What if online business model innovation is slowing down?

There is a widely held assumption that new business models will continue to emerge online – that statements like “how will Twitter ever make money?” will look as silly in 10 years as similar statements made 10 years ago about Google look now.

There is no question that, if they wanted to, Twitter could make tens of millions of dollars tomorrow, by, say, running ads or by licensing data feeds.   The big question is whether Twitter and other social media sites will figure out how to make Google-scale money and not just Facebook-scale money.  Google and Facebook get (ballpark) the same number of monthly visits to their sites.  Facebook made hundreds of millions of dollars last year and reportedly lost money.   Google made over $22B last year with huge profit margins.

The optimistic view (which I tend to hold myself) says that where people spend time, money will follow.  If people are spending all their time on Facebook and Twitter, the Proctor and Gamble’s of the world will eventually find an effective way to shift the bulk of their ad spending online.   The tacit assumption in this view is that the next 15 years will see as much business model innovation as the last 15 years.

On the other hand, what if we are mostly done creating big new business models for the web? History suggests that business model innovation is rapid right after the advent of a new medium and then slows down considerably.   If indeed it is slowing down, social media could end up like instant messaging – incredibly popular but basically lousy at monetizing.

The Twitter investment and the decline of venture capital

There has been a lot of talk the past few days about Twitter raising $100M at a $1B valuation.  To understand what is going on from the investor side, you need to know about David Swensen, the man who (inadvertently) destroyed venture capital.

Mr. Swensen manages Yale University’s endowment and is the inventor of the so-called “Yale Model.”  Basically this is a model for people who manage the largest pools of capital in the world – universities, pension funds, wealthy family funds, etc.  The major idea behind the Yale Model is to put significant portions of one’s fund into “alternative asset classes” like venture capital.   Yale had phenomenal returns for many years (until this year, which was a disaster) and thus was copied by fund managers around the world.  This created demand to invest hundreds of billions of dollars into VC funds.  This in turn radically increased competition amongst VCs, thereby driving down their returns (public pension funds like California’s release the returns of their VC investments and they aren’t pretty).

What this means is that there are lots of VCs out there with huge funds and very little chance of getting “carry” (performance fees), since most will have negative returns (and they know it).  So instead they are collecting management fees (typically, 2% of the fund for 10 years, so 20% of the total fund).  They need to justify collecting these fees, which is why if you hang out with VCs you’ll often hear them talk about needing to “put more money to work.”  I would bet that the new investors were the ones arguing for Twitter to raise more and more money, even if it meant a higher valuation.  I’ve seen it happen many times.

The inevitable showdown between Twitter and Twitter apps

People usually think of business competition as occurring between substitutesproducts that serve similar functions for the user.   Famous substitutes include Coke and Pepsi, and Macs and PCs.

In fact, especially in the technology sector, some of the most brutal competition has occurred between complements. Products are complements when they are more valuable because of the existence of one another – e.g. hotdogs and hotdog buns, PCs and operating systems.

There is inherent tension between complements.  If a customer is willing to pay $2 for a hotdog plus bun, the hotdog maker wants buns to be cheaper so he can capture more of the $2, or lower the price of the bundle and thereby increase demand.  (For a great primer on competition between complements, I highly recommend this Joel Spolsky post.  I’ve also been writing about complements, here and here).

Microsoft is famous for destroying companies that offer complementary products, either by bundling complementary apps with Windows (Windows Media Player, MSN Messenger, IE) or aggressively competing head-to-head against the most popular ones (Adobe, Intuit).  The surviving 3rd party apps are usually ones that are too small for Microsoft to care about.  The best (selfish) economic situation for a platform like Windows is lots of tiny complements that have little pricing power but that make the platform itself more valuable.

One of Google’s main complements is the web browser and desktop operating systems, which is why they built and open sourced the Chrome browser and OS.  Google’s other big complement is broadband access – hence their excursions into public Wifi and cellular spectrum.

So what does all of this have to do with Twitter?  At some point, significant (non-VC) money will enter the Twitter ecosystem.  I have no idea whether this is will be by charging consumers, charging businesses users, search advertising, sponsored tweets, licensing the twitter data feed, data from URL shorteners, or something else. But history suggests that where there is so much user engagement, dollars follow.

For the sake of argument, let’s suppose Twitter’s eventual dominant business model is putting ads by search results.   Who gets the revenue when a user is searching on a 3rd party Twitter client?   Even if Twitter gets a portion of revenue from ads on 3rd party apps, there will always be an incentive for them to create their own client app, or to “commodotize” the client app by, say, promoting an open source version.

I’m not saying this will happen in the immediate future.  First, Twitter and a lot of app makers* have raised a lot of money, so aren’t under (much) pressure yet to generate revenues.  Secondly, some of the lucky Twitter apps will get acquired by Twitter.  I think this is what many of their investors are hoping for.  But those that aren’t so lucky will eventually find their biggest competitor to be Twitter itself, not the substitute product they see themselves as competing against today.

* when I say Twitter apps, I mean any product, website, or service that eventually makes money and depends on Twitter’s API.

Notes from brown bag lunch at Betaworks

Lately it feels like a full-blown startup revival is taking place in NYC and betaworks is very much at the center of it.  So I was grateful to be included in a lunch discussion group they held at their offices yesterday.

Some things I left the discussion thinking about:

– In a discussion of the “real time” web (you can’t go to betaworks and not discuss the real-time web!) Anil Dash made the distinction between the value of real time as in the information being recent and the value of real time as in having a shared experience. The distinction strikes me as critical.  Speaking strictly from personal experience, most of the value I get from real time services like Twitter & Facebook falls in the latter category.  Reading my friends’ tweets helps me keep connected with them, the same way bumping into them on the street and exchanging small talk does.  The content isn’t as important as they connection shared and presence felt.

I think Anil’s distinction also explains why Twitter search is sometimes a strange experience.  Besides the (presumably fixable) problems of spam and relevancy ranking, you see a lot of tweets that are fragments of friends bantering.  There’s no context.  The major exception is when a news event happens, since then the related tweets are generally reactions to that event, so the event plus a single tweet provides the full context.

Caterina Fake discussed a few principles for designing successful user generated sites.

Among them:  make sure the minimum unit of work required of user contributions is very small (ideally, something that takes just a few seconds).  You can change something on Wikipedia in seconds, but writing a Google Knol page can take hours.   At Hunch, we think of one of our main product design innovations was to take something inherently large and complex (decision trees) and reduce the minimum unit of work to something small (submitting a result or question).

Another principle we discussed was what we at Hunch call the read-write ratio.  For every page created in Wikipedia (a “write”), there are thousands of millions of instances of people reading that page (“reads”).  The same holds true for YouTube (writes=uploads), Yahoo Answers (writes=questions & answers).  One goal in designing user generated systems is to get a high read-write ratio (for example, by avoiding duplicate writes).

Anyways, it seemed people enjoyed the discussion, since, as Anil pointed out, they weren’t doing much fiddling with their iPhones.