Marcelo Calbucci

 

    Michael Arrington just made a mistake in claiming that Outlook 2007 is the most popular feed reader of subscribers of TechCrunch.

 

    FeedBurner has very limited ways to measure subscribers and most of those are significantly flawed. On this case in particular I'll only pick on the Outlook 2007 issue.

 

    I believe the Outlook number is probably 3 or 4 times smaller than reported by that graphic. So, instead of having 519,904 subscribers from Outlook 2007, TechCrunch probably has only 130,000 or so, making it the 4th most popular source of subscribers.

 

    My case is based on how FeedBurner can compute unique readers. On the case of Outlook 2007 (or any email app), you are stuck between a rock and a hard place. Outlook 2007 doesn't send cookies. So, there is no way to do a similar unique count like Google Analytics does. The only number FeedBurner could use is the IP address.

 

    Now think about your typical TechCrunch reader (no, they are not developers). They are your regular corporate Joe. And what Joe the Employee uses as a device to carry his work? A notebook! You know, those little computers you carry around, sometimes called a laptop.

 

    Now, this laptop will know at least 2 IP address for sure. When it's at the office and when it's at home. But how about when you are having coffee at a Starbuck or Tully's. Well, that's another IP address. What about when you are at the airport, hotel or library... Another 3 IP addresses.

 

    I have no data to back it up, but it's fairly reasonable to assume a laptop sees 3-4 or even more IP addresses per week.

 

    This problem doesn't apply to Google Reader, BlogLines or any other online reader. It might not apply to other offline readers that support cookies as well, but who knows.

 

    The bottom line is that you should never trust in a piece of data if you can't validate the assumptions behind it.

   

 

   

 

 

 

    I looks like is the perfect time to trim the fat and to get rid of the unwanted from your company since you can always blame the economy. I don't mean to pick on CarDomain exclusively (I should pick on more startups), but the real reason of the layoffs is not the downturn of the economy, although that might have been the tipping point.

 

    The real reason is a failure to maintain traction with the customer base. The numbers don't lie:

 

 

 

I’ll spare you the long story of my life and how I’ve got to be above average when detecting BS on projections and assumptions (financial or not). But there are a few rules of thumbs that you can use to quickly validate some assumptions on your own projections or on someone else’s projection in case you are an investor.

 

Each industry is significantly different from the other as to assumptions not being easily transposed. So, mostly of what I talk here only applies Web-based consumer services. On the same note, unless two companies are on the exact same market, have the exact same positioning and exact same product, it’s hard to say that everything you know about company A applies to company B. So, whatever you do, you have to figure out how to validate assumptions on a case-by-case basis, however, there is the bullshit line. This is line that anything above it (or below it) is guaranteed to be BS.

 

The most easy BS threshold to detect is the growth of X, where X can be users, page views, widgets, etc. If someone tells they expect a 30% month-over-month growth for the next 2 years, you just detected the BS line. Why? Simple math. A 30% growth M-o-M for 2 years would mean a 41,654% growth in 2 years. That’s like saying you have 100,000 users today and in two years you’ll have 41 million users. Of course, the most used argument of a BS projection is to talk about examples like YouTube or Facebook.

 

Here are some simple direct numbers (BS lines) that you can use to validate your projections or your portfolio company projections that I use myself (see some exceptions after the list):

 

1. > $10 CPM (at niche): On the context of web advertising, even highly targeted ads are very unlikely to reach above $10 CPM. You’ll hear a lot about how the price of ads will go up, how profiling data on users will help drive up the price of ads, yada, yada. Above $10 CPM is BS to me.
2. > $4 CPM (at generic): On generic audiences website, it’s nearly impossible to get more than $2 CPM, and most sites are below $1 CPM. There is a step function in terms of volume and price because you can negotiate better rates (even w/ Google) if your volume gets really big (> 20 MM PV), but still, you won’t be getting $4 CPM for a photo-sharing site or any other kind of generic service.
3. > 20% Conversion Rate: Defining conversion rate as the number of people that sign up to your service divided by the number of people that visited the website cannot be above 20% and more likely will be between 5 and 10%. I can give you many reasons why this is true (another blog post), but if you think you’ll be getting more than 20% conversion at a high volume, stop.
4. > 5% Conversion to Premium: If you are on a “freemium” model it’s nearly impossible to get past the 5% conversion to premium of sign ups (notice: that’s 5% of the sign ups, not 5% of the visits).
5. < 50% Churn of New Users: If you are offering a free service, a free-trial or a freemium (free w/ ads or paid), you should expect a huge churn of new users. People might just be evaluating your service, they even might have been mislead into believing your service is a duck while it’s just a chicken by your hyperbolic marketing material. Either way, if your model projects a churn of new user of less than 50% I call it BS.
6. < 3% Churn of Active Users: The biggest problem with projections of churn of active users is not low numbers, it’s that is not even modeled. People think once they get a user, they get a user for life.
7. > 10 visits/UU/month: Really? You think you can get each user to visit 10 times per month? On *average*?!! Yikes.
8. > 30 PVs/visit: I love web stats and this is one of the most interesting ones. 30 PV/visit is impossible (see exceptions), and even that number is ridiculously high. More likely a site will have 5 PV/visit on average, anything above 10 PV/visit is suspicious, above 30 is BS. And the more your site’s traffic depends on referrals from search engine, the closer it gets to 1 PV/visit.
9. $0 Customer Acquisition Cost: Ah, what the heck. Your site is super-duper viral and signing up just 1 user will bring another 100, that will bring another 10,000, that will bring another 1,000,000… Yay!

 

The list above are the most obvious data points, and probably most used, on Excel spreadsheets all over the word trying to justify a business model. Feel free to complement or disagree with my list on the comments.

 

Exceptions: Mostly you can apply these rules to any web-based consumer service, except communication services, like email, IM, chat, forums, etc., because they have very high number of return visitors, high number of PV per visit, etc. On the flip-side, the CPMs for those service take a nose dive.

 

 

    I'm a big believer that we need more blogs in Seattle. There is too much knowledge living in the underwords of coffee shops, forums and distribution lists and not enough being surfaced. Because of that, Carolynn Duncan, Ian Lurie, Scot Herrick and I wrote a 4-piece blog spot, each one on his/her own blog.

 

    If you are considering having a blog or if you have one but it's stale, go ahead and read some tips on how to get the most out of it:

 

    Carolynn Duncan: How to Set Up A Basic WordPress Blog

 

    Scot Herrick: Marketing Your Career Through Blogging

 

    Ian Lurie: 15 Rules for Business Blogging

 

    Marcelo Calbucci: Finding Your Blogging Voice

 

    Maybe is the sign of times, maybe not, but yesterday's Entrepreneur University was smaller than 2007. They claimed to have 250 people attending the event, so, how many paid for it?

 

    Let's see...

 

    There was 22 speakers. None of them paid, of course. The EU Comission was mostly there, and that's 12 people. As far as I know, each speaker got to invite up to 2 people to attend, that's 44 free tickets. I'd think the EU Comission also got to invite at least 1 person each, and that makes another 12. Then there was the TechFlash and Seattle 2.0 ticket give away, another 6 free tickets (call it the marketing strategy) and finally, there was the sponsors that got free tickets as well, but those mostly overlap with the speakers and the EU comission, so let's not count any.

 

    That's for a total of 96 free tickets, which leaves us with 154 paid attendees.

 

    I think that sucks because an event like this is of great value for new entrepreneurs, business students and people around startups. But it hasn't had the best marketing, strategy or execution, and it might go away if it doesn't get the right kind of traction.

 

    I have this belief that if a conference sucks on one year, next year attendance will drop dramatically. This is what happened to WTIA Investment Forum this year, it had so few people. It wasn't present at the WTIA last year, but I heard it wasn't good.

 

    Just to be clear, EU didn't suck this year, but it didn't hit out of the park either.

 

    Yesterday I spent the whole day on NWEN's Entrepreneur University. I learned many things, but the day started on the wrong foot when I got there at 7AM and it actually started at 7:30AM. Yes, I do want 30 more minutes of sleep.

 

    The opening keynote was by Lynn Taylor (of Taylor Protocols) and it was just bad (see bad advices for entrepreneurs). You expect the opening act to make people joyful and hopeful for the day, but he was very much focused on how companies screw up and how you should fire 1/3 of your employees...

 

    Then I saw Kelly Smith (Curious Office/Pressplane) talking about "surrounding yourself with the right people". Awesome talk. The time went really fast and people could probably listen to him talk for 2 hours.

 

    I didn't see Mike O'Donnell (iCopyright) talk, but I saw him talk twice before and it's too easy to disagree with him and too easy to think he's giving bad advice (which he is, IMO). A friend told me the talk was bad (I told you so).

 

    Kabir Shahani (Appature) also gave a good and unpretentious talk on competition and when to worry and when to not worry about it. He gets it. One of my tweets yesterday said "talk about your business but don't invite competition". The context was some entrepreneurs that really like to boost about their profits, margins, easy sale cycle, etc. Those are bad things to talk publicly because you are inviting others to take a closer look at your industry.

 

    The closing keynote was from Clark Kokich of Razorfish. It was very good. It should have been the opening keynote. Razorfish went through a lot and made it.

 

 

Lesson #1: How to give the best presentation

Four of the presentations that I attended engaged the audience more than the others. I don't know if there is a correlation, but all 4 (Kelly Smith, Kabir Shahani, Todd Humphrey and Clark Kokich) repeated multiple times on how those things worked for them and might not necessarily work for you. I really hate when someone giving a presentation thinks he owns the truth. When you say something "this is my experience and it might not apply to you" you disarm the audience. No one will be trying to (mentally or verbally) challenge you. Note to self: Be humble when giving a talk.

 

 

Lesson #2: This is my last EU*

This is the fourth time I attended the event and it'll probably be my last. Like with books, I believe you can always learn something new at a conference. However, the value I've got from this event in relation to the 12 hours it cost me doesn't make it worth coming back (BTW, I've got in for free thanks to Todd Humphrey that gave me one of his free passes). That said, I probably will go back to schmooze with friends, entrepreneurs and service providers next year, but I won't stay for the whole day.

 

 

    I write this on my twitter:


    And within minutes had a dozen people asking me what did he say.

    Well, it's not a single thing, but a class of things. Lynn Taylor of Taylor Protocols was the speaker and he talked a lot about hiring/firing, but he used the context of medium and large size companies. That absolutely don't apply to startups.

    The one thing that didn't go well with me was when he said that "...don't let people define their own role in the company... that's a myth and it doesn't work..."

    In a tech startup environment that's not true. I will not say hire a person and then ask him what he wants to do, but you have to have flexibility and give people independence to find where they will contribute the highest value to the company.

    If you want to follow my thoughts through out the day on NWEN Entrepreneur University, you can follow me on Twitter: @calbucci

    I'll do some blogging as well if time/battery-life permits.