Earlier today, I was checking out some new questions in the TechCrunch topic area on Quora. One in particular caught my eye: How was TechCrunch traffic affected by their major redesign in July 2011?

This has been something I’ve seen asked here and there given the radical changes we implemented — and, I assume, given the audience issues Gawker faced after their recent redesign. Mostly, people seem to want to know: is TechCrunch tanking?

I was set to weigh in, when I noticed that someone else already had. This person (not affiliated with TechCrunch) painted a picture in which our site was essentially crashing and burning since the redesign (the answer has since been removed by Quora, presumably due to down-voting). Their source? Compete.

Dear Internet, I thought we’ve been over this? If you cite Compete or Alexa for anything other than making fun of them, you’re officially a moron. How bad is their data? Well, in the case of Compete, if you reversed their chart, then it would be much closer to being correct than it currently is. I seriously wonder if they’re tracking anti-visits or some new metaphysical stat I’m not aware of?

TechCrunch will set a new all-time record for traffic this month (both in uniques and pageviews), breaking the previous record — set last month. And that broke the previous record set the month before that. In other words, things are on the up and up. How do I know this? I have the luxury of seeing directly-measured results from both Google Analytics and WordPress.com’s own analytics area. Both confirm that Compete and Alexa are absolutely worthless when it comes to this type of measurement.

I mean seriously, the data from each service is so bad that I’m not clear how either is still operating. I have to cut the stat-searching public some slack because they are two of the only free public tools out there for gauging traffic data. But seriously, you’d be better off just guessing.

Yes, Quantcast is much better, but they have their own issues too, which is why we don’t expose data publicly that way.

The newer Google Trends for Websites seemed to be halfway decent for a while, but in the past year, they’ve also fallen off a cliff in terms of accuracy, it seems. Given that Google never talks about this product any more, I’m going to assume it’s one of many that has fallen into neglect — which is too bad.

As for Alexa, which is owned by Amazon, I have no idea what’s wrong with them. It seems to be more of a landing page for advertisements above all else now.

And yes, I know Compete focuses on U.S. Internet usage. But looking at our charts, that segment of the data is even more inaccurate when compared to reality. Compete has always been rather quiet about the way they actually gather data — and at times, it has sounded rather sketchy — but the fact of the matter remains constant: the data stinks.

But I will give Compete some credit: their data showcasing their own downfall does appear to be fairly accurate, as backed up by comScore.

As for the Quora question itself, I answered it there. Long story short: traffic is up pretty significantly since the redesign. Thanks all!

You can read the whole post over at Calacanis.com. He doesn’t pull any punches (in fact he goes on a tangent about punching bullies in the face as a kid) He suggests that companies refuse to pay Comscore for the service, and that investors short the stock.

Comscore investor Fred Wilson laid into Jason with a couple of comments on a copy of the post on Posterous. He also randomly dragged me into the argument (I think he’s still mad about the Zynga stuff):

jason, since you’ve slandered me, i’ll respond here.
you don’t know what you are talking about. comscore (SCOR) is a public company. you can go look at their financials. they aren’t exactly printing money. it’s hard to measure the internet and they spend well over $100mm per year doing just that. they aren’t “shaking down” anyone. their move to a hybrid model is a reaction to many of the criticisms that people have had of their panel model over the years. but it isn’t cheap to manage that data either. someone has to pay for this. or of course we could all just let google do it for free. we know how that will play out. eric schmidt has said “analytics are infinitely monetizable” well for google they are. if we want a third party keeping everyone honest, the market has to pay something for it. as i said, go look at comscore’s financials and you’ll see they aren’t exactly getting rich doing that.

and the “huge venture return i made in comscore” is in your imagination. i have not ever made any money personally on my comscore investment.

please don’t spew lies about me jason. with “friends” like you, who needs enemies?

and

hey Karl, i bet if you and i sat down and had a coffee or a beer and talked for a half hour or an hour you’d come away with a different perspective. if you get your data on me and my investments from Jason and his friend Mike Arrington, of course you are going to come away with an impression that isn’t correct. they like to sling mud at me and my investments. i am not going to get into a pissing match with them online. but i am “kinda sad” that you are getting the wrong impression. i don’t know where you live but if we are ever in the same town, give me that half hour and i bet you’ll have a different and better opinion.

Comscore’s CMO Linda Abraham also weighed into the argument on Posterous:

Jason:
You really need to get your facts straight.

1) First of all, we measure Unique People rather than Unique Cookies which web analytics systems erroneously can unique visitors. I would challenge you to find any kind of server side measurement system that measures people, not machines or cookies. To show you how absurd server side numbers are, AOL Inc. had about 259 MM Unique cookies which gives it over 125% reach compared to a true reach of 54%. The inflation is driven by cookie deletion, multiple browsers, multiple machines for the same users, multiple devices etc… Large companies do not complain about their numbers because they know their server side numbers are flawed as obviously evident by the AOL metrics, not because ‘comScore fixes your number”. This dynamic is less obvious with smaller sites—they don’t realize how inflated their numbers are until their reach starts exceeding 100%.

2) Our Hybrid measurement is not mere pixel tracking as you assert. Our panel, which allows us to distinguish people from cookies, is a central part of the system used to correct for the inflation of cookie based server-side measurement.

3) You are confused about our pricing, so let me explain it to you:

• We charge a one-time setup fee of $5,000 that enables us to audit the beacon implementation and make sure we are measuring everyone consistently. This means auditing beacons on every page to identify pages with multiple beacons that result in over-counting, and pages with no beacon that result in undercounting. We have found about 15% of sites have placed multiple beacons on a page, and over 30% of sites that have missed a number of pages on their site. This auditing function is crucial to protect the system from being gamed. Imagine what happens, if unchecked, sites start cross beaconing each other to inflate their audience. The ‘free’ services do not incur this cost because not much is expected of them. We have seen many sites where the Quantcast beacons ‘fire’ up to 7 times from a single page!
• The initial $5,000 setup fee pays for that audit and gives you access to our reports on comScore Direct $5K for 6 month period.
• The $10K annual price is for ongoing access to our comScore Direct reporting system. However, you don’t have to subscribe to continue being measured using the hybrid methodology. As long as you maintain your beacons we will measure you with our hybrid methodology FREE of charge.

4) You may be upset because you don’t get a free subscription to the reports. We make no apologies for charging for access to our reporting system. That is the only revenue source we have to cover our costs. In doing so, we make a ‘mafia like’ pre-tax margin of less than 9% . Google and Quantcast offer metrics for ‘free’ because they have an advertising supported model. They use the data they collect from users or publishers to sell targeted advertising. We chose not to have a business model based on selling advertising, because we do not want to compete with our clients who make a living selling advertising, and who need a neutral third party to provide audience data that is free from conflicts of interest.

5) As for the free trial offer we made you, you need to get your facts straight. When we rolled out this new hybrid system, we needed some sites to beacon with us early to test it out and get user feedback .This is a common practice you might have heard of—it’s called ‘free beta.’ You chose not to participate, which is fine. But there was no attempt to ‘buy your silence’ and we challenge you to prove otherwise.

We provide a valuable service and we are proud of it. We offer the most accurate 3rd audience measurement tools available which are paid for in real dollars by more than 1,200 companies who, unlike you, freely choose them despite available ‘free’ services.

It’s unfortunate that you were picked on as a child. It must have been difficult to you. But you’re an adult now. If you want to debate, please do so with facts, not just blind fury.

My take – Abraham is right. Comscore is by far the best analytics service available. Alexa, Compete and Hitwise are seriously flawed (I may dive into this more in a future post). Quantcast has its own issues and is subject to abuse, which we’ve seen directly. Comscore uses panels and statistical analysis to generate traffic estimates. The new product measure traffic directly off of website servers and should provide nearly perfect data.

And the fact is that the company probably does need to charge to do this properly, as Abraham argues. If a competitor can provide the same service for less (or free), God Bless Them and I’ll support them all the way. Until then, the market will bear what it can bear.

We always choose to use Comscore data first when its available, and will continue to do so. Here’s an example of how useful it can be.

So in this case I respectfully disagree with Jason on the merits of his argument. And I ask Fred Wilson to try to keep me out of his various fights.

Alexa Site Thumbnail was a paying service (developers were charged $0.0002 / thumbnail URL returned i.e. $0.20 per 1,000 thumbnail URLs) but in an e-mail sent out to developers Amazon admits that it never really took off and that the company will do the smart thing and focus their resources on more popular services.

Update: commenters are pointing to Girafa and PageGlimpse as alternatives. There’s also Websnapr, bluga webthumb, scURLr, etc.

Dear Alexa Developer,

We are announcing the deprecation of the Alexa Site Thumbnail service as of March 13, 2009. After this date, the service will be closed to new subscriptions.

The Alexa Site Thumbnail service will continue to be operational for existing subscribers for 90 days, until June 12, 2009. Use of the service has been relatively low, and we have decided to focus our resources on more broadly used services in order to provide the greatest benefit to Alexa customers. Thank you for your use of the service. We regret any inconvenience to you.

Thank you,

The Alexa Web Services Team

(Hat tip to Marc Hodgins)

The biggest change is Alexa’s decision to drop exclusive reliance on the Alexa toolbar for traffic data, with Alexa now aggregating data from “multiple sources” to compile its statistics and web rankings. As part of the move, historical data from Alexa is no longer available, with data now only going back 9 months (we presume calculated using the new methodology).

Alexa is spinning the decision as a step forward without admitting to previous flaws:

Your ranking wasn’t wrong before, but it was different. Alexa toolbar users’ interests and surfing habits could differ from those of the general population in a number of ways, and we described some of those possible differences on our website. While the vast majority of sites’ rankings were unaffected by such differences, we’ve worked hard on our new ranking system to adjust for situations in which they could matter. The new rankings should better reflect the interests and surfing habits of the broader population of Web users.

A search of tech blogs saw many with significant drops in rank, where as political sites have had big boosts. For example TechCrunch and the Drudge Report were tracking similar figures on Alexa prior to the change, where as now the Drudge Report is a mile out in front.

Although regularly derided in the past for its often bizarre results (like YouTube having more traffic that Google), Alexa has continued to maintain popularity due to its broad global reach and completely free service provision. Time will tell if Alexa has done enough to appease its strong and vocal critics.

This was a big surprise for parts of the blogosphere where Flickr is a hot topic.

I looked up these two sites on Alexaholic and found traffic results quite different from the Hitwise graph. Many people have long alleged that Alexa produces low-quality results, is easy to game and is worthy of lots of other criticism. If that’s is the case, is Yahoo! really the most visited site on the web? Is MySpace really number 5? Many of us talk about those numbers, from Alexa, often. (Though Hitwise seems to find similar numbers.)

Graph below: Flickr traffic in blue, Photobucket in red. Webshots.com in green.

Speaking of graphs, here’s some interesting ones that quantify what many people in today’s discussion are saying: the loudest voices in the blogosphere are missing the boat by talking about Flickr all the time. Flickr may be worthy of blog coverage for its innovation or it’s participation in innovative communities or its role in controversy – but among most of the bloggers online Photobucket is a much hotter topic!

Check out these graphs, measuring the times that the words Flickr or Photobucket appear in blogs with many inbound links (“high authority”) according to Technorati vs. in blogs without many inbound links. I think the results are remarkable.

Here’s some imprecise but telling math: high-authority bloggers appear to write about Flickr about 3 times as often as they (we) write about Photobucket. The blogosphere as a whole uses the word Photobucket 3 or more times as often as we use the word Flickr. (TechCrunch has used the word Flickr 11 times more often than the word Photobucket.) Does that mean high-authority bloggers are out of touch with the bulk of users? It may; it may also mean that being interesting doesn’t equate with mass adoption.

In the graphs below, “high authority” on top, all blogs on bottom, Flickr mentions on left, Photobucket mentions on right.

Amazon have been beta testing their web search platform and we have managed to get a hold of some screenshots as well as some information about it. Alexa is giving users and developers access to their crawler in order to build their own search engines. From the website:

The Alexa Web Search Platform provides public access to the vast web crawl collected by Alexa Internet. Users can search and process billions of documents — even create their own search engines — using Alexa’s search and publication tools. Alexa provides compute and storage resources that allow users to quickly process and store large amounts of web data. Users can view the results of their processes interactively, transfer the results to their home machine, or publish them as a new web service.

It seems that users will be able to create any type of search engine, indexing particular types of data or a single or set of particular sites. This service seems to be very powerfull and ambitious, how well it works has not yet been determined. There is a form on the site to register for the beta, but we know that there are only a very small number of users at the moment so your chances of getting in soon are small.

Amazon’s Alexa is opening up its 5 billion web documents and 100 terabytes of data to anyone who wants to use it. Included in this data are Alexa’s famous site rankings based on toolbar users.

As John says, this certainly opens up entirely new classes of search engines and other applications that can be built by leveraging this data.

Alexa is charging for its data, but it isn’t much.

The first 10,000 requests per month are free. Thereafter, requests are charged at a rate of $.00015 each (just 15 cents per thousand requests.). For example, if you make 100,000 requests to the Alexa Web Information Service during a given month, you will be charged $13.50. Your first 10,000 requests are free, while your remaining 90,000 requests are charged at a rate of $.00015 each: .00015 * 90000 = $13.50.

This is a developing story and some of the links are not live yet. More on this as things progress.

UPDATE: Richard MacManus, Om Malik and Dan Farber have more.