Archive for June, 2006
Posted on June 14, 2006 by Robbin Steif
I know that lots of readers don’t use SiteCatalyst, so I will write a minimal number of posts on the Advanced Omniture training I did (and I’ll space them out.) One reader did ask for some Golden Nuggets, so here they are:
Golden Nugget #1) 
When you see an icon that looks like a broken egg, that indicates you’ve enabled traffic correlation. When you see an icon that looks like a 
spyglass (the kind that every other WA solution uses to drill down), that indicates you’ve got a commerce correlation. Most traffic correlations have to be created, which may just require a phone call to LiveSupport – that’s how I got a correlation showing me exactly which terms visitors typed into the organic search to land on each page, page by page. (So I suppose they were correlating their standard organic search term report with their standard entry page report.) However, lots of commerce correlations are already there, you can just see them in the data — just look for the spyglass. (You are wondering why this is in the classifications post? Read on.)
Golden Nugget #2) Classifications always have to be children of a standard variable, like product, pages, etc. However, not all classifications inherit the correlation abilities of their parents. So if you take a paths variable like pages and create classifications (sample classifications for pages in a media site: arts and leisure pages, sports page, business page), you won’t be able to do the same kind of correlations that you might with pages. On the other hand, some variables (I believe they are product, customer loyalty and campaign) are “fully subrelated” — that means that they can always pass their relational genetics onto their children (classifications.)
So, for example, I might own a website for women’s sweatshirts, but all my products are just SKU numbers — not very helpful. So I could create classifications and group the sweatshirts by Type (hoodies and non-hoodies. Now I can pull up Products > Type and see all my commerce metrics for hoodies and non-hoodies.) Everything that I could normally correlate with products can be correlated here. For example, with Products, I can look at not just the Big Three Sales metrics — Revenue, Orders and Units — but I can also look at cart additions. Product views. Finding methods. (Some of those get pulled in by clicking on the spyglass icon, since that’s the way that correlation is signified in commerce.) The point is that all the children of products, customer loyalty and campaign can have the same correlations as their parents, but the same is not true of classifications belonging to other parent variables.
Golden Nugget #3) Omniture works hard to call things by names that are meaningless — but then, if you’re still reading, you know that already. WRT Classifications, they have a neat tool called SAINT — probably some kind of backronym. However, they should have called in the SC Excel uploader. It’s pretty easy, even for non-techies — you just create your classifications in Excel and use SAINT to upload them. (It’s useful for other uploads, but we won’t go there today.)
Robbin Steif
LunaMetrics
PS I am hoping that all the people who work with SC day in and day out will get on an make additions and corrections. For example, I was unable to find a Path or Traffic variable that had any preset correlations, i.e. correlations that I hadn’t created myself. So that addition would be really useful.
View Comments (1 Response) | Categories: WA Tools
Posted on June 12, 2006 by Robbin Steif
Just yesterday, I got some email from another web analyst asking me to buy his new .pdf booklet. I was really surprised that the price wasn’t in the email. “He should know better, he’s a web analyst,” I thought. “Clicks are precious.”
So I clicked through to his landing page — and the price still wasn’t there. And then I clicked again and finally got to it.
Do people bury their prices because they are embarrassed? Or, do they think that the customer will say, “I’ve spent so much time on this site looking for the price, I’ll just buy it now that I’ve finally figured out what it costs?”
Now, it’s not always easy to show your prices. Just today, I was talking to a customer about including prices on a category page. We didn’t want to say, “Starting at $10.00″ when the prices went up to $300 (feels deceptive.) We didn’t want to say, $10-300 (seems kind of useless.) We haven’t decided how to handle it, and I only point it out to show that I, too, am a sometimes-sensitive soul who realizes that not all cases are the same. Sometimes you’ve got a complex sale and you need to build the case before the customer has a chance to calculate the damage. But if the pricing is very straightforward and competitive, why hide it?
Robbin Steif
LunaMetrics
View Comments (2 Responses) | Categories: Usability
Posted on June 11, 2006 by Robbin Steif
As website owners, we often envision paths through our site (“First they’ll start here, then they’ll click here…”) But, if we do a good job with our search engine optimization, visitors enter in strange places. Every page (with the exception of pages off limits to the search engines) has to be clearly labeled as if someone were entering the site right there.
For example, if you do a Google search on T-Mobile call Canada, you’ll probably get this page high in the SERP:
www.t-mobile.com/templates/faq.aspx?PAsset=Inl_Pop_FAQ
But when you click through, the T-Mobile page is a list of FAQs to a specific, albeit unnamed T-Mobile program. It’s ok to land on the wrong page, it happens, but maybe this was the right page ? And the customer never figured out the answer, and there was no navigation so the customer finally called the 1-800-TMobile number (just what T-Mobile doesn’t want. Conversion on a customer service site is about creating great service on the site, not about frustration ending in a phone call that costs the company money.)
All they needed to do was label the page Frequently Asked Questions: Program Name and be sure to link the Program Name (whatever it was) back to the actual program. (After all, visitors who land there have no back button to see the path T-Mobile had in mind.) Some persistent navigation would be nice, too, so that visitors can go back to the site and find what they really are looking for.
Robbin Steif
LunaMetrics
View Comments (No Responses) | Categories: Usability
Posted on June 10, 2006 by Robbin Steif
I recently wrote on the Web Analytics Forum that I would be attending advanced Omniture training, prompting email, “Tell us what it’s like and if it’s worth it.â€
Summary: It was worth it, but might not have been if those two key success predictors, luck and preparation, weren’t both moving in the same direction. In addition to luck and preparation, I’ll also talk about the trainer and the materials here.
Preparation: Omniture (SiteCatalyst) Advanced training is a lot like business school. You can go even if you’ve never had a job but you’ll learn a lot more if you’ve had real-world experience. I definitely wouldn’t recommend it to someone who hasn’t spent 6-12 months working with the product. And all those little training videos, which can drive you a little crazy because they aren’t always updated for new code? They’re worth watching before you go. It’s also really helpful, before you go, to try to get something implemented by yourself (or at least, figure out what the implementation is and show it to IT.) That way, when the trainer says, “Oh, you mean you want to actually create sProps (or eVars, or events, or campaign code??),†you’ll have a vague idea what you and/or your IT department need to do to make that magic happen. If you’ve got campaigns and custom events and custom variables already enabled, the training will be that much more helpful because you’ll be able to further customize your application during the training (everyone has a computer during training.)
Luck: I walked in 10 minutes late and took the only seat available. I was sitting next to a quiet woman who turned out to be Connie Ross, the head of web analytics for LensCrafters. Connie probably could have taught the course, she knew so much about SiteCatalyst. So every time the trainer said something that I didn’t understand perfectly or that he didn’t exactly explain how to do, Connie explained it and/or showed me how to do it. Not everyone gets that lucky, but if you go, I strongly encourage you to learn from the others in the class who have implemented at a deep level.
As for the trainer and the materials:
Trainer: Nathan, the Omniture trainer, did a fabulous job of taking many pages of corporate-speak and making it understandable. For example, he was able to take this sentence, which sounds like it is from an academic paper, and make everyone understand it. (It is arguably the heart of SC, so it’s worth knowing) :
Commerce variables constitute the domain of a function and events constitute the range
I actually understood that one because two Omniture people had separately sat me down and lectured me on this topic: eVars and events are orthogonal. Perpendicular. Every time a visitor triggers an event, all the eVars get credit. Nathan, however, threw that sentence out the window and instead made everyone understand croquet: Visitors roll along like a Velcro croquet ball, picking up information. When the ball rolls through a hoop (an event), all the information sticking to the ball scores a point. If you don’t understand the relationship between eVars and events, you sure will after Advanced Training with Nathan.
I think Nathan (and all the trainers, I suppose) would be a little more effective if they knew what all basic licenses include — the minimum. This is because we got through few sessions without someone asking, “How much more does that feature cost?” It’s true that everyone has a different license deal with Omniture, but I often talk to Live Support and when I ask “Do I have to pay for this?†they (sometimes) answer, “All licenses include [two free custom reports, one Excel Client “create†seat, at least two seats that can talk to LiveSupport, etc.]â€
While this is not an implementation class, most of the students really wanted to understand the implementation. Even though every customer is different, Omniture would be smart to include a copy of the Quick start guide in every student package. It would be less expensive than the fancy giveaway and more helpful (especially when the classroom computers don’t have Adobe on them, so bringing up the guide yourself isn’t possible during training.) So if you are going, print it out before you go — it’s only 25 pages.
Is it worth it? Well, YMMV, but if you need to learn
- more about advanced classification and segmentation
- how all those features you’ve been ignoring work (data warehouse, excel client, VISTA, report suites, ASI, Discover)
then, it’s definitely a good use of time and money.
Robbin Steif
LunaMetrics
View Comments (4 Responses) | Categories: Web Analytics
Posted on June 9, 2006 by Robbin Steif
Like most people, I read the news earlier in the week about Microsoft choosing WebTrends for their WA solution. So I was really surprised to get here to NYC – Advanced Omniture Training – and find three people from MSN in the class.
“I thought you guys are using WebTrends!” I blurted out in my usual undiplomatic fashion. Later, I went up to the MSN attendees to get a better fix on why they were there. In fact, they pointed out to me, Microsoft uses at least ten different WA solutions, at least one of which is home grown. Business units are allowed to choose the solution that they want, they said (hence MSN’s choice of Omniture – SiteCatalyst.)
Robbin Steif
LunaMetrics
View Comments (No Responses) | Categories: Web Analytics
Posted on June 6, 2006 by Robbin Steif
Last night, I was working at my computer while my 15-year-old was buying something on line, on her computer. I wasn’t paying any attention to her until I heard her ask, “What’s the difference between Address1 and Address2?”
Before you mentally reply, “But she’s only 15,” remember that teenagers are a highly sought demographic group – after all, brand preferences are sometimes set at an early age. Further, I have heard adults say the same thing (“Do they want a backup address in Address2?”)
Here’s a great example of two address lines without the confusing 1-2 nomenclature:
The website where I got this also does a particularly nice job of sizing the input fields to the length of information that is expected, which is a subtle way of helping users put in the right kind of information.
Robbin Steif
LunaMetrics
View Comments (2 Responses) | Categories: Conversion Science
Posted on June 5, 2006 by Robbin Steif
So maybe you aren’t doing e-commerce, but you probably still have a call to action: an opportunity for the visitor to raise her hand and say, “I’m interested!” That’s when you convert her from a visitor to a prospect (or maybe just a suspect.) In any case, I am always awed at how companies hide their conversion opportunities (aka their calls to action.)
Recently, I was working with a very close friend on a website. The whole purpose of the site was to get the visitor to pick up the phone and call, or to send an email. “So why do you bury your call to action behind a ‘Contact Us’ link?” I asked her. “If the idea here is to get them to take contact you, and this is the only action that they can take, why isn’t the phone number in the persistent navigation, right at the top of the screen (on every screen?) Why isn’t the email contact info up there too? And if possible, why don’t you have a “contact us” form on every page?”
I don’t think my own web site is always a model of best practices, but in this one area, I think it’s worth looking at. The potential customer can contact us with a form on every single page. I don’t have lots of data yet, but I can see that I get more “contact us” forms filled out from the home page than from any other page.
Robbin Steif
LunaMetrics
View Comments (2 Responses) | Categories: Conversion Science, Usability
Posted on June 1, 2006 by Robbin Steif
Earlier this week, I explained how to calculate a very simple value for the kinds of conversion events that happen on lead generation websites (e-mail registrations. Demo downloads.) I also wrote about calculating the value of an event when someone might convert more than once before actually buying (i.e. they might register for the email as well as download the software.)
Today, let’s talk about how to weight the value of those events. All the data here are fake, so that I can make them do exactly what I want.
The trvial example: Let’s start with a website that sells basement waterproofing. In order to show how awesome their service is, they let you download a canned presentation, but you have to choose a presentation that is compatible with the Mac or Windows. No one does both conversion events, so we should look at them each in isolation. Let’s say a thousand people downloaded the presentation. 200 were Mac presentations and 800 were Windows presentations. Of the ten people who ultimately purchased (at $3000 a pop — waterproofing is expensive!) three watched the Mac presentation and seven downloaded the same presentation for Windows. Now, instead of lumping all the conversions together like I did in Part I, let’s calculate them separately:
Value of a Mac presentation= (3 purchases*$3000)/200 Mac downloads=$45
Value of a Windows presentation= (7 purchases*$3000)/800 Windows downloads= $26
Now you can put a separate value for each conversion event into your web analytics, and convert your return on investment for your marketing dollars that lead to those events.
A harder example. Many companies that don’t do e-commerce have multiple conversion events, and visitors will often convert more than once, especially if the ultimate purchase is a high-ticket or “high consideration” product/service. So now let’s consider a trade organization that has two conversion opportunities on their website: you can register for the e-newsletter or sign up for a webinar. Once again, 1000 people have converted, but this time, let’s assume that 100 signed up for just the e-mail newsletter, 300 signed up for just the webinar, and 600 did both. Ultimately, 43 people actually joined the organization, which costs $2000. Of those 43 people, one had originally signed up for just the e-mail, 12 had signed up for just the webinar, and 30 had signed up for both.
So what are the two different kinds of conversions worth?
Value of an email registration: (1 purchase*$2000)/100 email registrations= $20
Value of a webinar signup: (12 purchases*$2000)/300 webinar signups= $80
I know, you are dying to know how the people who sign up for both conversion events affects the value of each event. When I designed the dataset, I was very careful to make sure that the probabilities lined up perfectly so that we could completely ignore it. Notice that there is only one email converter who also becomes a customer (1/100 email converters – 1%), there are 12 webinar converters who also become customers (12/300 webinar converters = 4%) and there are 30 “both” converters who also become customers (30/600=5%). That means when you add the chances of an email converter becoming a customer (1%) to the chances of a webinar converter becoming a customer (4%), you get exactly the chances of a “both” converter becoming a customer (5%). It looks like just numbers and statistics, but it means that there is no synergy between the events — they don’t have a 1+1=3 effect. This means that we don’t have to pull apart the “both” category and start assigning value to the two different events.
Don’t quite believe me? Let’s do it another way to check ourselves. Since there were 100 email only converts, 300 webinar only converts, and 600 “both”, that means we ultimately had 700 (100+600) email converts, and 900 (300+600) webinar converts. 700*$20 per email convert = $14K, and 900*$80 per webinar convert = $72K, for a total of $86K. And how much revenue did we bring in? 43 customers at $2000 each, which equals…. $86K.
Robbin Steif
LunaMetrics
View Comments (5 Responses) | Categories: Conversion Science
When you see an icon that looks like a broken egg, that indicates you’ve enabled traffic correlation. When you see an icon that looks like a 
spyglass (the kind that every other WA solution uses to drill down), that indicates you’ve got a commerce correlation. Most traffic correlations have to be created, which may just require a phone call to LiveSupport – that’s how I got a correlation showing me exactly which terms visitors typed into the organic search to land on each page, page by page. (So I suppose they were correlating their standard organic search term report with their standard entry page report.) However, lots of commerce correlations are already there, you can just see them in the data — just look for the spyglass. (You are wondering why this is in the classifications post? Read on.)
Subscribe to this blog's feed
