Dave Naffziger’s Blog

I’ve long been a fan of Amazon’s E2, S3, FPS, SimpleDB, etc. These services are changing the way that early-stage startups make their platform decisions.

News outlets are quickly picking up that tonight Google is launching a hosted platform for web developers: App Engine.

I’m extremely excited about this for several reasons:

• Google is one of a handful of companies that have truly understand Internet scale. In addition, they have regularly dealt with the challenge of taking a small application and scaling it very quickly (nearly all of Google’s acquisitions go through this)
• By offering application hosting, Google is stitching together many of the components that Amazon’s services provide, making app development even quicker.
• While Amazon has been doing a great job pushing itself, the competition will be great and will raise the bar.
• It started off with Python support and is modeled after Django (which BrandVerity is built on and I’ve been very impressed with)

I look forward to seeing AppEngine develop further. I’m especially curious what it will take to launch an existing Django application on AppEngine.

I also learned for the first time that Google uses Python internally for its scripting. I guess that could explain all of the Google Python developer ads that keep popping up in my gmail account.

Update: I took a look through the documentation and now have a better handle on how AppEngine operates. It definitely borrows a number of aspects from Django:

• Django Templating Engine: The SDK provides Django’s templating engine (although it allows others to be used). The engine has been criticized by purists for allowing too much logic in the presentation layer, however I’ve found that the templates allow sufficient logic (if statements, for loops) to actually build webpages, while preventing the more complex view logic from making it to the front-end (variable assignment, method calls, etc.)
• Similar modeling system: In Django, you define your model classes and the framework provides common object methods that handle the sql underneath. The AppEngine models look very similar although it looks like the similarities are only conceptual (they definitely don’t share the same code and are interacted with in different ways). I can easily see how frameworks like Django (or Rails) could be modified to treat App Engine as another storage backend.
• Similar request and user frameworks: Django provides a few frameworks that are commonly used request.user (via contrib.auth) and its session middleware. These handle things like authentication, permissions, cookies, etc. Google’s APIs seem to function pretty similarly, although this isn’t exactly groundbreaking as these frameworks are fairly standard.

I don’t see a way to run python scripts that are independent of web requests. The O’Reilly article suggests that developers may utilize EC2 for batch operations. However, I have to believe that Google will support batch processing in some form as this forms a critical component of just about any web application (and I’m sure they don’t want to encourage people to write batch applications that are initiated by get statements)

I’m thrilled to announce that BrandVerity is now in closed alpha.

What is BrandVerity?

BrandVerity provides tools and services to identify and combat online trademark abuse.

The majority of the technology developed in this space has been developed by trademark abusers, while trademark owners typically resort to manual discovery techniques and are typically only engaging reactively.

Our first offering (still to be named) is a pay-per-click monitoring service that identifies trademark poaching by merchant affiliates.

What is Trademark Poaching?

If you’re a merchant with an affiliate program, you are probably already familiar with the practice of trademark poaching if not the name itself.

The most profitable search terms are those of a merchant’s own brand. Searchers that search for terms like ‘Macys’, ‘Amazon books’ and ‘Blue Nile coupons’ have already become familiar with the brand and are typically looking for the online store. The conversion rate of these visitors greatly exceeds that for generic searches.

Most e-commerce companies do not allow their affiliates to bid on these terms. The high conversion rates typically mean high bid prices and significantly higher cost of acquisition.

Grey and Black hat affiliates have found many ways to bid on these terms without the merchant noticing. They will day-part (run ads at night), reverse IP geotarget (run ads that target every city, but the one where merchant is based) and even copy the ads of merchant so that a visual scan of the search results won’t identify the poacher.

How does BrandVerity identify trademark poaching?

To put it simply, we know the tricks that the poachers use and our service uses technology that counters each of them. Some of the techniques are obvious and I’ll be talking about them later, but others aren’t and we’ll keep those private.

We understand that this is an arms race and are already working on the techniques we expect we’ll need when the poachers counter our techniques.

How do I get an alpha key?

I’m sure all of the readers of this blog have a burning need for trademark monitoring.

If by some chance you do have a need to monitor a few search terms on Google, and you’re willing to work with a bit of wet paint, drop me a note. The key bits of functionality are there but there is a lot that is still pretty rough. I’m watching load pretty closely so it may take a little bit before I send you the key.

Companies can easily spend $10K+ on identity and brand consultants while creating the company’s logo and related identity materials (letterhead, business cards, etc.). However, you typically only have access to the ideas and creativity of a handful of designers. Logo creation is much more art than science, so there is a good chance that you actually limit the process by working with an expensive consultant.

I spent $200 on a Sitepoint Design Contest to produce a logo for a new venture I’m working on. I received over 50 entries from ~20 designers. Some of the entries were definitely amateurish, however others were really well done.

The diverse ideas submitted by the designers were extraordinarily useful. I was able to look at many different ideas and would reinforce themes that I liked.

Logos are personal decisions
I solicited input from a handful of friends and colleagues. There was zero overlap on the designs - each person selected a completely different design or two as their favorites. The selection of a logo is always going to be a highly personal decision that reflects the unique likes/dislikes of the person running the process.

Which leads me to my final point:

The exact logo you use is one of the least important aspects of your business and you don’t want to spend a lot of time or money creating it. Find something that works and get back to creating value.

The guys at Cooler Planet, put together a very cool Google Map mashup of the growth of solar power in California.

You can see how solar panel installations have grown over time. I took these two screenshots of the installations in Northern California. I find it fascinating that there is more solar generating capacity in Berkeley than in San Francisco or Oakland.

2001	2008

They used hard-to-access, but publicly available data to build a yearly snapshot of the CA solar installations. I think its a great example of what you can do with government data that hasn’t really ever been put together in a compelling manner.

I’m thrilled to see them take a ‘content-rich’ approach to the solar space. While they don’t make money directly from work like this, it helps build their credibility and awareness in the space and helps establish themselves as a leading solar content destination.

(Disclosure: I made a small investment awhile back in Cooler Planet)

Conventional wisdom has held that if your’re selling software solutions to large companies you need an enterprise sales force.

The type of sales force is most frequently determined by your average selling price (ASP). If you can support $100K sales, then you use a field sales force. $10K to $100K and you use telesales (although I’ve seen it attempted at lower ASPs).

There are a number of reasons (beyond ASP) that would justify the use of a sales force (I’m sure I missed a few):

• The sale requires buy-in from multiple people (often in different organizations). Software hosted within internal IT departments is a great example.
• The decision makers are difficult to reach / difficult to discover. Some decision makers within an organization make themselves known and are easy to find. Others are uncomfortable with the spotlight and prefer to not have their name listed on the company website.
• The product is complex (or requires configuration, or has subtle benefits). The product can’t be fully conveyed on a website or in a white paper (that customers rarely read). Customer’s can’t figure it out on their own.
• The product is a commodity (or is perceived as one). The sales force gives customers a reason why they should buy a product from your company instead of your competitor’s.
• The solution isn’t a priority . This could because the pain your product solves isn’t great, or the capability it provides isn’t substantial enough.
• The product requires a commitment. It could be time invested to understand or configure the product, dollars to purchase it, or the opportunity cost of not selecting another solution.

The very nature of the products and the solutions they address requires a salesperson to sell them.

However, I’d argue that many products require a sales force because they are poorly designed. They are too generic, too complex, too undifferentiated…

Google (OK, Overture) completely overturned many assumptions about the enterprise sales force by creating massive amounts of value and by making it trivially easy to get started. 6 years ago it was inconceivable for a business to spend thousands a dollars a month on advertising without ever speaking to a sales rep.

I expect that one of the next waves of enterprise innovation will be software that doesn’t doesn’t require a sales force. It will require simpler products, that require minimal user investment, that address well-defined problems, and will very often be delivered over the web.

The products and services that accomplish this are going to price significantly below conventional enterprise-sales-based products. The conventional products will be structurally unable to compete at those price points.

Bijan Sabet at Spark Capital started a healthy debate on getting rid of non-compete agreements (which already aren’t enforceable in CA and other states). His assertion is that non-competes are a significant barrier to startups and innovation.

Jason Mendelson and Fred Wilson have countered that non-competes are very much in the interest of portfolio companies and their investors.

I actually think that both Bijan and Jason/Fred are correct (although they’re arguing non-opposing points of view).

Are non-competes good for companies (and startups)?
Yes. Non-competes create barriers in the talent market that reduce the likelihood of an employee from directly competing against their company.

What company wouldn’t want that?

Are non-competes good for economies?
No. The negative impact of non-competes is most likely felt by employees within larger companies. Very few employees at large companies have the negotiated non-competes that Fred & Jason suggest. These employees tend to get standard non-competes that offer neither just compensation nor limited scope.

The simple threat of a lawsuit is often enough to discourage the employee from leaving to start a company. Scott Kirsner points to HBS research (pdf) that reinforces the impact of this market inefficiency:

Marx and two colleagues [at HBS] looked at what happened after legislators in Michigan accidentally made non-competes enforceable in 1985. Inventors were suddenly 34 to 51 percent less likely to move from one company to another. And the “star” inventors were the least likely of all to move.

I’m not quite sure how this translates into economic impact, but it does suggest that it could be substantial.

Non-disclosure agreements provide companies with significant protection. Non-solicits provide even more protection (and have been enforced in states like CA). The non-compete does not create net economic value.

What is a company to do?

Bijan ended his post with a call for VCs, CEOs and founders stop requiring non-compete agreements. While I applaud his willingness to take a stand, I don’t think that it is reasonable advice for a startup.

Business can and should take every ethical strategic advantage available.

Amazon thinks that the patent system needs reform badly. They aren’t boycotting the system by refusing to file patents or allowing competitors to infringe their existing patents. Instead they’re lobbying for reform.

Don’t begin reform by placing your business at a strategic disadvantage. Instead lobby hard and change the system from within.