Qualities of a successful web startup

Providing value from customer one

Thriving communities support even the most flawed premise. The classic chicken and the egg problem must always be kept in mind. Even ideas that are inherently social need to provide significant value in their own right when there are only a handful of users. The idea must sustain itself for long enough to build a userbase.

Content

As a technically oriented person, it is often tempting to devote myself to developing the infrastructure and consider the content as an afterthought; the miscellaneous minutiae of online business development. The fact is that in most cases, content matters significantly more than the technical. Quality content is the key asset that can drive growth and maintain repeat customers. As with all assets, content can be expensive. Unique, well developed content can be prohibitively expensive for a startup.

Conversely, content is ubiquitous.

The Creative Commons licenses provide a wealth of resources to draw from, many of which are commercially viable. Attribution is a small price to pay, and at times can even be a selling point. The company Threadless has built a thriving business around community sourced fashion design. Artistically minded members of the community submit designs to a vote. The winners receive a cash prise and the exposure of having their work go into production. The kicker, of course, is that Threadless keeps the profits from these subsequent sales that although undisclosed likely dwarf that handed out to the community. Although they assume some risk, the social voting component acts as a focus group to gauge interest, maintaining small runs and print on demand production further mitigate this. Attribution has never hurt Threadless; in fact it’s one of their key marketing techniques.

Projects such as DBpedia and Freebase aim to extract structured data from semi-structured sources such as Wikipedia. These datasets are then exposed for third party applications to query and build upon. OpenStreetMap aims to provide freely available geographical data and street maps unfettered by the licensing constraints and associated costs of proprietary solutions. Filtering these disparate sources for suitability and intertwining them in a meaningful way is a difficult challenge, but an achievable one.

Not to mention purely user content-driven sites.

Residual income

When generating sustainable business ideas, a common trap equates the freelance path with business ownership. These two approaches couldn’t be more divergent. It is a cliché, but you want to build a business not a job. If a business isn’t self sustaining after the initial setup phase and grinds to a halt when you are out of the office you are doing something wrong.

Revenue stream

It is my contention that potential customers willing to pay for a quality product or service still exist. The ease of setting up subscription payment models makes this approach feasible for even the smallest of companies and also allows potential customers to trial your services with minimal outlay. Subsequently, by keeping the price unobtrusive you can achieve customer lock-in through inertia: although I disagree with preventing a customer access to their data, there’s nothing wrong with making migration unattractively painful compared to a small monthly credit card debit. This also provides a reliable revenue stream that can be forecast and used to build greater assets. The tiered plan approach is common and tested, however I would like to note that I have read commentary (unfortunately I lack the attribution) musing that the cheaper accounts are often loss leaders. If this is the case, one would have to look at their conversion rates carefully to calculate if offering these lower tiers is worthwhile. (Dan, please provide your opinion in the comments. I need an economist’s perspective.)

As I’ve likely mentioned here before, I have long felt an affinity towards contrarian investing that I like to express using the oversimplification “if the majority are invariably wrong, doing the opposite will invariably be right.”
There’s no point following the crowd if you can reasonably determine they’re wrong.

Not to mention advertising and hybrid approaches.

Marketing

Marketing starts with idea generation. “Build it and they will come” is fun, but if you can ascertain who they are this will inform your actions. Although arguably overzealous, the approach of defining your customer not only in broad generalities but in specific, human terms is a useful one. A few brief sentences describing your customer’s name, personality, interests and desires gives you someone to aim towards.

I read an story in Business Week this morning on the use of mirco-blogging services to reach out to customers. It’s predominantly a fluff piece, but provides an adequate high level introduction and case study.

Unimportance of legalities

In contrast with my focus on licenses in the content section, it isn’t hard to think of disruptive businesses that threw caution to the wind: YouTube, Napster, etc. Deploy now and ask questions later.

I have switched sides on this point several times in the past, but have finally concluded that I would rather fight to save something tangible than protect something imaginary.

Had I developed something that received enough attention to raise the rabid dogs (read: lawyers) I would be rather chuffed.

Financing

Bootstrapping works.

I am rather amazed at the multitude of people frothing at the mouth over Google entering the Venture Capital space with Google Ventures. Technological enabled startups are predominantly appealing as their initial and ongoing overhead is minimal. Chances are that buying one less coffee per week will cover the hosting costs of a web application of simple to basic complexity. Costs then generally scale with bandwidth, which is easy to cover if you’ve designed your revenue stream correctly. The only reason I see justifying going for VC is to make connections. Connections are important, money is easy to come by and startups are cheap.

Crowd sourcing

Ever since Amazon Mechanical Turk (MTurk) launched back in 2005 I have been generating ideas on how to leverage this unique offering. Billed as “artificial artificial intelligence” MTurk provides the ability for even the smallest company to outsource small, repetitive tasks for humans to complete on a pro rata basis. With an API that allows for submitting tasks automatically, it can be used perform aspects of a services workflow that cannot easily be automated by software. Additionally, I also see it as an avenue to temporarily implement components that could be automated but financial or schedule constraints make it cheaper or faster to outsource until the userbase expands beyond a certain threshold. Likely less risk is involved as well.

Imagine piloting a new feature on a photo cataloguing website that automatically tags photos that contain people wearing hats. You could take a week training your complex hat recognition algorithm only to find out that users don’t particularly care about hats. A better approach may be to choose a subset of dedicated users to pilot this new feature for, and tweak their workflow to send each image they upload to MTurk for human categorisation. This way you get immediate feedback on the idea without the upfront risk. Perhaps you will discover the users wanted moustache recognition instead.

Shoulders of giants

I came across this funny anecdote of how Lisp destroyed one man’s programming career.

Existing software exists to solve problems you haven’t even thought of:

• Scarling/Kestrel provides persistent queues written in Scala to bolster Twitter
• Jabber/XMPP protocol and its various implementations have seen commercial deployment from the eminent Google Talk to the niche ChessPark
• JQuery
• etc…

These solutions are tried, tested and vetted by greater developers than you or I. Conversely, I have learnt the pain of using immature solutions and unstable versions the hard way. This is one area where I have to work to reconcile the idealist and pragmatist sides of my personality.

Ubiquitous connectivity

Two camps have developed when it comes to connectivity. One says that connectivity is now ubiquitous and local replicas are unnecessary while the other maintains that connectivity will never be reliable and these issues must be addressed before web applications can be accepted as serious desktop application replacements. Google released Gears and later Google Chrome as a response to this risk, correctly recognising that the web browser represents the greatest point of weakness for their entire business. Other emerging technologies such as Apache CouchDB provide alternative solutions to these problems, having taken inspiration not only from IBM Lotus Domino server’s storage model but also its replication abilities. Android and iPhone OS are also jostling to provide a component to this solution, as are the various developers looking to go along for the ride.

Break the rules

Everything above here is rubbish. Break the rules. Be different. Be daring. Be unique.

Speed

“You can build prototypes in the time it takes to have a meeting” - Simon Willison