I know about. I am hoping I can get entrepreneurs interested in working on these problems. Of course, most of these will not be fully vetted or even viable. That’s where customer development comes in. Honing in on the specific problem set will be something that needs work.

The problem I want to talk about today, is pricing. As more software companies offer their products on the cloud, and more companies are becoming “full stack“, there is a big need to help them optimize pricing to capture value instead of usage alone. Pay for performance instead of pay for usage is not a new concept.

Pricing by usage means that as people use more, you will charge them or they will pay more. That’s typical of mobile phone plans for example, the more data you consume, the more you will pay. That’s consumption pricing.

Pricing by value or pricing by outcome are two other means. As companies get more “full stack”, it is becoming more clear that they intend to not price on standard known means. For example Uber’s surge pricing is “value” pricing based on demand. GE has started to price its jet engines on availability and uptime instead of a maintenance fee, which increases its ability to execute differential pricing. If the engine is available more, GE benefits as does the airline, which leads to lesser downtime and hence more profits for the airline. This extra profit is what GE wants a part of.

More specifically, in terms of software defined pricing, it is becoming clear to me that granular pricing, the associated billing is equally important.

The solution is “Pricing as a Service” or PraaS.

I can envision an API driven offering, which is used by the developers of any SaaS company. The offering will manage the pricing pages and the billing for the company (different from recurring billing and transaction that’s done by Recurly, Chargebee, etc.)

The API’s will also be available to customers who want to bill granular components of their product. It should provide the ability to manage pricing tiers or create “packages” and also allow for multiple SKU’s to be created from base offerings by segment of customer.

I think this is a sketch of the idea, and there needs to be a lot of work done to talk to SaaS companies to understand their problems with pricing their products.

The other part of the pricing problem is metering and billing. I know of many SaaS companies who are looking to expand their customer footprint by offering their products via public clouds (AWS, Microsoft Azure) to large customers. For these companies, new metering and billing by components, or value based is a lot more profitable than plain usage or user based pricing.

I think Pricing as a Service is an important solution, so if you have a background (been an entrepreneur – failed or succeeded before) in a SaaS company, and a team of 1 or 2 cofounder who can build large scale API based systems and have the desire to build a company, you should reach out to me. I’d love to learn more and if your team fits the bill, I’d be willing to fund this idea. If you are a solo entrepreneur, I’d not be interested. If you have not been an entrepreneur before, again, it would not fit my criteria.

Lessons from 3 founders on surviving the “near death” of your #startup

Startups were largely meant to be an experiment. An attempt to solve a problem (that may or may not exist), or to bring to bear an idea whose time “has arrived”.

Experiments, though unlike many startups, have a hypothesis, are conducted largely in test environments and have the ability to come to a logical conclusion after their period of testing.

Most founders I know have survived at least 3-4 “near death” experiences at their startup. Most of the near death experiences come within the first 18 months of forming the company. Surprisingly, they dont come in the first 3-6 months, but after that.

A near death experience is categorized, as a point where the founders believe there’s no point in continuing to build the company any more and they would like to shut down. Most times it is because they ran out of money. A few times, even when they have the money, they decide to shut the startup down since what the founders signed up for is now different from reality.


statistics-on-failure Credit:

I had a chance to work with many founders over the last 10 years, and the thing I have noticed about founders who survive the journey is that there are 3 things they all seem to have in common:

1. A bias for action, not deliberation. It is not as if you wake up one day and suddenly feel like your startup is going nowhere. The buildup to your closure is usually a series of events and feedback that points to inconsistencies with the assumptions you made when you started your venture. The best entrepreneurs have a very strong bias to take action on the data and quickly put new experiments in place to validate the new ideas and tactics.

2. They take an extremely short term (hours, day) view of their survival needs. There is an old saying in cricket, paraphrased – “When faced with a daunting score or a large total, instead of trying to focus on the target, focus on the next ball. Leave the rest to time. If you focus on the next ball and surviving the next ball it is likely you will worry less about the many other distractions that come with figuring out scale and other “to be solved later” problems.

3. They focus only on revenue generation activities, after cutting costs to nothing. I notice that many entrepreneurs say they cut costs to “bare minimum”. Which, in my opinion is still high. If you are spending any money at all (for development, marketing, etc.) you should cut them to zero until you have your revenue plan in place. They remove their “small office” space, eliminate “web hosting charges” but looking at creative ways to use free resources to “keep the lights on” without spending money.