52 Bicycles

This map (from xkcd.com) requires no comment:

Steve Poland (a regular at TechCrunch) has a blog called Techquila Shots that is a fun read. He does a drive-by on what he calls the Stupidest Web Feature Ever Created: Tag Clouds:

Tag clouds are the stupidest creation on the web I’ve seen in all the years I’ve been on the web. I don’t use tag clouds — and I can’t imagine anyone that does. A list of unsorted single words, some bigger than others? Does nothing for me — the only words that are ever going to catch my attention in a tag cloud are ones that might swing me into “I’m a male” mode, such as “Jennifer Love Hewitt Naked”. Otherwise, a random assortment of words just distracts me, doesn’t make me focus in on one of them or give me more ideas of what I could be looking at. If I’m looking for something, I do a search for it.

As pointed out in his comments, tag clouds are sorted alphabetically. They are most useful as value-added indices of otherwise disparate content, telling one not only what subjects are covered by a given set of content, but also which of those subjects have a greater intensity of content. They are not infinitely useful, but they are helpful. I use one here on our internal knowledgebase to keep tabs on the subjects we are writing about, partially to make sure we have a balance of documentation among the various areas of integration.

Personally, I find them a good value-add, made moreso when mouseovers reveal the actual number of articles/posts/whatever that are associated with the tag.

We’ve got a Left Coast online business celebrity saying hello on this (very very) very long tail journal, and explaining that my previous post missed his point:

I am not for a second suggesting that “software: the ability to create systems, born out of the heads of bright people, that add value for people and organizations” is going away. My point was only that the world of shrink wrapped software, and increasingly enterprise software, is a dying breed. Software as a Service and internet service models are going to dominate the software landscape going forward. In which case, at scale, scale will matter (is that recursive?).

I’d like to take this opportunity to point people both to Ventureblog and Venturecast, which I avidly consume. I am not a cheerleader for “web 2.0″ the way he is, but he is really smart and articulate about important business-technology issues, and he’s well worth listening to. Especially when he is provocative like he was in the last post, and even more so when when I disagree with his conclusion, because these get my intellectual juices flowing. I hope they do so for you as well.

As for the substance, David makes a point above about enterprise software being a dying breed. This is a broader theme that he and others have discussed in the past, probably a good subject for a post - for I think he is utterly wrong. Thanks for visiting, David!

David Hornik writes about The Softwareless Software Company, in which he essentially claims that the ability to provide a major back-end infrastructure will be the core of a SAAS venture:

I believe that a new era is upon us in the 2000’s. I believe that we have progressed from the “Computerless Computer Company” to the “Softwareless Software Company.” Taking the evolution of computer company one step further, “computer” companies are no longer about selling software, but rather about delivering services. Hosted services have the distinct advantage of meeting all three of the “new” rules suggested in the Computerless Computer Company: 1) they compete purely as a utility; 2) they monopolize the true sources of added value; and 3) they are designed to deliver the greatest possible sophistication with the simplest possible user experience…

In the era of the Softwareless Software Company, in which value is measured by utility, simplicity and reliability, the greatest asset may ultimately be the near infinitely scaling data center. It will certainly be important that the new computer company deliver great utility through its software-delivered service. But the most significant differentiator may ultimately prove to be the capacity to scale with massive demand. And those companies best situated to deliver that scale will be the winners. Thus, it is no surprise that just up and down the river from Microsoft’s new datacenter in Quincy Washington, both Yahoo and Google are contemplating building their own gargantuan datacenters. The Softwareless Software Company may have come full circle from the Computerless Computer Company and be more about hardware and infrastructure than about software after all.

Tim O’Reilly has an interesting corollary in his post, YouTube on MySQL:

Every web 2.0 company is ultimately a database company. Some roll their own, a few, a very few, use proprietary commercial databases. Most are running on MySQL.

I see why Hornik and his ilk (i.e. Venture Capitalists) would glom to this idea, much as he (and others) have to the idea of advertising-based business models: it paints a picture of a capital-intensive business, where money can be put to work, and scale begets differentiation and value with a smaller number of winners. The idea that “software” companies are really data handlers (bolstered by O’Reilly’s comment) means that the ability to scale to serve data will be the key to the future, and so is attractive to an investor.

Were it so. It is possible that there is short-term VC money to be made in the provision of back-end services, but I look at the likes of Amazon or possibly MediaTemple as examples of companies doing that right - providing this kind of infrastructure to other, customer-facing software enterprises. Drew Houston’s DropBox does not depend on an internal ability to provide an “infinitely scaling” infrastructure - he just looks to Amazon S3 to provide that, and he focuses on really exceptional software. The good people at 37Signals also look to Amazon to provide back-end

And if there is a differentiation between front and back end, there is a rapid path to commoditization in this back-end service. The empty vessel model might mitigate front-end risks, but it is easier to duplicate and scoop up. And while a poorly (or hastily) designed service might have a hard time switching back-end infrastructure providers, the very phenomenon of many new customer/consumer-facing startups means that the market remains liquid- the big existing guys do not necessarily win the new business without major price/value competitive pressures, even if their old business is more locked in. In this situation, pricing will flow to commodity levels.

So there is some money to be made right now: in this situation, early entrants can make margin on being there for the leading edge, but late entrants will focus on cramming the costs out of the business. And those who build it all themselves will see the advantage from that early model erode, as their front-end investments in data centers stop paying competitive dividends.

Turn this around for a minute. In 2005, Joe Kraus wrote his last and most influential blog post, It’s A Great Time To Be An Entrepreneur. He gives a number of reasons, but this one is at the top:

Hardware is 100X cheaper
In the 10 years between Excite and JotSpot, hardware has literally become 100X cheaper. It’s two factors – Moore’s law and the rise of Linux as an operating system designed to run on generic hardware. Back in the Excite days, we had to buy proprietary Sun hardware and Sun hard drive arrays. Believe me, none of it was cheap.

Today, we buy generic Intel boxes provided by one of a million different suppliers.

I posit that what he was really seeing was the dropping of infrastructure cost, both in aggregate (hardwardwcosts drop, value per MIP increases) and in piece (through the ability to rent just the amount you need). If this cost is going away, or at least is dropping to commodity levels, then the entrepreneurs for whom this will be a really great time are the ones who build at the higher level.

That’s not to say everyone makes $100 million (or even $50 million) - or even that a few people will, as the market looks less like a winner-take-all.

But a great deal of innovation and margin will be  - however it is delivered - in software: the ability to create systems, born out of the heads of bright people, that add value for people and organizations, building on the components that are out there and inexpensive.

I shamelessly reprint Jitendra Gupta’s chart from How to Build a Profitable Startup by Knowing Your Users Better (which is itself a summary of Jeremy Liew’s post, Three ways to build an online media business to $50m in revenue.

Wow - good analysis.

Rockwell has a great comment on the source piece:

I think the lesson here is that trying to achieve large returns on online media businesses is currently a losing proposition. From what I’ve seen, the trend is towards greater compartmentalization - instead of people hanging out on MySpace, they’re hanging out on sites that are explicitly geared towards their interests. Unfortunately for VCs, most of these sites don’t require much in the way of capital to get started.

Increasing revenue per person you interact with will increase profits, because same-person/same-customer/same-store increases do not involve acquiring more people/customers/stores, so more of the top line flows to the bottom.  This might not be the path to $50MM in revenue, but it might be the best path to maximizing the bottom line and total cash output of the enterprise.

At BarCampBoston2 I attended a neat session on advertising revenue. Out of a number of interesting data points, I heard him make a great suggestion for aspiring entrepreneurs who want to monetize through advertising: build your own ads.

Here’s the plan: join an affiliate program (e.g. Amazon Associates plus API) and make your own ads that point to products. You make money on the sales that are actually driven, creating a short-term revenue opportunity.

And the transactions you drive create metrics that you can show to potential advertisers/sponsors as to what they will get for participating on your site. Your profits will drive advertiser acquisition, which lets you get out of the business of creating ads yourself, and move more resource into the site and managing advertiser relationships.

Or, if the ads you build for yourself are profitable enough, you have an alternative business model, even if advertisers are banging down your door!

This fixes the chicken-egg problem for advertisers, and shortens the path to profitability at the same time.

I think this model is particularly important for sites/services that fall out of the adwords mold - dynamic pages that are by their nature harder to index by the likes of adwords (which are otherwise the easy drag-and-drop approach, though very low revenue on average).  

I apologize for not recalling the name of the presenter, but if he or a fellow attendee lets me know, I will update.

Shimon Rura, Mike Walsh, Susan “Sooz” Kaup, Jeff Potter and a host of others made this past weekend’s BarCamp a great success. Smart durations (45 minutes for sessions, 15 minutes for walkabout), great hallway conversations, and I had the pleasure of running a session called “Drop Your Pants” (subtitled “Not Literally). At least 150 attendees on day one. And all this in the wake of a major snowstorm!

I have some notes of substantive lessons I got from the sessions which I hope to post, but for now let me just say congratulations to those who made this happen by force of will.

Glad to have taken part both in advance and during, though I had to bail on much of the second day on account of my “real work.” I look forward to more community events - DevHouse 2? BarCamp 3?

And I am reminded to get my own event, the sales & marketing breakfast, back into gear!

Brad Feld’s Ask the VC has a a great post, How Do I Create Projections For My New Web Service?

The more interesting piece is the user adoption – how it’s going to work, what is going to drive it, and how this links back to the underlying costs. 

Recognize that the costs are not just hardware / software / bandwidth / hosting costs.  These are going to relatively minor in most cases compared to your “user / customer acquisition cost.”  How are you going to get new users?  If your answer is “they’ll just come”, that’s probably wrong (although not always - you might have one of the insanely effective web services that spreads rapidly by word of mouth and nothing else.)  So – assume you are going to have to spend some money somewhere getting users – what do you think that actually looks like.

This is true for more than just web services. Geoffrey Moore says in Dealing with Darwin that particularly in a volume operation, innovating the business model is crucial to success. (He contrasts this with a complex systems environment, where s 10x improvement from the technology is the focus of innovation.) Cost of customer acquisition (COCA) is usually the most crucial lever to pull in the business model, because that is how dollars come in. Especially when the back-end is built on commodities, the ability to generate top-line either at lower cost or at a higher effectiveness is going to be critical.

For web services that essentially want to operate like media properties, they have two issues: getting readership, and selling inventory. So 2/3 of the business is really about getting customers (either indirect, as readers, or direct, as advertisers).

For those of us in the ”we sell stuff” space, it’s a little more straightforward, because rather than matching advertisers to readers/clickers, we just need paying customers. But that doesn’t necessarily mean easier, and it is snowboarding to the media model’s skiing - they both get you down the hill, but require very different skill sets.

Either way, the great entrepreneurs of history were all great marketers - they put customers together with their products. Show the VC (or anyone else looking to invest time or money) that you can do that!

The inestimable Mike Walsh has Photos on Flickr of the posters so far.

It’s less than two weeks off, sign up! And please, if you can afford to, make a small donation to help keep it free for those who would otherwise not be able to attend.

Rob Finn at Edison Ventures discusses Desktop Applications:

While not a new concept, the desktop is also becoming more web-connected. More and more desktop apps are talking to software and to web content that are consumed through a browser. Richard MacManus describes this as the webification of the desktop. Example webified applications include iTunes, Qumana (blogging), Quintura (search), Webaroo (offline search), Picasaweb (photo organizer), Smilebox (greeting cards), Weatherbug (weather), Omnidrive (storage), Bubbles (communication tools), Trillian, Pronto (comparison shopping), Invisible CRM, Slide, FeedDemon, and Visokio. There is a long list of such desktop apps thanks to web services, APIs, open source communities.

Some of this recent desktop centric growth has also been evangelized by the portability of content due to consumer demand for personalization and decentralization of web content. Exhibit A: Widgets on sites like Myspace on the desktop (e.g. Konfabulator/Yahoo and Vista’s Sidebar).

The article referenced above, Webified Desktop Apps vs Browser-based Apps, is actually by Ebrahim Ezzy (but published on McManus’ blog as an opposing viewpoint) and contains the following insight as well:

I rely on various web applications to create documents, presentations, spreadsheets; share images, videos, data; manage and organize tasks, projects and life. But I still believe the future of computing isn’t entirely web-based. It’s necessary to have the desktop as the pivotal point, because the power of the desktop is important for a rich user experience - and will be, for a very long time to come.

What we require then are smart, webified, internet deployable desktop applications - that can reliably store data, serve it robustly, and interact with both remote and local databases. This connected model will ensure that applications will function in both online and offline states - for a seamless, uninterrupted experience. 

I have written about this general idea under the rubric of “tech cosmopolitanism” before, but the point these guys make specifically is worth noting - that the experience of a user who is intermittently connected to back-end systems is going to be better when that experience is closer to the user.

That said, there are best practices from web-based services that all should keep in mind:

1. Updates: software should self-repair relatively silently and easily.
2. Collaboration: connectivity should help people work together, both in real time and in a stateless mode.
3. Redundancy: data should be protected, and having it both in the cloud and on the desk/laptop/handheld is valuable. This creates a synchronization complexity, which we can deal with another day.

Am still thinking about these issues.