Here is the visitor cloud from my blog:

So I will ask the readers in Edinburgh and Zurich and Haifa and everywhere else but W. PA to forgive me for making a local post: Since a number of people are from the Burgh, including yours truly, I am taking this opportunity to tell everyone that we’ll be doing our first Pittsburgh Web Analytics Wednesday on Wednesday, April 4, 6-8 pm. Location TBD soon. I am leaning toward the S. Side or Oakland, suggestions are welcome. Post here or send me email, steif at lunametrics dot com.

Robbin
LunaMetrics

ps I know you are dying to know, did I exclude myself from the data? Eventually I did set a cookie to do just that, but I definitely skew the results. So maybe I will be talking to myself…

Last Friday, I wrote what I thought was a pretty nice review of StatCounter, a little free web analytics package. I included this line, “And one should never forget that, like many client-side analytic packages (i.e. the kind where you don’t need server logs), they don’t collect data about bots (like the GoogleBot), because most bots don’t read javascript.”

Being a good cross referencing blogger, I then went to the StatCounter forum, where I have never written, and referenced my post in their “Do you like us?” section. I was really surprised to see the Master Member, Christine, write back this post:

“Compliment or Competitor: Forgive me if I’m wrong or paranoid. Bur aren’t you in fact a competitor of Statcounter’s? Your flagship website from your sig doesn’t use Statcounter…”

Wow. I was expecting, “Welcome new member” or maybe, “Thanks for the great review.” I never figured out how using their code on only my blog made me a competitor.

But she went on,

“BTW, you do have some factual errors when discussing Statcounter’s capabilities. The most strinking (sic) of them is that bots don’t get tracked by Statcounter. Statcounter tracks image enabled bots. The bots that don’t get tracked are not tracked because they are not image enabled, not because they are not javascript enabled.”

I wasn’t sure what she really meant, so I went back to Hack 23 in Web Site Measurement Hacks and read, “…a solely client-side data collection model (page tags) may not be able to collect all robot/spider traffic information, because some robot/spider agents do not execute JavaScript and generally do not accept cookies.”

Well, the author (Eric Peterson), wrote “may not” instead of “will not.” So I wrote Fred Kuu from HBX Uncovered. Fred is “the Web Metrics Technical Lead at Adobe Systems” according to the HBX website. Here’s his answer:

Hi Robbin,
Most bots (aka spiders or crawlers) cannot parse and execute Javascript. This is why all vendors (based on JS tagging approach) tout that they track and report only human activity. Granted, it is possible for a hacker to program a bot to parse the Javascript but it’s not easy and there’s not much of a gain by enabling it.

Now, regarding images, almost all bots (especially search engine ones) will be able to track if a page contains images but majority will not actually request the image. So in the logs, virtually all bot activity are to web pages and not to images or other binary files.

-Fred

Then I went to my SiteCatalyst user manual (as another reference point - it is also a client-side package) and it said, “SiteCatalyst does not track spiders since they do not load images.”

So finally I wrote Jon B. in London. The great thing about writing the other side of the ocean is, you write them at night and the answer is in your in-box before you wake up the next morning. Here’s what he said:

Hey Robbin, how goes it?

Strictly speaking, bots don’t execute Javascript. The javascript is responsible for loading the image hence there is some indirect truth in the sense that the bots don’t load images. Also bear in mind that bots can load images - that’s exactly how Google images sources its images database-index…

So, I put all this data into my blender, turned it on high and came up with this: If a bot reads a picture (like the Google Imagebot), a client-side solution like StatCounter can pick them up if the company decides to enable that ability. However, most bots are not about pictures, they are about finding your text, and those bots don’t get picked up by client-side solutions because they don’t talk to the javascript.

Robbin Steif, CEO
LunaMetrics

I have referred so often to this older post on scent and websites that it is time to update it (especially before I forget everything I heard at the UIE roadshow.)

You may remember that I orignally wrote about looking for a Honda. “Think of it this way. A red, 5-speed Honda Accord can be classified as a red car, a manual transmission car, or a foreign car… If I click on the “Manual Transmission” link and just see the Honda listed as “Honda EX,” I lose the scent. ‘This isn’t for me,’I think to myself, ‘I was looking for a car with a stick shift. I’m in the wrong place…”

What I was describing, without knowing the term of art, was the concept of target words. Visitors come to the site with certain target words in mind, and if they don’t see their target words, they lose the scent, just like a bloodhound does when he gets to the river. That’s why it’s so wonderful to link to the same page using different terms - you create different kinds of scent for different people. (When I bought my first stick-shift, I never would have known to call it a 5-speed.)

Short links don’t give off much scent but they can emit a little. This is something I have always known - would you rather see “You can learn more about web analytics?” or “Learn more about how web analytics can improve your conversion rate?” The worst are completely scentless links, like this one: Learn more. Spool & Co. say that the best links are 7-12 words. The problem, of course, is that the navigation bars aren’t long enough for those long links, so they have to be embedded in the copy. There is a conversion opportunity here: Instead of having buttons that say, “Submit,” we’d all be better off with buttons that say, “Submit my automotive question and hear from a dealer in 24 hours” or similar.

Something I didn’t always know is that users feel good when they perceive that they are getting closer to the target information(the blouse they want to buy, the car registration they want to renew online) but they lose the scent when they expect the link to take them to more specific information and it takes them to more general information.

Finally (for today) - while descriptive pictures can create scent, they post a problem when people don’t know to click on them. I wrote earlier about having the design vs. marketing debate with my designers. During the presentation, the designer used an example site, that he had designed, to show how great the visualization was. He clicked on the graphic and another whole set of information came up. I stopped him immediately. “You’re right,” I said. “It’s fabulous. It explains the problem beautifully. In a million years, I wouldn’t have known to click on it.”

Robbin
LunaMetrics

Today, a colleague paid me (and my designer) the ultimate website compliment. He asked if he could copy my website structure. “I figured you had it all worked out,” he wrote.

There was a lot of truth in this — enough that I could feel good about telling him to go ahead. I really did try to work in best practices, because I had to. How can I tell customers with lead generation sites to put a contact form on every page if I don’t do it? How can I tell them to link to their privacy policies right by the email field if I don’t do the same?

On the other hand, I didn’t win every battle with my designers and still have a lot of work to do on my own. For example, I don’t have a great 404 error page, so that’s still on my list. I don’t have on-site search that can handle spelling errors and stemming — if you type in “emarketing” instead of “e-marketing,” you get a No Results page. (It’s a nice No Results page, but plain old Results would be better.) I didn’t have enough negotiating capital to get my navigation below the banner, where people would actually see it.

Which brings me to my point. Big companies like Amazon or a small web conversion company like mine aren’t always worth emulating. You assume they’ve tested everything — but maybe they haven’t. Sometimes the politics of an organization force upon companies sub-optimal solutions that everyone can live with.

Because my colleague wrote to me first, I was able to point out one change that would improve his site based on the experience I had with my own. But if you’re copying Amazon — well, the employees at Amazon are better about keeping secrets than the CIA is. Even if you do have a friend at Amazon, she won’t tell you anything. In that case, you have to measure and test, test and measure.

Robbin
LunaMetrics

Let’s talk about free web analytics.

While Google Analytics are really awesome (I’ve worked with customers who already have them), I’ve been on their waiting list since November. So I don’t even bother to recommend them to customers anymore. If the website is mission critical but not a profit center, I recommend NetTracker or ClickTracks. (Note: I hear that NetTracker 8.0 is going to “do money.”) For e-commerce, SiteCatalyst is my weapon of choice, although I’m sure I’ll love HBX eventually. And there are lots of other great packages that I haven’t touched upon. But for customers who have ten-page sites, I really encourage the use of StatCounter. Here are the top reasons I like it:

#1. Most other free analytics are server side, i.e. you have to have access to the actual web server to be able to install the software and then run it against your logs. Ten-page sites are almost always hosted at some ISP, and installation of StatCounter is much easier.

#2. StatCounter makes it really easy to exclude your own computer from the analytics, even if you have a dynamic IP address.

#3. StatCounter provides rudimentary clickpaths, so I can watch how people come to the site and then where they go (and if they reach a “Thank you very much” page, how they convert. It’s not Omniture, but one has to keep the price/performance ratio in mind.)

#4. StatCounter has drilldown. So, for example, I can choose the referrer tab (what they call “recently came from”) and next to each referrer, I can click on the spy glass to see the IP address, which is often resolved into a company name that I recognize. It’s not NetTracker, but again, there’s that price/performance ratio…

#5. In addition to drilldown to IP/domain, StatCounter shows you where the visitor is physically, on a map. Or at least, where her computer is.

#6. On their summary page, StatCounter lets you choose the time period you want to measure.

It also does all the things you would expect of a free package. Entry pages, exit pages, search terms, etc. Of course, there are lots of things StatCounter doesn’t do. Other than letting you choose a time period, you can’t customize anything - no customized dashboards, no campaign management. And one should never forget that, like many client-side analytic packages (i.e. the kind where you don’t need server logs), they don’t collect data about bots (like the GoogleBot), because most bots don’t read javascript. So for the first time today, after I had a new client put up a “coming soon” site, I wrote the developer and said, “Please load AWStats. [A log-based, not-as-interesting-but-still-free-package.] My only goal right now is to see if the bots/spiders are coming to visit.”

Robbin
LunaMetrics

Recently, I had a Design vs. Marketing debate with my design firm for some customers. The designer-debater made some really good points. But I was delighted to be at the UIE seminar in DC and see lots of my debating thoughts validated. Here are three:

1) Although designers often say, “Don’t write long text, people don’t read it,” the usability guys validated my thought that long text is just fine if it matters to the visitor. Long pages, they pointed out in DC, tend to have more scent.

2) I couldn’t get Christine Perfetti, who spoke for half the day in Washington, to agree that blue underlined links work better than other kinds of signals that a link is clickable. “Test it,” she said. But when Jarad Spool got up to speak, he said, “Hey, a Swiss astrophysicist [Tim Berners-Lee, who really did create the Internet] chose blue underlined links, and now we’ve all been conditioned to click when we see a blue underlined link.”

3) And my favorite: I always want to hyperlink lots of different words on one page to the exact same page. Designers say to me, “Why? You’ve already linked to that page twice. It just breaks up the flow of the copy.” But the usability guys said something like, “Users just want to get their job done. If they get to the correct page using one link instead of another, they don’t care. They don’t even notice that there were five other links on the homepage taking them to the same category page.” In fact, Spool & Co. suggested that multiple links using different words work in the websites favor, because they increase the chances that a particular visitor will see the words he is looking for in a link - and click.

Robbin
LunaMetrics

I’m in DC at the UIE Roadshow conference (so, my apologies for not participating in the testing conversation that Offermatica and SiteSpect are having on yesterday’s permalink. I have to get this post out at a T-Mobile hot spot before my battery runs out.)

Anyway, Christine Perfetti from Jarad Spool’s company, User Interface Engineering, spoke about one of my favorite topics today: does anybody scroll anymore? I had this fight many times with the company that designed my website. They complained to me that my text was too long, and that no one would scroll down to read it.

In fact, Perfetti pointed out, people do scroll, but pointed out four major reasons that stop visitors from scrolling:

1) You have tiny text right above the fold (i.e. before they start scrolling) so it appears to be the copyright. When visitors see the copyright at the bottom, they know the page is over.

2) You’ve got a big white space at the bottom of the live window, so they assume that the page is over

3) You’ve got some kind of long horizontal line or graphic that signals “Page Over” to the visitor

4) The visitor gets what she calls Iceberg Syndrome. He doesn’t find any scent in the top of the page, so he freezes up and stops scrolling. Maybe bails completely.

I haven’t done the kind of user testing that UIE has, but I’d add a fifth reason to this list: Only a small part of the page is below the fold. For example, I tried really hard to buy this book (which is only coincidentally on usability), but bounced between two pages three times before I figured out that if I only scrolled a little more, I would find the “Buy” button.

I’ll write more about scent and usability (which affects conversion, because if they leave the site, how likely are they to buy?) in future posts. If Christina and/or Jarad get time between cities, maybe they will join us.

Robbin
LunaMetrics

So what is A/B testing, anyway?

A/B testing springs from old fashioned direct marketing. Mailers would have a mailpiece that worked well, but wanted to see if they couldn’t improve on it a little bit. So they changed just one variable — maybe it was the snipe (”Your free gift inside!”), or the offer - and tested the new piece against the control. Because they changed just one variable, they could scientifically measure not only whether the change made the mailpiece perform better than the control, but also what about the new mailpiece made the difference.

And so it is with the Internet, although it always requires technology. The idea is the same - some people get served the current page and some get served a test page, with only one variable changed, and then we use web analytics to measure the performance difference. If you own Eric Peterson’s Web Site Measurement Hacks, you can check out his Hack 63, Run your own Split Path test, which teaches you how to send some visitors to one page and another to a test page. Alternatively, you can use the services of a company like Offermatica or SiteSpect. In fact, a comment from a reader at SiteSpect gave me the idea for this post (so keep those comments coming…)

Let me take one last minute to distinguish between A/B, multivariate and Split Path, since I’ve mentioned them all. A/B is when you test one kind of creative against the current creative. Maybe you just change the headline on a test page. When doing A/B testing, never forget what your 8th grade science teacher taught: if you change more than one variable, you don’t know which one was responsible. Multivariate is when you change a bunch of elements on the test page, and then use statistical technology to figure out which elements mattered. (I will do another post on this if anyone is clamoring for more…) Finally, Split Path is just the technology to send some people to one page when they click on a link and everyone else to a different test page when they click on the same link. You can do 3-way or 4-way or nth-way tests, but you generally need more traffic so that your sample size is large enough to mean something.

(MarketingSherpa did a very interesting piece on testing multiple variables with a small audience size. I will dig it out when I get back from the UIE Usability conference.)

Robbin
LunaMetrics

Recently, I subscribed to BlogPulse Analytics. (Well, ok. I haven’t handed them the $24/year it will cost me to be a real subscriber. But I will.)

Anyway, BlogPulse has a number of free analytic tools that anyone can use to find out what kind of buzz is going on among bloggers. Need to know about the buzz on Mark Warner vs. Hillary Clinton? (After all, Warner was on the front cover of today’s NYTimes Magazine.) Put each of their names into the BlogPulse Trend Search and Get Trend. Here’s what you see:

So, I want to know how the blogosphere handles web analytics. The first time I did the trend report, I found there were two big peaks: when Google brought out its free Google Analytics in November ‘05, and when Google purchased MeasureMap in February ‘06. I tried to correlate Google Analytics with web analysis the way that I did the same correlation on Mark Warner and Hillary Clinton but found that the buzz for Google Analytics was so many hundreds of percent higher than the buzz on just web analytics that the web analytics line was reduced to an almost straight line on the bottom of the graph. Instead, then, I did a correlation of web analytics and MeasureMap:

Notice that web analytics has its peak in November, with the Google Analytics announcement. It hits the smaller February peak with the Measure Map announcement (probably for posts that include both the terms “Measure Map” and “Web Analytics”) but the yellow Measure Map line is off the charts (no pun intended) in February, for all those posts that include just the Measure Map name (and no reference to web analytics.)

Anyway, use the tool for your own purposes and enjoy.

Robbin
LunaMetrics

I’ve been pushing all my customers to validate their sites with Google Sitemaps, especially now that the initial buzz is over and there’s no waiting list. Since this is a long post, I’m going to skip all the great ways Google Sitemaps shows you whether they are reading all your pages, and just talk about the analytics provided once you’ve validated that you’re the site owner or webmaster.

Lots of people don’t have web analytics but need them. I can’t tell you how many customer engagements start with me asking the customer to please install web analytics so that we can actually achieve something. Once you establish yourself and your site with Google Sitemaps (and they are so good about walking you through the validation step by step that I won’t do that here), you get some immediate analytics. They call them stats. There are two sets of stats that are very interesting, Query Stats and Page Analysis.

Query stats. Sitemaps includes two kinds of query stats: 1) Which search terms most often brought up your site in the Search Engine Response Pages, and 2) Which search terms both brought up your site and people clicked on them. For both of those metrics, they tell you where you rank on the page. So this is a mini tool for Search Engine Optimization, and the most interesting part about it is what is not there. For example, if I didn’t have web analytics and I weren’t using Google SiteMap, I might think that my site ranks well in Google for website conversion rate, the way it does in MSN and Yahoo! In fact, I’m still buried for that term in Google.

Page Analysis is incredibly interesting. It shows keyword density for your entire site. For example, here are the “Common Words” for my company’s website:

lunametrics
search
conversion
rate
email
software
marketing
analytics
visitors
engine
business
analytic
keyword
customers
pay
package
engines
phrase
privacy
guarantee

They don’t say that the words are in order of density, but it seems to be that way. And if so, I just learned something really important - we don’t use the word “analytics” enough. BTW, there are a number of free tools that do this kind of analysis, too, but you don’t get to do the entire site with one swoop.

Page Analysis also shows how words are used in the links to one’s site. Here are mine:

lunametrics
www
com
http
luna
metrics
consulting
website
analytic
conversion
rates
company
improve
your

This is important information to me because I want to know whether my links use more than just my company name. I’ll get more “credit” in the search engines for the word conversion if it’s in a link. Sure, I get credit for the word LunaMetrics when it’s in the link, but my site will always come up #1 for a search on “LunaMetrics.”

I hope.

Robbin Steif, CEO
LunaMetrics