For all you analysts: should you want to brush up on your knowledge of statistics (or even start from scratch), here is a link to the Statistics Open Learning Initiative at Carnegie-Mellon University. It is completely on-line, self-paced, and free. I road tested most of the course about a year ago and really loved it. The professor who wrote it does a wonderful job of using real world examples. For example, he teaches box plots and distributions using temperature — he shows that the average high temperature here in Pittsburgh is the same as the average temperature in San Francisco, but oh! does the distribution vary.

I see now that you can do this course with either Excel or Minitab.

Robbin Steif
LunaMetrics

As promised, here is installment VII of my Regular Expressions (RegEx) tutorial - parenthesis. I am learning and sharing at the same time. I am only learning them to use for Google Analytics.

I wanted to get this one out soon after my last RegEx post, because the last one was on the use of pipes, which stand for OR in Regular Expressions. Pipes (OR symbols) and parenthesis often go together.

My tutor, Steve in Australia, does a really good job of explaining parenthesis. In the same way that this mathematical statement,

6*(2+3)

is equivalent to 6*2 plus 6*3, parenthesis in Regular Expressions make sure that the stuff outside of the parenthesis get applied to the stuff inside of the parenthesis equally.

For example — and remembering that the pipe symbol | stands for OR — we can have a regular expression like this:

grand(mother|father)

That will match either grandmother or grandfather.

Or, here is another, similar but not identical example:

Ste(ph|v)en

that will match either Stephen or Steven

What if the two terms are really different and there isn’t much in the way of grouping to do? For example, what if we want to filter out Robbin or Luna (which I do all the time in my GA)? Then we can go back to the last lesson on OR and just use a simple pipe:

Robbin|Luna

(Often, even people who know me well misspell my name, so I could use what I learned in lesson V, question marks, to make the second “b” optional, like this: Robb?in|Luna)

In Google Analytics (I won’t speak to other languages) we don’t need to use any parenthesis if there isn’t any grouping — the pipe can stand on its own. Or as Justin always tells me, keep it simple.

[Incredibly techie addition: My last comment about never needing parenthesis when there is nothing outside the parenthesis is not always true. At the eMetrics Summit, Nick from Google and Justin from Epikone taught me a lot about creating custom filters and during that process, explained how parenthesis define a variable. I will revisit this topic later.]

Backslashes \
Dots .
Carats ^
Dollars signs $
Question marks ?
Pipes |
Parentheses ()
Square brackets []and dashes -
Plus signs +
Stars *
Regular Expressions for Google Analytics: Now let’s Practice
Bad Greed
RegEx and Good Greed
{Braces}
Minimal Matching
Lookahead

Robbin
LunaMetrics

I really do love FeedBurner, and it’s such an appropriate gift - I just found out that I am speaking about feeds and blogging in November.

Robbin Steif
LunaMetrics

ps I promise that I will write Part 7 of my RegEx series, maybe even today.

I got tired of waiting for Crazy Egg to have FAQs so I just wrote them myself. I put them together based on the many questions Hiten Shah from Crazy Egg has tirelessly answered for me, as well as my own perceptions. No one from Crazy Egg has seen this or blessed this, and any mistakes are mine (mine, mine, all mine.)

Q: How is Crazy Egg different from other overlays?
A: Other overlays show you where people click on links. Crazy Egg shows you where people click, even if they are clicking on a picture that doesn’t have a link associated with it. This is very helpful because it tells you where visitors expect to be able to click — where you should have links . Also, some overlays give the same amount of “credit” to two links on a page that go to the same place, and Crazy Egg only counts a click where it happened. Finally, Crazy Egg has not one but three different kinds of overlay reports.

Q: What are the differences between the three reports that Crazy Egg provides?
A: The Overlay (#1) displays your test page with buttons next to the places where people click. When the buttons are close to dark blue, you have few clicks, and when they are at the other end of the rainbow — the reddest — you are getting the most clicks. You can click on any Crazy Egg Overlay button to see how many actual clicks came to that spot. Alternatively, you can use the Flower Marker (the tab next to Overlay) to show you all the results associated with all the buttons. The List (#2) is just that — a list of the different kinds of clicks you get and how many of each there are. And the Heatmap (#3) shows you exactly where on a page people clicked (did they click in the middle of the picture? At the end of the link?) so you can see if links need to be longer or easier to reach.

Q: Can you describe the List better and explain what “Type” means?
A: Even though your most common type will probably be “A,” you won’t see B or C or D — when you see A on the list, it stands for an “a tag” like a href. (A link.) Drop down boxes usually have type Select. Free form large boxes usually have type Text Area, and free form one-line text areas have type Input. IMG is an image.

Q: Why does my dashboard say, 1324 views, 649 clicks, 589 shown? Why can’t Crazy Egg show all the clicks?
A: This means that there are clicks on the page that Crazy Egg is not identifying in the overlay. Many times it has to do with Flash elements, ads and even elements that are not on the page anymore. Javascript dropdown menus often have this problem.

Q: If someone clicks into a field and then enters a name with five characters and then hits the tab or enter button, how many clicks does Crazy Egg count — one, five, six or seven?
A: It is supposed to count that as only one click.

Q: What does it mean when Crazy Egg says that they can archive up to a certain number of snapshots?
A: The snapshots are pictures of a test when the test is finished, since Crazy Egg archives the way the test looked right when it ends.

Q: I thought that I knew how long I needed to run my test, but there weren’t enough visits so the data isn’t interesting enough yet. Can I extend the length of my test without starting a completely new test?
A: No, you can’t. Sorry. (Wouldn’t it be great if everyone’s FAQs were written this way, they actually answered a question like this without marketing doublespeak?)

Q: I can put the Crazy Egg code on four pages at no charge, right? And Crazy Egg will archive four tests at no charge, right? So does that mean that after I have ended one of my four free tests, I can start a fifth test, but I lose the archived snapshot of one of my first four tests? Or does that mean that my fifth test will not allow me to archive the final snapshot? Or does that mean that I am done with my free stuff and now I have to pay?
A: Wow, what a great question.

Q: I noticed that the code Crazy Egg gives me is always the same for my site, no matter which page it is on. Why can’t I add the code to an include file and let it run on every page? I would be happy to pay.
A: True, true. Read the editorial part of this post.

Robbin Steif
LunaMetrics

p.s. If you haven’t tried Crazy Egg, you are missing out on a great analytic tool.

On Tuesday, I talked to yet another investment banker (Morgan Stanley, the company that took Omniture public.) I don’t want to turn these guys away, because it’s good for the WAA and the industry to have this kind of exposure, but damn! I wish they could link to my site. Can’t you see it now, a link from Morgan Stanley’s web analytics research to LunaMetrics? Well, I can dream, right?

We agreed that we would keep it to a half an hour. I don’t remember all the questions (and should leave something for him to write himself), but as usual, he started with the assumption that web analytics “has made it,” and I maintained that web analytics will not really make it until marketers can do most of the work in really easy-to-use interfaces and not have to wait for IT to code an onclick event or a custom eVar. At a certain point I said, “OK, this is your last question.” So he asked his question and I answered and then he rolled right into another question. “Sorry,” I said, “That last one was your last question.” He was very nice and sent me the research on the pricing for the Omniture IPO as a thank you gift and then — he asked his last question again, in email. It was an interesting one, though. He wanted to know when I thought that Google would release their upgraded, non-free version. (This was one of the topics that they raised in their research report, “Google also plans to sell a more comprehensive product offering shortly.”) I considered telling him that Paul Muret writes me daily and tells me all the insider knowledge, but decided to stick with the truth: I really had no insight into that particular issue, and thought that Google would be smart to spend their time bullet-proofing the current free version.

Robbin Steif
LunaMetrics

This is the sixth installment of my Regular Expression lessons. I am actually learning more than teaching and just sharing as I go along. These are Regular Expressions for Regular People (c), so all the tech-talk is removed. My motivation for learning RegEx, as they are called, is Google Analytics.

OR gets symbolized by the pipe symbol |. The pipe symbol is on my US keyword just above the Enter Key but for some reason looks like two vertical dashes on the keyword itself. It’s incredibly simple, and even Google Analytics does a fabulous job with its description:

This was a hard one to screw up, although they have done a good job of screwing up other easy Regular Expressions.

Here’s an example. One of the sites that I work with is an engineering education site, and they teach both Statics and Finite Element Analysis (FEA). Some people also refer to the latter as FEM (Finite Element Method). If I were only interested in the statics keyword searches, I would probably want to get references to FEA and Finite Element Analysis and FEM out of the search reports. I could do that easily in GA by going to the little filter box that can be found on each report, making it into a red minus sign (so that I am filtering out) and typing in FEA|Finite|FEM . This has the effect of saying, “Get rid of references to anything that includes FEA OR Finite OR FEM.”

Backslashes \
Dots .
Carats ^
Dollars signs $
Question marks ?
Pipes |
Parentheses ()
Square brackets []and dashes -
Plus signs +
Stars *
Regular Expressions for Google Analytics: Now let’s Practice
Bad Greed
RegEx and Good Greed
{Braces}
Minimal Matching
Lookahead

Robbin
LunaMetrics

Good Experience must be the classic barefoot child of a cobbler, unable to do for themselves what they do for others.

The Good Experience website says that their mission is “is to encourage the creation of good, meaningful experiences in business and life.” That says, “usability” to me, because that’s the mental slot I have to put them in.

I met the Good Experience guys back in the winter at Shop.org in Atlanta. I gave them my card and vaguely remember telling them that they could put me on their mailing list. I was very surprised to get their first mailing because it was such a bad experience. It was all text (no html) with lots of naked links, lots of hyphens used to break the text up so that they could advertise job openings — just unreadable. I wrote Mark Hurst, the CEO, after my first disappointing experience and he explained that text pulled better for them.

I never read a single issue of their newsletter, and this weekend, I finally unsubscribed. I immediately got a “personal” email from Mark. He called me by my last name and asked me if I could help explain why I was unsubscribing. (Note: I try very hard to never let someone subscribe to my email marketing with poor information. If they have signed up with their first name in all lower case, I go in and capitalize the first word so that my marketing does not come through as, “Hi jim.” Since I know that the GE people had my business card, I was surprised, again, that they hadn’t gotten this right. I don’t remember the last time someone called me “Steif,” when they weren’t writing about me in an article.)

Anyway, I wrote back and explained that it was too hard to read the newsletter, as explained above. Mark replied that I could always go read it online and provided a helpful link. So I clicked on the link in Firefox and was confronted with the tiniest type that I have ever seen online. For completeness sake, I asked Taylor, our 22-year-old analyst, to look at it, and it was way too tiny for him to read, too. Of course, we could increase the type size, but is that a good experience? Eventually we figured out that if you open it in IE, it is not bad at all. So I guess Internet Explorer Users all get a good experience once they get past the email and Good Experience’s “personal” email.

To be fair - Mark really did reply to me in person. And it really is hard to do for yourself what you do for others, often because great gets in the way of good and you end up doing … nothing. Been there, done that.

Robbin Steif
LunaMetrics

Check out my interview with Marshall Sponder of WebMetricsGuru. My spouse looked at my picture, which Marshall included in the interview and said, “Can’t you come up with something better than that?” I guess it is time for a new photo shoot… but as usual, I digress.

This is an interview in advance of the e-metrics Summit. (You’re going right? I really just want to see all my friends….)

Robbin Steif
LunaMetrics

What do you get when you integrate –mentally — your analytics overlay with your analytics numbers? A whole lot of insight into what your users are doing. In this case, my insight was all about internal search.

1. What does the overlay say?

We like to use Crazy Egg for Google Analytics sites because the GA overlay is so lousy. And while we still have to wait for two weeks of data to feel more conclusive, we are seeing that the search box is the most used link on the home page of one site. (Note: if the page under analysis — in this case, the homepage — has three separate ways to reach another page, like “About Us,” the sum of those three sets of clicks may exceed the total clicks for a single link like “Search.” This is in direct contrast to Google Analytics, where you always get the same information on links that go to the same page unless you add special code. But I digress.)

2. What do the analytics tell me when I drill down on the overlay issues?

The fact that Search is so popular rings all sorts of not-pleasant bells. It’s true that some people like to use the search box, just like some people walk into a department store, ignore the signs around them and ask the first salesperson where they can find what they need. However, if internal search is too popular, it probably means that people aren’t finding what they want at a glance - in your navigation and in the links from the home page.

Are they at least finding what they want when they search? I went to my Dynamic Content report and filtered by the word “search” to get the search page. Then, I exported it into Notepad because that gave me more control over how it would end up in Excel. I used Find and Replace to get rid of all the properties and other junk that comes with the search query data. And then I sat back and looked at it.

I was initially interested in sorting by search term, so that I could group identical or near-identical searches together. But as I was straightening up the data, I realized that very few searchers continued after the search - many if not most of the searches had a 100% exit rate. Very strange, right? So I started typing search terms into the company’s search engine. And I was awed at how often the internal search engine came back with “Can’t find it” — even for the company’s most popular products.

3. In summary…

… in summary, in about an hour, I learned that too many people use the search, probably because they can’t find what they want from the navigation and other links on the home page. Furthermore, most of them leave because the internal search engine doesn’t work very well.

What a great application of web analytics.

Robbin Steif
LunaMetrics