jump to navigation

Why Color Is More Than “Yet Another Photo-Sharing App” March 28, 2011

Posted by David Card in Uncategorized.
Tags: , , , , , , , , , , , , , , , , ,
add a comment

Much of last week’s buzz surrounding the launch of Color was justifiably skeptical. The startup, after all, raised $41 million to enter a crowded space without a business model or customers, and many wonder whether the world really needs another mobile photo-sharing app. But two components of Color’s vision — implicit networks (connections created without user effort) and place/time tagging — extend far beyond photo-sharing, and make the company worth watching as a potential indicator of social media and data-mining trends.

The Color app for iPhones and Android lets users share photos in real time with other nearby photo-snappers. The sharing network is determined by proximity rather than by a user explicitly specifying who his friends are. Users are anonymous and all content is public.

Early reviews are pretty negative. Om writes that Color is attracting more attention from pundits than users because the app may not deliver obvious fun or utility. Matthew Ingram wonders if the big funding bet is on Color’s all-star team — which includes Bill Nguyen (Lala), Peter Pham (BillShrink) and former LinkedIn chief scientist DJ Patil — rather than its product or ideas.

But some of those ideas matter.

Implicit Networks

Angel investor and Hunch co-founder Chris Dixon says he’s intrigued by Color because it is pushing the envelope on implicit social graphs. Color’s implicit networks aren’t specified by users, but rather are based on underlying contexts like geography or shared interests. I’ve written before about context-based social networks, and how Facebook Groups is struggling to deliver them. Peter Yared, a VP at WebTrends, writes that Facebook is also experimenting with implicit neworks of friends.

If Color builds on its implict network concept it could deliver instant groups of friends for different occasions or interests, and expose recommendations based on common tastes. Marketers could target advertising or offers within a Color network to real-time groups around an event or location, or aimed by shared interests.

Place and Time Data

Search pundit John Battelle goes a little overboard on how Color could push augmented reality. But he’s right about the importance of geo-tagged data. In a presentation last week at GigaOM’s Structure Big Data 2011 conference, IBM Distinguished Engineer Jeff Jonas showed how adding place and time to data objects can power big data analysis, predicting a person’s likelihood of being at a give location with astounding accuracy, and assisting in identity management. Again, if Color is a leader in gathering this data, it could build out a powerful — yet still privacy-protected — targeted advertising network.

Business Model to Come?

Color chief Nguyen says the company is really about data-mining rather than photo-sharing. He says combining place and time data with implicit networks can help services or marketers parse the difference between entertainment and work activities. That information will affect the elasticity of Color’s networks — how broadly it expands or contracts its sharing range — and power its algorithms for ranking photos and, presumably, other content or advertising elements.

Nguyen also talks about a future news API that could spawn a curated news app for journalists. He describes a pretty dumb restaurant service that would help waitstaff know customers’ first names and interests. Before he sold Lala to Apple, reportedly for $85 million, Nguyen took the service through at least three different business models. Lala started as a CD trading service, morphed to a digital music locker, and then offered Web songs with perpetual streaming rights for ten cents each. With its talent and cash hoard, there’s no doubt Color will evolve as well.

Question of the week

Is Color more than just another photo-sharing app?
Advertisements

Is Quora Worth the Hype? January 17, 2011

Posted by David Card in Uncategorized.
Tags: , , , , , , ,
2 comments

Around the end of the year, the hype surrounding Quora kicked into overdrive. The Q-and-A site, founded by Charlie Cheever and ex-Facebook talent Adam D’Angelo, first raised eyebrows with a round of financing last March that valued it at $86 million. When it went into public beta last summer, the tech and business press got excited, and lately it’s being called the savior of search and the next Facebook. But is Quora worth all the fuss?

Quora enables anyone to pose and answer questions, and users can “follow” other users and topics. Much of the site’s charm comes from its audience: Famous and influential personalities from tech and VC regularly ask and answer questions. Bloggers and mainstream reporters are using Quora for story content and leads. And as über-blogger Roger Scoble pointed out, Quora has cleverly adopted key social media and real-time innovations to the Q-and-A space.

Search Replacement?

Arguably, Quora’s biggest innovation is “crowd curation.” Lately, the same blogger community that has taken to Quora has been complaining about Google. Google search results are cluttered with spam and links and low-quality posts from content farms like Demand Media, the bloggers charge. The solution? Relevancy enforced by human beings rather than algorithms. But hiring editors doesn’t scale as well as writing software, that is, unless you can crowdsource those editors for free, which is exactly what Quora is doing.

But Quora is also wisely allowing Google to index its content, and practicing SEO well enough that Quora answers are starting to show up in Google results. Google’s own PageRank algorithm has always harnessed some wisdom from the crowd by analyzing link popularity. To add relevance and force out spam, Google engineers are smart enough to create or license other indicators of authority and influence — whether that means baking in to its algorithms something like a Klout Twitter authority score or ratings derived from professional content databases.

Differentiating from the Crowd

Quora is far from alone in the Q-and-A space. Facebook’s barely launched Questions appears aimed at generating status update activity and real-time responses. It feels more like personal expression than knowledge management, and thus may be a bigger threat to a company like Formspring, whose Q-and-A pages Om likened to blog comments without the blog. LinkedIn Answers is geared to its professional audience, but doesn’t have much traction. Yahoo Answers, the granddaddy of them all, generates lots of page views but little in the way of revenues. Yahoo Answers are often cute or funny, rather than useful. To avoid a similar fate, Quora is scrambling — so far quite successfully — to impose protocols on its users for asking, answering, editing and tagging questions and answers in order to preserve their quality and add structure to the Quora data folksonomy.

But for all the talk of its usage “exploding,” Quora’s community and traffic is tiny. It caused a minor scandal over its self-reported registered user count, which remains below half a million. Quora’s traffic is half the size of Formspring’s and dwarfed by Yahoo Answers. True, Quora could grow, and probably maintain at least Wikipedia-like quality, but it has a long way to go.

Of course, Quora has no revenue streams. Advertising or job boards a la programmer site Stack Overflow require scale. Quora’s evolution might take the following paths:

  • Expert network. Gerson Lehrman Group, for instance, has built a multi-hundred million dollar business by brokering paid one-to-one communications between experts and questioners. So far, Quora depends on altruism and self-promotion to incentivize its answers.
  • Interest graph supplier. As with any robust social medium, Quora could collect — and license — information on personal interests. But it still needs scale to build privacy-secure personal info into anonymized segments useful for marketers.
  • Magazine. Quora could indeed survive as an independent, engaging content destination — what it is now — but its model would be that of an online magazine. It could sell brand-oriented or contextually related advertising aimed at a small, but desirable techie audience. But does Quora really want to be Salon?

Question of the week

Is all the hype surrounding Quora merited?

Can Mining and Filtering Monetize NewNet? December 20, 2010

Posted by David Card in Uncategorized.
Tags: , , , , ,
2 comments

One of the keys to monetizing NewNet technologies like real-time feeds and social media will be harnessing the massive amounts of data they create. In recent weeks, there have been a handful of announcements illustrating creative ways of using this data to enhance products, often via recommendations. But most of them have not shown clear revenue strategies.

What the initiatives have in common is their use of information from feeds or social graphs. Foursquare posted a job listing for a data scientist to assist in mining its own data to enhance product features, but there may be more opportunities — and competitive differentiation — in combining data sources. The recent initiatives display at least two ways of tapping those veins:

  • Mining happens behind the scenes. Companies license and/or utilize APIs to extract information and apply it to applications and services to aid in targeted marketing, aid personalization, or create entirely new products.
  • Filtering is more visible to the end customer. Like mining, filtering adds relevance, but is generally controlled by the user.

Who’s Doing It, and How

Mining NewNet data from multiple sources may require the resources of a company with a big, established business —rather than a startup — for deployment if not development. Social media buzz-monitoring companies like Cymfony (part of ad agency giant WPP) and Buzzmetrics (part of Nielsen) sold themselves to ad agencies and market research firms. Because changing an established user interface is a tricky thing, innovations in filtering multiple data sources will likely originate at startups.

Examples of each include:

  • Wowd filters Facebook’s feed. It applies its own algorithms to Facebook APIs to automatically create natural groups of a user’s friends by analyzing users relationships to each other and posted info. Wowd allows the user to filter by time, topic, and trends.
  • Clicker, that makes an Internet video guide, is one of the few companies that pulls in Facebook data via “Instant Personalization.” It maps a user’s self-professed Likes into genres and topics to produce recommendations it shows alongside editorial suggestions, friends’ viewing, and popularity.
  • Google mined its own traffic and embedded content for YouTube Trends, and tweaked its social search presentation. Microsoft appears to be using Facebook data in its basic Bing results, as well as offering an alternative social view. MTV Networks created a new music discovery space by mining social data.

But Payoff Remains a Challenge

A simple ad revenue model for a site or app that filters a Twitter or Facebook feed produces pretty small dollars. I used traffic data from Compete, “visits” as a proxy for page views, and assumed a low-cost ad (CPM of fifty cents to a dollar). If a filter company showed a single, relatively untargeted ad per page, and siphoned of 10 percent of Twitter’s site traffic, it could generate yearly ad sales that would be measured in the tens of thousands of dollars to perhaps half a million. If the company managed to appeal to one percent of Facebook’s US users, the figures are in the same ballpark.

My model is very simple, and very conservative. If Facebook is really approaching $2 billion in revenues, it generates roughly $2 to $3 per user per year. Google is more efficient: it gets $25 per user/year. To get to multi-million dollar yearly ad sales, a filtering company would have to attract a million users, preferably of a distinct demographic, job description or sphere of interest. That would enable it to offer a better-targeted audience and a richer palette of ads and marketing opportunities to advertisers, and charge a CPM in the $3-plus range.

Active personalization — convincing a user to set up a customized experience — is tough. Yahoo never got more than 15 to 20 percent of its users to build out a My Yahoo page. Those who did were its most valuable users, the ones that used multiple Yahoo products and converted to paid services. The passive personalization enabled by mining could indirectly contribute to customer monetization via retention and increased usage frequency.

Question of the week

How can you make money off of social media and real-time data?