It’s possible that I’m the only web analyst around who didn’t understand the interplay between Google Analytics and Urchin (but I doubt it.)

I have a customer with Urchin logfile software whom I wanted to upgrade to Google Analytics (and I even got them an invitation from GA while I was at the eMetrics Summit.) While I was at the Summit, their Urchin reseller set up their current Urchin 5.0 software for their new website, and told them that there would be an upgrade. I was sure that Urchin is going to die and will be replaced by Google Analytics completely; after all, Google bought Urchin for just that purpose. However, Tim Seward from The Unofficial Google Analytics Blog assured me that I was wrong — he even called me from his son’s ball game on Saturday morning to tell me how wrong I was. Server side Urchin is not only alive and well, but will be upgraded to Urchin 6 sometime in 2006.

The Google Analytics site says,

Urchin 6 software will not be free when it is released in 2006. However, if you paid for an Advanced Support contract that expired after March 2006, you will be offered a free upgrade to Urchin 6 software when it is available. If you are looking for a free analytics service, please sign up for Google Analytics.

And of course, if you want client side analytics, you have to use GA.

Robbin Steif
LunaMetrics

Although the search algorithms are closely held secrets, despite search engine patents, it is no secret that the title tag of each page is an important part of the search engine algorithm. The major search engines all use title tags on their search engine response pages (SERPs) But conversion starts on the SERP, and we don’t often think enough about how humans look at our title tags.

Sure, we can get high up in GYM (Google, Yahoo and MSN) with keyword rich title tags, keyword rich pages and a variety of other strategies. But will anyone click on our sites? (The converse argument is, if we aren’t there, we know no one will click on them. Fortunatley, this is not an SEO blog…) The problem is harder still when you consider the constraints placed by the search engines, particularly Google. Although Google will index over a thousand characters and spaces in your title tag, they only show 63 or 64. You can write a long title tag but the search engines will truncate them and then you don’t get your message across in exactly the way you like. So you may not want to waste precious words with calls to action, like “Try” or “You’ll find,” even though they may help your conversion.

At the very least, though, write to the reader and not just to the search engines. And remember that your title tag gets picked up in many places besides just the search engines. I can think of four right now:

• It’s the name that most browsers give a bookmark that visitors set, unless the visitor changes it
• It’s the description that del.icio.us gives a page on your site, if someone bookmarks your site there.
• It’s the description that users see in their browser’s back button
• It’s the name that users see when they minimize their browser (and your site is there.)

These are all conversion opportunities, and some, like minimized tabs and bookmarks, depend on the first two or three words.

Robbin Steif
LunaMetrics

Earlier in the year, I did a post on RevealSite Analytics, which arguably make the website owner into Big Brother. Then today, I read the MarketingSherpa article about Kevis, the “grow your hair back” product with a website that greets site users with a popup chat box within seconds. Naturally, I had to try it - it was too similar to my earlier post to ignore.

Because I skimmed and scanned the Sherpa article, I hadn’t realized that the popup box would come so quickly. I assumed that it would wait in the background until it saw that I was having trouble, or was following a clickpath pattern sure to end in doom. But the box was in my face within seconds:

Personally speaking, I hated it. It was like walking into a department store and having the salesperson in your face before you had a chance to figure out what department you even wanted to check out. There are four logical and very different conclusions to this problem:

1) Different people have different purchasing patterns. I hate to be sold to and I love to buy. Not everyone is the same.
2) Lies, damned lies and statistics. If 15% of the people who see the box engage in the chat(according to Sherpa), and then 28% of the 15% give their phone number because most people need to be sold on the phone, and then 12% of the 28% become long-term customers, that’s .15 x.28 x.12= .5% — not a conversion rate I’d be excited about. I know that companies who do case studies have limits on what they can say, and so you will see, if you read the case, that there is some fuzziness around those numbers. For example, it’s unclear if the percentage of people who give out their phone number and then make a single purchase is much higher than that 12% (one would think that it is.) Then there are all the people who use the box and purchase without giving out their phone number.
3) Maybe the people who give their phone numbers out to a woman in a chat situation have ulterior motives.
4) And last, there is the possibility that I’m wrong. People love to have a popup box shoved in their face in the first 10 seconds of their shopping experience — or maybe, it just works for this product. Every product and every site is different. A good reminder to test and not let your feelings get hurt when you are wrong, as Avinash Kaushik pointed out at the Summit.

Robbin Steif
LunaMetrics

Everyone in analytics has moved on to the WebSideStory conference but I’m still finishing up on eMetrics in Santa Barbara…

One of the speakers I enjoyed the most at the eMetrics Summit in Santa Barbara was Avinash Kaushik, Sr. Manager of Web Research & analytics for Intuit. He succeeded in putting a lot of web analytic issues into perspective. I should also point out that Intuit uses ClickTracks because, in Avinash’s words, “it doesn’t puke out data.” He loves clickTracks so much that the next day, at the vendor roundtable, he jokingly asked, “could everyone tell me in what ways they are better than clickTracks?”

One of Avinash’s first “Intuit Insights” is that data quality sucks. Here are the major reasons data quality is so lousy, and I have added to his list.

• People delete cookies
• Robots and spiders do weird stuff to your data
• Tagging has limitations - not everyone accepts javascript
• Weblog analytics have limitations - lots of them
• People behave strangely
• Unique visits are never all unique

So just get over it.

• Assume a level of comfort with the data and its limitations.
• Make decisions that you are comfortable with
• Over time, drill deeper in micro specific areas and learn more (this was a continual theme of the Summit: Don’t try to boil the ocean.)
• Keep your calculations consistent and look for trends. Here’s my example: there’s a debate about how to calculate conversion - should it be actions/visits or actions/unique visitors? Following Avinash’s logic, don’t change from one metric to the other. It won’t necessarily be more accurate, and it will certainly obfuscate any trends you had in your historical data.

In addition to his Intuit Insight #1, which was of his most insightful, he added some other nuggets of wisdom:

• Spend 10% of your money on your tools/services and 90% on your people. (Or, “Working part time, your admin can’t extract enough value from the greatest tool on earth, but she can publish tons of reports.”)
• 80% of the time you are wrong about what a customer wants/expects from a site experience. Don’t let your experiments hurt your feelings.
• Bosses always think they represent site users and they want to do site design. Listen to your customer, not your boss. (Comment: Easier said than done…)
• The most phenomenal website is stale tomorrow (or as one of the designers I work with says, a website is like a plant. You have to water it every day.)

Robbin Steif
LunaMetrics

I don’t often write, “And then I did this, and then I did that…” but I wanted to share all the cool things that happen at the eMetrics Summit in between the “scheduled” learning. Two months ago, in fact, I wrote Jim Sterne and asked him for more structured mentoring at the Summit but he wrote, it really happens by itself. And he was right:

This is what I achieve by being there:

• I complained to the Google Analytics team that I have been on the waiting list since November and they gave me an invitation to GA.
• I complained to Jeff Veen, formerly the CEO of MeasureMap and now a Google employee, that I have been on their waiting list since before the purchase. I had an invitation in my mailbox before I got home that night.
• I dragged Matt Belkin, VP of Best Practices at Omniture, over to my computer and showed him my GetQueryParam G code and he showed me exactly what to do. (Now that’s Live Support!)
• I talked to June Li from ClickInsight, who is part of the Toronto Usability Group and learned what I need to pay users who are testing a real estate web site. ($50/hour, give or take.)
• I accidentally sat down next to the founder of Omniture (I was just looking for a shady seat), and he taught me how to install some very cool SiteCatalyst capability.
• I met, in person, Akin Arikin and Clint Ivy and Ian Houston and Jim Novo and Jim Sterne and Eric Peterson and all sorts of other people whom I’ve been corresponding with or reading on the Web Analytics board but who never had faces before.

And just think, there were lectures too.

Robbin Steif
LunaMetrics

At the eMetrics Summit, I unknowingly ended up sitting at dinner between two people I would never have met outside the summit: Paul Muret, the former CEO of Urchin and now the head of Google Analytics, and Tomas Remotigue, formerly an Omniture Sitecatalyst implementer and now the product manager who integrates GA with Google AdWords. I probably got that last description a little wrong, but it’s close. (Inside Adwords thinks they are the source of help on this topic, but now I really know who to talk to…)

On the one hand — the left hand — was Tomas, who taught me a lot about SiteCatalyst, and in particular, how companies who have enormous sites use SC in “sample” mode so that they don’t have to incur high charges. Tomas, who has now been at Google for about a year, also pointed out to me that it is so important for him to get out of the Googleplex and meet real users. “I get to work every day and there are over 500 emails in my inbox,” he said. “Real emails, not spam. I have to get them done before another 500 come through, and oh, I have to get some real work done too. So I have to rely on autoresponders. The chance to sit and talk to real users is wonderful.”

Then I asked Paul Muret, on my right side, how he was enjoying the conference. He thought it was excellent. “Why?” I asked. “It’s not like you need to sell your product, the way the other vendors here do.” (Google Analytics was one of the sponsors.) And then he explained to me that a conference like the eMetrics Summit helps him understand cutting edge analytics and what users want and need in order to build better functionality into GA.

We also talked about the benefits of selling Urchin to Google. He felt that the ability to give the product away for free was wonderful. It is amazing, he pointed out, how much time and effort they used to put into pricing, licensing, etc, and now all of that is gone — their only job is to create functionality for users. (And to figure out who gets GA, I suppose. I would like to be a little more cynical about Google Analytics, but Paul is such a wonderful guy that it is hard to revert to my usual mode.)

Finally, he told me about the best Google benefit: the food. (I’m serious.) “You know how it is when you’re an entrepreneur,” he said, “You’re always running, no time to eat, when you do eat it’s something fast and unhealthy.” Apparently, Google has many cafeterias, one of which is California Healthy with lots of vegetarian options, and he really loves it.

Just thought you would want to know what the GA team eats.

Robbin Steif
LunaMetrics

At the eMetrics Summit today, there was one session where the vendors got up at the longest table in the world, and the audience could ask questions (no matter how tough.) All the vendors were web analytic software vendors or testing vendors. Vendors were, in alpha order: ClickTracks, Coremetrics, ForeSee Results, Google Analytics, Maxamine, Offermatica, Omniture, Optimost, Unica/Sane, WebTrends. I did a pretty good job of capturing what most people said but rarely captured their words exactly. When I couldn’t understand what they were trying to say, I didn’t include their comments at all. And I missed the first question entirely, so sorry.

Question: I’m curious to know how you handle the situation when a customer is not allowed to use cookies(asked by a government customer.)
Google Analytics: We do work with sites without cookies
Omniture: We do too. They are only one way of measuring. Usually, software has a fallback to IP user strings when cookies aren’t to be found. Other options are unique user identifiers (UUID).
NetTracker: Not all tools are all created equal. There are tools where they don’t have cookies, it shows you more visitors than visits (but I won’t say who I’m talking about. ) Everyone bursts out laughting.
ClickTracks: Just about any tool that can accept a log file can work without cookies. Also, there is an interesting intersection w/customers who want to use cookies and javascript but don’t want the data to leave the company. Our product can do that, Webtrends can do that, I *think* that Urchin [GA] can too. So can NetTracker.
WebTrends: We’ve worked a lot with the government standards group on this. Yes, you can work without cookies but what reports are you going to miss?

Same person asks another question: We don’t use cookies at all, we don’t use session identifiers at all on our static site, how do I connect the static site with my dynamic site (which does use session identifiers?)
ClickTracks: This is not an unusual problem. Lots of times the software has to reconnect sessions.
Omniture: Suggest that you use a service like HitWise, gives you some broad pathing perspective.
Another vendor: Create a non-cookie that identifies the session

Question: In a world where there is nothing like a page (he is referring to an earlier presentation on measuring Flash, AJAX, etc) how will you measure?
Omniture: I came from MacroMedia and I don’t see it as a technological barrier.
Foresee: Customers don’t think in terms of pages, we need to think in terms of what matters to them.
Google Analytics: Lots of times, it is about building in the analytics to the application.
Offermatica: We don’t track pages. We track elements.

Question: I want to ask about data integration and multi channel issues. Each customer can put up a NASCAR chart with all their vendors. Why should my web analytics vendor be the center of my measurement universe - why should I import to you instead of export from you to other applications?
Clicktracks: You have so much data in the analytics piece that it is just easier
NetTracker: I agree. But if you prefer, you can take everything out of NetTracker and put it into the other applications.
Maximine: We’re focused on a complete solution. I’m probably the only guy at this table who is really focused on structure. Think about an airline company that is so preoccupied with booking, meals, ticketing, etc. They will have lots of happy customers going into mountains. He says something like, “I don’t want to own the universe” Summit organizer Jim Sterne interrupts to ask, “How many people up here aspire to own the universe? Google Analytics raises his hand for fun and everyone applauds.

Question: What are your solutions for measuring RSS?
No one jumps to answer.
Omniture: At our recent summit, we did a whole seminar on measuring RSS.
Google Analytics: We recently aquired MeasureMap, it is analytics for bloggers.
Anyone else?
ClickTracks: It’s hard

Question: As a Yahoo person, my fear is the amount of data that gets measured, we’re building big data warehouses and tools. Are you thinking about creating statistically significant models where I don’t have to measure everything? How will we deal with the volume?
Google: Sampling is definitely done for the largest of sites. Companies with small conversion rates need all their data for statistical significance.
Omniture: We work with the largest customer. I assume he means eBay and we don’t sample. Never forget that we are talking about people. Web analytics are very uniquely positioned to show you information all the way up the funnel. There is a lot of information and (no disrepect to the original Webtrends) that must be what brought logfiles to their knees.
Webtrends: We have customers with hundreds of thousands of page views per day.
Omniture: That is so 2005.
Jim Sterne: Do you guys want to take that outside? Audience bursts out laughing.

Jim: Any other sampling info vendors want to share?
Clicktracks: We have sampling tools and we can’t persuade people to use them.
Maxamine: The vast majority of enterprises are sampling already. We routinely see 60% of tags missing. The audience breaks out in laughter and applause.
NetTracker: It’s important to keep all your clicks to do, say, data mining. Should you keep it forever? Probably not.
WebTrends: You may capture all data, but selectively keep some of it.

Q: What happens when the analyst is happy and his boss is not? What do you tell to your clients in that position, and how is that different if they come to you as a new customer?
Audience appreciates what a loaded question that is.
WebTrends: We always just want to know what your issues are, no matter what your situation is. When the customer comes to us and has WebTrends 3.0 (from 1996), we can understand why they aren’t satisfied - we’ve changed so much.
NetTracker: Dissatisfaction generally stems from training.
WebTrends: Lots of times, people are dissatisfied because they are using logfiles, and just moving into tagging will fix that (WT now supports both.)
CoreMetrics: Any sales person is happy to hear that the analyst likes it.
Maxamine: If I am the vendor and I don’t see the problem coming, I’m not doing my job.
Everyone weighs in and says, our job is to help the client.

Q: Following up on the scale and volume question: most ASPs are charging by pageview, so aren’t you penalizing us for growing our business? Applause from audience.
Google Analytics: The answer is, don’t charge for pages (everyone applauds because GA is free.) Education is important
ClickTracks: You can buy our software and own the data and still get tagging. No cost per pageview.
CoreMetrics: We are just passing on the load from our servers onto our larger customers. That way, we are not penalizing the smaller users. Like electricity or cell phone minutes, people who use the most, pay the most. Using sampling would help this.
Omniture: Our prices are on a server call, not a page view.
Maxamine: We sell at a flat price. Any customization or innovations are included, so the bigger and more innovative you are, the cheaper our product is.
Foresee: We have a really simple model so we base our pricing on our costs. Cell phones are not based on costs, the incremental cost of a phone call is almost zero. That’s the best pricing model for both sides.
ClickTracks: If you are going to do sampling, you can’t do sampling at the page view level. You have to decide if you are going to sample that entire session or not, but you cannot decide if certain pages get included in the sample.

Q: Where is the line between web analytics and business intelligence?
NetTracker: There are two sides to this question: WA includes BI sometimes, but there is also a very nitty gritty side of WA that is not about BI
CoreMetrics: A lot of our data is being pumped into Cognos, Business Objects, etc. It is surprising that the BI tools have not integrated well with WA. Once the web group is not some guy sitting in the corner, but is an important piece of the business, everyone wants to know about the data and we’ll see more integration.
Maxamine: Consolidation is happening. You’ll see it.

I tried hard to get all the responses- any mistakes are mine, and vendors are welcome to correct me in the comments section here.

Robbin Steif
LunaMetrics

I’m at the eMetrics Summit in Santa Barbara, and there are lots of cool lessons to be heard, some of which I’ll write about in the following days. But today, I’m writing about lunch.

I didn’t really notice what I *ate*, but I sat with Matt Roche from Offermatica (whom I referenced a few days ago but only met today) and Bill Bruno from Stratigent. During the course of lunch, I asked them both what they thought about ABA testing, which companies sometimes use when they have too few visitors (too little data) to do standard split-path testing.

With ABA, the company does an A/B test with three groups instead of two. They randomly split visitors into three groups. One of them sees the test page (the B group.) The other two groups both see the same control page (Those are the two A groups.) When the first A group has the same conversion rate as the second A group, the company who is running the test decides that it has collected enough data, and then feels that it is in a position to declare either the control or the new test page a winner.

Matt and Bill both dismissed ABA testing as a not-too-great idea. Matt drew a picture on the back of Bill’s business card, showing how data usually presents itself:

It is never very clean, he said, and it doesn’t usually go in a beautiful curve, but it bounces all over the place. So in the above picture, the two “A” tests are in black and red, and the B test is in blue. At which arrow should we say that the two A tests are the same — the first one, where the blue is a winner, the second one, where they are all tied, or the third one, where the blue line is the loser?

Robbin Steif
LunaMetrics

A key component of trust is keeping your site up to date - changing the copyright, pulling down content that is clearly dated or event-specific. It’s not that hard to keep the main pages of your site up to date — you probably look at them all the time. It is hard to remember to update less traveled pages, especially when they are response pages that you rarely look at.

Since Future Now is exemplary in every other way (and would surely agree as to the importance of keeping one’s site up to date), I thought I’d start with them as an example. I send customers to their WeWe Calculator all the time, and I even reference it in my What is Link Bait? article. But if you use their calculator and hit the enter button, you get a page like this:

Notice that the lower right corner, that I highlighted in yellow, has dates from 2004.

You know what they say about glass houses and stones, right? You shouldn’t think that I am throwing anything here, because I found these results when I used the (not terribly great) on-site search on the LunaMetrics site:

See the out-of-date copyright? Despite the fact that LunaMetrics has an .asp site (so we have lots of elements that we can change once and they get changed everywhere), our copyright is not included in those “includes”. It needs to be changed on every page by hand - and it wasn’t changed on the response page to the on-site search.

I fixed it before I sent out this post, but it makes one wonder about all the other “hidden” pages. Responses to email sign-ups. Responses to “Contact my Company.” Email responses that are sent out when a customer requests a white paper — maybe the page is up to date but the white paper is not.

Robbin Steif
LunaMetrics

Verizon made it very hard for me to set up an automatic payment plan using their online tools, because their website is outdated. Ultimately, I had to talk to someone on the phone just to set up an account.

After two emails and the phone call, they sent me a “How did we do?” survey. Here is what the buttons in the survey looked like:

Notice how “Next” and “Previous” are not in the most intuitive positions — “Previous” should be all the way to the left, since left connotes “Go back” and right suggests “move forward” (in most of the world). You might be interested in this post I wrote about buttons on the web two weeks ago.

Their third button, “Suspend” is a little mysterious. I assume you use to save your work and come back later. But they should have it on a different line and put enough words into the button so one knows what it is. Matt Roche from Offermatica shared a trick about button wording:

I have a simple rule of thumb. The button text should “finish the sentence”.

For example, if I am signing up for cell service, you could say “Start my service” or “Get my phone”. For shopping, “Place the order” or, frankly, “Buy now”.

Now, let’s look at this one from Quickbooks:

Every time I use this, I am dying to click on “Change PIN.” After all, it’s the big button right where I am looking when I put in my PIN. I wish they would relegate “Change PIN” — a maintenance function — to a small link at the bottom of the screen. The wording is find, it’s the placement that is problematic.

Robbin Steif
LunaMetrics