Cogblog

VentureHacks writes a great post: How to deal with lawyers in a start-up.  Now, some people might argue, “that is pretty tough, what incentive will they really have to do work?”, but the obvious answer, that VentureHacks points out, is that they weren’t really feeling incented anyway.  It was a want more than a need and that is a really, really expensive way to make someone feel good for 10 seconds while, in the big scheme of things, affecting very little.

If they think they need the equity to feel engaged, then go find a new lawyer with less emotional baggage.

One of the things I talk about when people start to talk bubble these days is the structure of the bubble and how different it is than the last bubble.  It seems to me like during the last bubble, many VCs were able to get their investments public and cash out, leaving it to public investors to take a bath.  This time around, M&A has slowed a bit and very few companies are going public.  There will be a lot of funds that fail to come out of the first decade of the 21st century showing good returns.

AdReady has now raised $12 million.  Can they really find a buyer that will give those investors 10x returns?

What is interesting about AdReady is that they do not own a network.  They are really a front-end tool for media, but it kind of assumes that all inventory is the same and that performance differences tend to be more about creative quality.  At Advertising.com, I would say that we tend to skew the other way.  We see performance differences in creative, but we also recognize that all inventory is not created equal.

For them to raise $10m, they had to convince Bain Capital and Khosla and others (smart money) that ad inventory is a commodity.  To me, that is a hard sell.  But then, the AdReady team is great and could sell ice to eskimos.   I love those guys, I just think these deals are hard to do.  Probably I think too small.

I did a quick analysis of the last 30 years of my High School and students/former students participation in Facebook. Here is the graph:

The accumulated 70’s and 80’s graduates only amount to 80 people.

The 90’s graduates are less than 500 strong.

The 21st century graduates are almost 2000 and the decade is not even over. There is no single inflection point, but 97 had 46 people and then 98 had 88. 99′ had 93 people and 2000 had 136. Those are the only years where, on a percentage basis, “interesting” growth happened (by that, I mean absolute growth of more than 15 people (so I rule out the years when it went from 3 to 9 or 1 to 5) and percentage growth greater than 30% (91% and 46% respectively))

So this population skews pretty young. We are talking about high school here, so ~75% of the people on Facebook that attended my high school are 25 or under.

My year (Class of 91 – wooooo) had a graduating class of more than 700 and only 27 are on Facebook. Facebook’s efforts to target an older population and business crowd are just going ok, I would say.

It is a shame I don’t have data about when they joined so I could look at how this curve is changing over time. Maybe I should check back in a month or quarter.

Update: Thomas makes a great point that you could compare this to demo data and actually try to build a model of when people stop identifying with their high school – and this might be a better/more accurate thing to do.  He points to a TechCrunch link that shows some basic demo data for Facebook that indicates that ~50% of Facebook members are over 25.   Pretty interesting!  I still want to go back and run these numbers in a month or so and see how they change.

Cogmap is starting down the path of introducing collaborative filtering. Soon, very soon, we will start to tell you maps that you might like, based on maps you have looked at or edited in the past!

We are attempting to use Vogoo, which is a Slope One engine perfect for what we are trying to do. Unfortunately, the documentation is a bit weak. Not terrible, but not great either. I wanted to document a few of my questions and some of the things I did for the next lucky guy!

Set-up

The tutorial tells you how to set ratings, but the installation guide tells you that you should do other things also when you are going to set ratings.  And there are no examples.  Sad face.

I have sat down and tried to implement all of the things they explain at different points and here is what I ended up with (An example, YAY!):

define('VOGOO_DIR','<PATH TO VOGOO DIR>');

include("<PATH TO VOGOO.PHP>");

$vogoo_check = $vogoo->connected;

if($vogoo_check)

{  $rate = $vogoo->automatic_rating($userid,$chartid);

} else { generate_error(0,"Vogoo DB not connecting"); }

Now, I used the automatic rating function because I wanted every logged in view to count as a vote.  Then I use subscribing (which happens automatically on a chart edit), as a “purchase” or indication of love++ for a chart.

Questions

It sure sounds like you can’t do both user filtering and item filtering at the same time.  For a site like this – kind of a social network wrapped around an application – that is kind of bleh.  Ideally I want to say here are the kind of maps you like and then also have a “here are users who dig the same maps”.

Despite my bashing of the documentation, when I sent an email with a question, I got a response 24 hours later…. twice!  So that’s pretty good support and I was using the free version.   Of course, he failed to find my bugs, which in retrospect were dumb bugs.  Hmmmmmmmmmm.

Big Conclusion 

So now we are up and gathering data.  My plan is to run it for a month or so and then look at what kind of data set I have.  According to the Vogoo manual, you have to run a separate php program to compute the links between each item, so I assume (although once again the documentation doesn’t discuss it), that I need a fairly good sized sample.  With more than 3,000 maps, I need a little time to develop and expose relationships.  Once I see what the data looks like, I will begin to figure out how to crowd my UI and add that in.

Now, I am not the world’s greatest programmer by any means. In fact, many would testify that I may be the world’s worst. So take all of my advice with a grain of salt, but enjoy!

– brent

TechCrunch, Read/WriteWeb, GigaOm and others are having a contest.  I figure if everyone that reads this types in “Cogmap”, then I might get nominated.  How many start-ups are there that have raised no money?

I was this close to nominating myself for best business model.  That’s a lock!

Great press release today from Advertising.com announcing their Facebook Application Ad Network.

Federated Media made an announcement the other day, AdBrite has made an announcement, also.

Sounds like tough times ahead for Cubics, Lookery, and Social Media. It is hard to imagine they have the scale to support strong RPMs on a significant amount of inventory. Large network’s strong marketplace of advertisers should position them extremely well to take share in this market. Once they start offering higher RPMs on more impressions, that will make it difficult for small networks to compete.

Popular applications that generate a lot of page views will not be able to get the same eCPM by doing a boutique deal on a small amount of inventory that they can getting significant fill rates from a large network. As the big guys commoditize the inventory, the ability to sell this in an interesting way for small networks should get tough.

Fledgling networks that started with a strong advertiser message by pricing on CPA (installs) like Social Media and Slide will find that the pricing they offer publishers rapidly becomes uncompetitive. The big guys have some of the best CPA deal structures in the world, they will simply have more competitive ways to fill inventory on an eCPM basis.

Crunch, crunch, crunch.

Stylehive used to be a go-to site for me when I felt like looking for home decorations or clothes for my wife. Unfortunately, I feel like that is no longer true. The front page now contains so much stuff, compressed so tightly, that it is no longer easy or fun to breeze through. It has been made super-efficient for the Stylehive power user. Also, they have worked so hard to incorporate text links into product pages that the user experience is virtually destroyed every time I look at a product.

Here is a screenshot:

Correct me if I am wrong, but half this page is ads and I have to scroll to see the full-size image of the picture I just clicked on. That’s awful! I can’t imagine that eHub is as proud to show this as they were the initial site design.

Why would Stylehive corrupt their own site? Well, they have taken $4m in venture capital. Probably hoping some revenue will offset anemic page view growth – here is the comScore data:

What’s crazy is that they are seeing strong unique visitor growth. comScore data:

I would maintain, because it selfishly serves my purposes and I know nothing about what I am talking about, that the page view growth is flat in the face of growth of uniques because visitors are coming to this site with good buzz and then are driven away by the morass of advertising.

With $4m in capital, they need to get to breakeven pretty quickly or start to show the kind of revenue ramp that might make people pay more than $50m for this company. Ouch. We have talked a lot about how easy it is to build a company with a little capital today. The downside of raising a little capital (and in this instance we will consider $4m little) is that you can never run out of cash. So if you aren’t getting to profitability and the looming threat of bigger and bigger raises with what is, in many instances, essentially niche sites that probably won’t justify 9 figure valuations, the real concern that a company has an unsustainable business model emerges.

I go to Stylehive less and less and today I could not figure out why I should ever go back. They are pushing community features out the wazoo, but I am not interested in being part of the community, I just want to find out what the hot new stuff is. Before, I knew Stylehive would give it to me. Now, I feel like Stylehive tells me about communities I don’t care about then tries to get me to click an ad.

I am reaching here, but it is how I feel. It is a shame they make me feel this way.

I recently tracked some of the events that unfolded around Sean Taylor’s shooting death by following Wikipedia.

When I first went to Wikipedia on November 26th to find out the details, after reading about it on CNN, Wikipedia reported that he had died.  Sadly, I cannot find the exact version I was looking at because it said after the death date: “Trust me admins, I know”, which I found hilarious.

In a tribute to the power of Wiki’s, the next revision of the article comes less than a minute later and deletes this bad data.

I spent a little time on SEO on the site recently and saw a definite lift in traffic. Now, I am no SEO expert and at some level I would say that the site is more or less what it is. There are not a ton of things I can/should/want to do regarding changing H1’s, adding anchor text, blah blah blah. Cogmap is an application. It does what it does. I don’t want to add a bunch of unnecessary verbiage to improve search results, etc.

But, what I did has resulted in better traffic! 30%+ better with just a few changes. I was talking with a friend about some of the things we did and some of the things I learned as we went through the process and he was interested and surprised, so I said, “Man I should blog that!”

Here it is!

First, PageRank is huge, but checking PageRank is kind of bogus. See Matt Cutt’s commentary.

PageRank from the Google toolbar is some static number that gets dumped out quarterly, while your actual pagerank is constantly being computed. So all it provides is some raw quarterly directional indicator. That is kinda weak!

Second, I used SEOmoz’s Search Engine Ranking Guide, which seemed like a relatively complete and reliable resource, to rapidly generate the small list of high impact things I could do to rock my pagerank.

Frankly, the thing that caused me the most agitation is that I would love for the URLs to all be Wikipedia-esquely the company name of the org chart. Unfortunately, we continue to refuse to code all of the pain in the ass code to make it happen and make sure that we don’t get 404’s if someone links to a chart and then someone changes the name. There are a host of version control issues there that are Pandora’s box that we have shied away from. We have API’s that let people access pages that way, but we use UIDs to identify each chart in our internal linking scheme. That makes us kind of weak! I would pick up so much pagerank steam if I overcame this.

Moral of the story: SEOing applications is hard because you do not want to make significant functionality changes that hurt the user experience simply to improve rank. Most SEOing articles are assuming that you are simply improving content on the page for machine consumption.

I recently received an email question that I offered to post to the blog:

Currently I use Cogmap as a reference for a company’s general reporting structure.  Is there a way to get in touch with other Cogmap users to pick their brains about what programs they use for their finished product org charts?

Speak up, although my response was that everyone uses Cogmap for the finished product!