Web Analytics software compared

One of my favorite quotes, I don't know who said it originally but I got it from a high school math teacher "If you change the questions, the answers will change accordingly."

Public kudos to Stone Temple for their 2007 Web Analytics Shootout, a head-to-head comparison of how several different web analytic packages running concurrently on the same sites reported different outcomes. A must-read for anyone who produces or consumes data like site page views, unique visitors, and conversion rates.

Too bad WebTrends and Omniture were not included in the broader study, each one was only used for one site in the study. It'd be great to see if the findings were true on more heavily trafficked sites.

Mint.com resuccitates a failed concept

Mint has been getting a lot of good press, kudos to them for that. But the idea is not new. It was tried during Web 1.0 by OnMoney.com, a division of Ameritrade. I worked there for a couple months.

OnMoney tried to make a business as a website where you got manage all your financial data: banking, insurance, investments. Total net worth, with articles from experts advising of all of this.

After I left, I heard they got the results of an $80k usability study: "You've got some great tools that *no one* knows how to use."

OnMoney went on to blow a big wad on a Super Bowl commercial. Not too long after, they closed shop.

Hurdles Mint will need to overcome:

- consumers don't know they need these tools. They aren't fun and sexy. You going to give me a facebook app that lets my peeps know "Steve has maintained a positive balance for 6 weeks in a row!" People -- not the Ivy school grads that fund and launch startups like this, but real live real-world people -- don't talk about finances to their friends. So how go you get a community, how do you spread the word?

- all my financial data in one place? No way I'm going to trust anyone with that. Mint needs to prove to the world that they have a 100% bullet proof infrastructure that the NSA and hackers all over the world agree is insurmountably robust. New code just can't do that. So I'll wait 10 years and if they are still around, then I'll assume their code is good. Can they wait that long?

They could put all their code open-source to address the security concerns, but I'm not sure their business model can support that.

Huppahs.com launched

My sister Maria and I launched her web site this week (on her birthday, in fact). It's very basic php and for now the only method of payment is thru PayPal -- kudos to Maria for that idea and for coding hooks for their cart.

The final version will be fully LAMP. My friend/employee Jeremy Price is hosting for us on his Linux box. Apache (duh), pHp 5 and mySQL on its way, after we get the credit card processor account set up.

So check out Huppahs.com, even if you not currently planning a Jewish wedding.

Eclipse 3.3.1 in Ubuntu

If your Eclipse install under Ubuntu just took a crap like mine did due to a bad update (shhhrrrr, it got rolled back), you may be looking to upgrade to the latest and greatest version of Eclipse, and don't want to wait until it finally gets added to the official repository.

Many thanks to Ivar Abrahamsen for writing How to install Eclipse in Ubuntu. He posted it to the Ubuntu wiki as Eclipse Web Tools so find an up-to-date version there.

domainsystems.com site down?

What's up with domainsystems.com? Their site has been down all day. Not just me; even Netcraft can't reach them.

Update: Day 2, still down. Has this site been down for awhile? Not one that I've used before, but moniker.com is still pointing to it.

VoilaBot, behave!

My websites are in the local advertising space: employment, online dating, business listings. My audience is in the USA, Canada, and Mexico. Because of that, I have no chance of making revenue from traffic from Europe, China, or Nigeria.

So when I see a spider called voilabot, for Voila.fr, pounding my server farm, that's a spider I'd like to disallow. No problem, just add them to robots.txt. Only thing needed is how the bot identifies itself when scanning that file.

Typically, a spider will identify it's useragent and, parenthetically, give a link to info regarding the bot. Voilabot does not, it just points to their homepage:


2008-01-09 05:00:12 GET /robots.txt - - 193.252.149.16 Mozilla/5.0+(Windows;+U;+Windows+NT+5.1;+fr;+rv:1.8.1)+
VoilaBot+BETA+1.2+(http://www.voila.com/) - - 200 1100 358 546


which redirects to http://www.voila.fr/, which sucks for me since I don't know French. I am able to find a page on their site about robots.txt, and how to block *all* spiders from visiting my site -- no thanks! I heart Google -- but nothing that mentions what user-agent top specify to block Voilabot.

The Google index, interestingly, includes the robots.txt files they scan. Examining these, some people specify Voila and just as many specify Voilabot. Other Google results include rant like mine -- apparently this bot has been in Beta since 2001.

So Voila.fr Webmaster: please tell us how to block your bot. Thanks!

Update: 1/15/08
No response from Voiala.fr so I'm now denying them at the firewall. If you don't want this bot, I encourage you to do the same.

We've seen it from

81.52.143.15
81.52.143.16

193.252.149.15
193.252.149.16

Selenium and CAPTCHA

Selenium is a great tool for the automated testing of web applications. CAPTCHA is a way to prevent automated tools from accessing your web applications. So the two are mutually exclusive, by design.

Selenium can only be used to test a web app protected by CAPTCHA if a human intervenes for the test. So, if you are testing an 8 page transaction that includes a CAPTCHA on page 2, for example, you can still save tons of time by automating everything else with Selenium but have the test pause for human interaction during the specific section that requires a CAPTCHA response.

This example assumes you are already familiar with how to write a Selenium test, in HTML.

... the first part of your test goes above here...



<tr>
   <td>clickAndWait</td>
   <td>login</td>
   <td></td>
</tr>

<tr><td rowspan="1" colspan="3">human tester must enter CAPTCHA...</td></tr>
<tr>
   <td>waitForPageToLoad</td>
   <td>60000</td>
   <td></td>
</tr>
<tr>
   <td>waitForTitle</td>
   <td>Title of Next Page Here</td>
   <td></td>
</tr>

<tr>
   <td>type</td>
   <td>FieldOnNextPage</td>
   <td>Test</td>
</tr>

... continue on with the rest of your test. That worked for me in TestRunner, at least when the speed was set to slow.

If you are only going to be using TestRunner, as opposed to using a tool like ant to run Selenium tests during a scripted build, you could more easily just insert the following Selenium command into your test, forcing the test to wait until you click the Continue button in TestRunner:

<tr>
    <td>break</td>
    <td></td>
    <td></td>
</tr>

Note: if you are a scammer, screw you. Further note to scammers: the preceding code won't help you in your scam pursuits, because it doesn't defeat CAPTCHA, it just allows a human to do what only humans can do and let the bots do the rest.

Vintacom now under Think Partnership?

Relationship Exchange, the online dating white labeler, is Vintacom Media Group. But now they are under the umbrella of Think Partnership. When did that happen?