I watched this in its entirety earlier. If you're familiar with Ries' work on Lean Startup the last year or so, you won't get much new out of it, but it is probably the best single "This is what we do, this is why we do it, this is how you can get started RIGHT THE HECK NOW on it" I've seen yet. It would be a great thing to show someone if you're trying to get, e.g., buy-in from management.

My favorite quote: "60 to 80% of features do not change customer behavior." So painfully, painfully true.

Agreed. I wish Eric didnt repeat so much of what he's already said. Repetition in an age of infinite content availability is unnecessary.

But if scoble had heard his other interviews, i doubt he would have been as surprised or asked those questions. Scoble is great but sometimes like all journalists he can forget to 'know his audience'. And for each of his interviews, he shares the audience with the interviewee.

I appreciate the feedback. I find it hard to balance between introducing a new audience to the material and having something new for people who follow online.

I love this quote: "Repetition in an age of infinite content availability is unnecessary." I'm just wondering how to cope with it; infinite availability is not the same as infinite attention.

Always open to suggestions,

Eric

Whenever you have a thought that merits a good chunk of discussion, give that thought a name. In everything you say after that, don't rephrase your previous discussion, just reference the name—it's a verbal #include. In a textual medium, of course, every name-drop can just be hyperlinked to the previous discussion.

Well, I was taught to:

    1) Tell them what you're going to tell
    2) Tell them
    3) Tell them what you told

Didn't quite take ;-)

Don't repeat yourself to computers. DO repeat yourself to people. http://breckyunits.com/with_computers_dont_repeat_yourself_w...

He's not repeating himself, he's just being Lean!

I like the angle that it's not just ignorance that keeps us away from early customer contact (and therefore facts about product fit) but the perhaps unconscious attraction of fun and lack of constraints. There is a comment directly about that (part 2, 4:30 in): when you are free of the burden of knowing how your product fit is wrong, it is very entertaining to be creating in this blank world where you have only your own cleverness as a design opponent. This seems really dangerous, especially if one of your motivations to start a venture is to not feel so constrained.

He also totally nails the resistance to testing caused by fear of what testing will tell you.

I cost myself a pile of money on Facebook ads this week, trying to do an Easter promotion. That happens sometimes: new initiatives don't pan out, it is inevitable. What wasn't inevitable is after I had evidence of it not working -- and not just a little off, I mean nuke-it-from-orbit-its-the-only-way-to-be-sure monstrously off -- I decided that in spite of the evidence the problem was probably not Facebook users or my ad campaign but my stats code. ("What do you mean I'm paying $30 to sign someone up for the free trial? It costs me twenty six cents sending AdWords traffic to the same page. Clearly this math is incorrect!")

So I spent two hours rewriting my stats code to fix the "bugs" that must have been happening, then let my ads run another day. I might as well have taken $70 out of my pocket and set fire to it.

Well given what you've told us before, didn't the hours you spent fixing 'bugs' cost you several hundred dollars in time, plus opportunity cost?

I think, what patrick meant to say is that, his stats code was correct. It really was costing $30. He just didn't want to believe it at first.

> I might as well have taken $70 out of my pocket and set fire to it.

That was my experience with AdWords a few years ago. Consequently, I haven't clicked on a google ad since.

A nice quote in the context of how Google made "big company" mistakes with Google Lively (part 2, ~9m in): "Most startups don't realize that one the big advantages you have is that you have a pathetically small number of customers"

Part two of the interview: http://www.youtube.com/watch#!v=eubdKflIVoQ

Perhaps I'm overly sensitive to this, but [at least the start] seems to just be making new buzz-words.

key takeaways:

1) judge split test (or new features?) strictly by top 5 metrics of customer behavior: "macro metrics"

2) have engineers only work on the top feature in the queue, and not put it down til it's done; managers may rearrange queue as often as they desire

3) 5 why's: invest improvement time proportional to size of problem, at each of 5 points of "failure"

4) product feedback is about your CUSTOMER, not about you.

Question: Why does the behavior of current customers necessarily predict the behavior of future customers? I understand why it would in the case of IMVU (people aren't paying monthly/one-time fees; they pay based on consumption).. and some other business.. but not, say, for the quintessential web2.0 web app. Can someone please clarify? Don't get me wrong, I think these are killer insights :)

Is there any software already that will transcribe this to text, or should I batch it out as a Mechanical Turk process?

Why exactly do we bother taking advice from loud-mouthed celebrity types like Scoble or Arrington who have NEVER shipped software. Period. No software that rocked any customer's lives. No software that even caused problems.

Following this trend, we will soon be taking software advice from Ashton Kutcher.