There has a been a lot of chatter out there around Startup 2.0 philosophy (I am staking my claim on everything 2.0). At the end of the day, a lot of the discussion revolves around how to make technology startups cheaper to build. The traditional rule of thumb is that an enterprise software start will take about $20-25MM investment to make it profitable. I know this seems like a lot, but when you consider the investment in developing complex enterprise software products, a direct sales model, marketing, facilities, customer support, etc.. all dimensioned by the time needed to penetrate deep enough into the market for brand and maintenance revenue to kick in - the costs start to rack up. Some, of course, have done it for less, and many have spent much more on their way to success or failure.
So, when people talk about the next generation of startups and venture capital, they are really asking "How do we make this cheaper to do?". Well, i wanted to lay out three approaches that I see emerging.
1 - Open Source as a Starting Point (not as a model)
While there are a few successes (MySQL, RedHat, Novell, etc..) of the "give it away for free and charge for maintenance" model, in my opinion is yet to be proved widely as a sustainable model. There is another derivative of the open source model that I am seeing emerge: companies who fundamentally use open source as a core component of the initial solution they deliver. In fact, I am invested in a company that started this way. StillSecure built it's first generation of security products on top of Snort (an open source IDS system). This enabled them to quickly penetrate a growing field of vendors in the security market with a 3rd generation level product, allowing them to moved cheaply towards more classically proprietary extensions and products.
From the entrepreneurs perspective, building on top of open source can reduce cost and increase speed to market. However, i wrote a long blarticle-based cautionary tale on the use of open soure on Parallax. Handing over your core code base to someone else (or a group of someone elses) is a blessing and a curse. Buyer beware.
2 - Offshoring
About 2 years ago when Newmerix was raising its second round, there was a lot of chatter from the VC community around offshoring. "You're going to offshore development, right?" was a pretty commonly heard phrase in pitch meetings. Does Nantucket count as offshore? For many reasons, we consistently argued that doing it "in house" at our stage of a company was a better way to go. In general I think the VC fervor around offshoring for startup companies has died down a little bit as ROI's have proven to be smaller than expected (at least in India, who's salary rate has increased double digits per year for the last few years). In addition, anecdotal evidence is saying that large companies who can open their own shop in India are the ones really reaping the rewards. There is enough chaos in a startup as it is, that waiting 12 hours to ask a pertinent technical question is a pretty debilitating problem.
3 - Saas
While I have a longer piece coming on the "real reasons" why SaaS will take over the world - here is a little snippet related to starup costs. One of the biggest costs involved in a startup is the N to the Nth platform support problem you face when shipping out CDs. Take a normal application today. It runs on a browser (which versions?), a database (which vendors and which versions?), an operating system (which vendors and which versions and which service packs?), and possibly interfaces with something else in the application stack (application server, BPEL services layer, packaged application.. which versions?). This causes an unbelievable amount of cost in the development and support cycle. I'd estimate it takes about 50% of your development time and resources to actually properly QA a piece of enterprise software in all relevant environments for your customer base. Not to mention the cost of managing a lab of all of these permutations (thank the heavens for VMWare). But then, when you move forward, you have the additional problem that customers are going to be using different versions and patch levels in the field. What upgrade paths do you support, which do you not - you have to test every upgrade path. Then, when a support call comes in, you've got to rebuild the exact customer environment to find the problem many times. Its a cost management nightmare, but one we've been living with for so long that we sort of don't talk about it anymore and assume its part of the cost of doing business. Frankly when Microsoft slices and dices the next version of a platform (e.g. Vista) into 6 marketing bundles, my stomach turns at the thought of how many new permutations we'll have to test (or not test and just "hope" it works).
SaaS changes all of that. You only ever have one platform stack to test - the one you run. You only have one version upgrade to test - the one you run. And putting out patches to your software can be done at your own pace, incrementally, and there is only one IT department to wrestle with - yours. It's a phenomenal model for startup companies. The classic virtual urban legend around this is the mere 100K Joel Kraus apparently spent to start JotSpot. Even if he is an order of mangnitude off on the build costs, the maintenance costs alone will be dramaticallly different than classic enterprise software.
So, while I'm not arguing the death of classic enterprise software (some things just need to be written as desktop applications - e.g. Newmerix Automate!Test product), I am arguing that startup entrepreneurs and VCs look very hard at viable SaaS business models, as the risk/cost equation has all of a sudden moved heavily in their favor.
Comments