Dec 5, 2013 · 24m · mad
"The Business of Data" Panel // October 2011
Roger Ehrenberg · 7m spoken Matt Turck · 4m spoken Joe Fernandez · 4m spoken Sean Gourley · 4m spoken Event MC · 1m spoken
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Hosted by Matt Turck, 'The Business of Data' panel features venture capitalist Roger Ehrenberg alongside startup founders Joe Fernandez and Sean Gourley discussing how entrepreneurs can monetize, architect, and scale data-centric enterprises. The session provides tactical insights on data infrastructure, business models, team structure, and market defensibility within the big data ecosystem.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 21.2% of the talking time here. How this is scored →
Matt as informed peer 3.2 Guest teaching 3.7 Guest disagreement 1.5 Matt pushing back 1.3
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 18:35 Sean Gourley undercuts Silicon Valley data hype
Sean bluntly reframes mainstream Silicon Valley data engineering as mostly being used to manipulate consumer ad views and shopping behavior.
Hardest push from Matt ▶ 7:27 Matt prompts Roger on contributory business mechanicsMatt steers the conversation past vague startup answers by specifically pressing Roger to detail his theoretical model on data network effects.
Biggest teaching moment ▶ 15:49 Roger reframes the panel's focus on PhD credentialsRoger interrupts the founders' brag contest over PhD counts to explain that real-world data hackers with non-academic backgrounds are often far more effective.
Matt holds his own ▶ 7:27 Matt demonstrates deep familiarity with VC thought leadershipMatt shows substantial domain preparation by citing Roger's specific published theories on contributory data models and non-linear network dynamics.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| The Evolution and New Reality of Data | 4 | 3 | 1 | 2 | Matt opens the panel with sharp historical framing comparing modern data startups to legacy giants like Bloomberg and Axiom. The guests elaborate on real-time web-scale data growth and monetization obstacles while maintaining a friendly, collaborative dynamic. | |
| Data Stack Layers and Network Effects | 5 | 4 | 1 | 2 | Matt demonstrates preparation by bringing up Roger's writing on contributory business models and network effects in data. Roger and Sean detail the technical difficulties of data normalization and architecture across the stack. | |
| Sourcing Data and Finding the Right Problem | 3 | 4 | 2 | 1 | Joe gently corrects Matt's outdated stat on Klout's user base, noting they reached 300 million rather than 100 million. Sean re-anchors the conversation on starting with a core problem to solve rather than hoarding raw data. | |
| Team Structure and the Data Hacker | 2 | 5 | 2 | 1 | Matt prompts the founders about their team PhD counts, leading to discussions about string theorists reducing data dimensionality. Roger interjects to reframe the discussion, arguing that practical non-PhD data hackers are more crucial than academics. | |
| Small Data Value and Consumer Applications | 4 | 3 | 2 | 2 | Matt pushes the panel to explain how smaller e-commerce businesses can leverage data products. Sean offers a slightly cynical reframe that most Valley data tools are merely built to get consumers to buy more stuff or view ads. | |
| Audience Q&A and Panel Conclusion | 1 | 3 | 1 | 0 | An Event MC takes over moderation to ask audience questions, leaving Matt inactive as host. Roger breaks down the structural differences between non-scalable content/agency businesses and scalable software platforms. |
Statements from this episode (13)
Ehrenberg: AWS and modern CPUs eliminate need for big iron architecture
“On the one hand, you've got, you know, things like, you know, EC-II and AWS, which facilitate, you know, the storage of massive amounts of data without your own your own big iron architecture, and at the same time, the CPUs to be able to process that magnitude…”
Klout API monthly volume grew from 100M to over 10B in 2011
“In January, we were doing a hundred million transactions a month on our API from about a thousand partners. In, ah, September, we did over ten billion from 5000 partners.”
Gourley: Customers pay for decision insights, not raw data
“The truth is, if you can do that well, and you can give people a better insight to billion dollar decisions, they'll pay a lot of money for it.”
Ehrenberg: Contributory databases trigger non-linear network effects
“There's a whole new wave of businesses around this notion of, ah, contributory databases, where users will actually put their data in, so they, let's say they contribute X, but then they actually get X plus one back because they receive back intelligence from …”
Ehrenberg: Data cleansing and normalization are valuable, non-commodity skills
“The sheer act of pulling together these disparate data sets and then putting the data in a form where they can play nicely with one another is really, really hard. I mean, that itself is not a commodity. The skills and efforts to do that are very valuable”
Fernandez: Klout refreshes scores daily for 300M people
“When Cloud launched, we had data, we had scores on less than 2000 people when we publicly launched, and now we have scores on three hundred million people that we refresh every day”
Gourley: Quid hires string theorists to perform dimensionality reduction
“For us, one of the big problems is dimensionality reduction, and you want to go from a big, complex world to something that's sort of bite-sized and accessible for humans, and so we employ a lot of physicists, and particularly string theorists, because they sp…”
Fernandez: Data startups can out-recruit tech giants by providing unique data
“Yeah, I think the diversity of skill set is critical, and the good news is if you're doing interesting things with data I know at least for us, like, I feel we have a shot at anybody, whether they're at Facebook or Google or doing, you know, wherever they're d…”
Ehrenberg: Graduate schools fail to teach practical data hacking skills
“You need someone who's applied and skilled and who can hack code and try a bunch of things to actually see signal in data, and that is a very specific competency, and they do not teach that in graduate school.”
Ehrenberg: Querying domain experts often beats large crowdsourced datasets
“Wisdom of crowds and large data don't give you better answers. In fact, it's much more valuable to go to a few extremely knowledgeable people to get a particular answer.”
Ehrenberg: Early marketplace startups like Uber and Airbnb are not super tech-intensive
“Those businesses themselves are not super tech intensive, right? They're just leveraging very basic building blocks”
Ehrenberg: Content creation business models do not scale well
“A business like Veri, where the actual, actually creating content to deliver to customers is a core part of their original value prop, that's, that just does not, that does not scale well.”
Ehrenberg: Most tech startups rely on heavy services initially
“I think almost every tech company that I've been involved with has a heavy service, service component at the beginning, because the bottom line is you're engaging in hand-to-hand combat with that customer, and you're trying to get as much feedback as you possi…”