MMP8: Actionable Web Analytics for Higher Education

[Intro Music]

Announcer: You’re listening to one in a series of podcasts from the 2009 HighEdWeb Conference in Milwaukee, Wisconsin.

Joshua Ellis: Thank you, everyone, for coming out. I'm Joshua Ellis. This is Shelby Thayer. We both work for Penn State Outreach. If you think of college as the regular undergraduate students and graduate students who go to classes, Outreach is all the other stuff; so adult students, online education, public broadcasting. And we both work in the marketing, in the communications department. Shelby works in the strategy side. I work on the techie creative side. We like to joke that I'll tell you how stuff works. Shelby will tell you what it means.

So we're going to divide this presentation in half. Shelby is going to talk about the idea behind analytics. She is going to provide a few case studies. And then I'm going to spend 15 or 20 minutes actually going into Google Analytics and showing how you can do some of the things that Shelby talked about.

So to start out I'm going to turn it over to Shelby.

Shelby Thayer: Hi, can everybody hear me? Of course I just came down with the cold like a day ago, brilliant right? OK, so I'm going to be talking about the strategy side of Web analytics; goals, objectives, things like that.

So first let's get started. What is Web analytics? And that really encompasses three things; competitive intelligence, which is if you've heard of companies like Alexa, Hitwise, Compete, you can get competitive insights into your competitors and then measure yourself against them. Do gap analysis, things like that. So you can get that there.

There is voice of customers, which are your feedback forms, your surveys. Usability testing is really under here, your voice of the customer. And then you have your website behavior on-site analytics, your Google Analytics, Omniture, Web trends, Yahoo! Web analytics. And that's really what we're going to go over today.

Is anybody here running analytics that is not Google Analytics? OK, a few. OK, great. OK, so let's get started.

I don't know if any of you know who Avinash Kaushik is. He's probably arguably the Web analytics guru in the industry. And everybody knows what the 10/90 or some people say the 90/10 rule is of everything. And he coined this for Web analytics.

And what it really says is when it comes to analytics, it's really 10% about the tool and 90% about the analysis and what you do with the data that you get out of your tool. And then fortunately a lot of times in higher ed and in other industries, it's the other way around. You say, "Oh, Google Analytics is free. We're just going to slap it on there." But it takes time to implement and tweak. And you're looking at these things and not really get anything out of it. So 90% on the analysis.

So why do we measure website behavior? Well it really can be boiled down to two things; optimizing your marketing efforts and optimizing the user experience. And for those of you who are not in marketing, you're not marketing, well if your website is indexed by search engines, if you have links to it from other sites, whether they're internal or external, you are marketing your website whether you think you are not. And all of that can be measured and optimized.

So is your website effective? Is it meeting its objectives and how do you know? So let's forget about your website for a second. Let's go back, and I don't know how many people were in Karlyn's presentation right before this. She was talking about business schools, business objectives.

What's your business goal? What's your business objective? If you're in the admissions unit, it might be to increase student enrolments. If you're in the alumni association, it could be to increase alumni association memberships. If you're in IT, it could be to decrease the amount of viruses on campus in students and in faculty.

So you need to have a business goal. And then once you have your business goal or objective, then you talk about what are your website objectives. How can your website help you attain that business goal?

So for the admissions, for increasing student enrolments, one of your website objectives could be to increase online application submissions by 20% this year over last year. That's a measurable objective. If you're in the alumni association, same thing. You could increase alumni donations or alumni memberships online via your website 10 or 20 or X% this year over last year. You always want to give it a timeframe and make it measurable; this year over last year, this month over last month, things like that. So not until you have that objective can you really measure user behavior.

So let's talk quickly about key performance indicators or KPIs. And these are the measures that we use to see if our website is actually meeting its objective. So let's go back to either the alumni donation or the admissions site. So we have those objectives. We want to increase student enrolments or increase alumni membership or alumni donation.

So we come up with our key performance indicators. And when we come up with those, we want to ask ourselves, "So what?" A lot of times you go into your analytics. You look at the reports and you say, "OK, great. We got so many visitors from Ethiopia." What does that tell you? That doesn't tell you anything. Don't waste your time with it.

The so-what test, again, Avinash Kaushik came up with this test. He calls it the three-layer so-what test. So when you come up with your key performance indicator, you ask yourself "so what?" three times. And if at the last "so what?" you can't come up with an actionable step, then it's really not a good key performance indicator.

So I know a lot of administrators or whatnot love to look at that map and see where visits are coming from. So let's see, you got 100 visits from Australia or something. And they are all excited about that. But you go back and say, "OK, so we got 100 visits from Australia. So what?" And if they can't answer, "So what? Let's do this," or "Let's take an actionable step," it's not a good KPI for you. Forget it, don't worry about it.

So on the flipside, for increasing student enrolments or increasing alumni donations, looking at a key performance indicator like conversion rate, that's measurable. So what? When it goes up, you can take an action. You can do more of your marketing or whatnot. If it goes down, you can increase your marketing effort. You actually have an action to take.

An easier way to look at this might be if your KPI increases 10% or decreases 10%. OK, so if our conversion rate increases 10%, who are we going to go to and congratulate and pat on the back and say, "Awesome. Keep doing what you're doing"? Or if it decreases 10%, who are you going to run to and demand answers? And if you're not going to run to anybody and you're not going to take that action, then don't worry about that KPI. Move on to something actionable.

So let's dig a little deeper into KPIs. One thing that's critical is segmentation OK? So let's say that one of your website objectives for this year is to become more iPhone-friendly. You're seeing an increase of usage of iPhones on campus. You really want your websites to become iPhone-friendly.

So one thing you want to look at before you've been coming up with that objective is your trends of people coming in via iPhones. So let's say you only have about 3% of your visitors coming in via iPhones, that doesn't seem like that's very small. But if you look at the trend and you say, "Well six months ago it was less than 1%," well that's a pretty significant trend. So six months from now it's going to be maybe almost between 5 and 10%.

So you want to look at the trend. If the trend is there, then that would be a good objective. If the trend isn't there, you can say, "Well being iPhone-friendly, it's great maybe for next year. But really right now let's not focus on it."

So let's say the trend is there, we want to be iPhone-friendly. So one of the KPIs or key performance indicators that you could come up with is page views per visit. So page views per visit is great but it really doesn't tell you anything about iPhones. So what you want to do is segment out your iPhone visitors from your non-iPhone visitors and then compare their page views per visit. So if your non-iPhone visitors are 10 page views per visit and your iPhone users are one, you know that you have an issue there and that's an actionable step that you can take to solve that issue.

So we've talked a lot about trends. Trends are key with analytics. A lot of times we go to our bosses' office and we say, "Hey, we got 100,000 visits today or 5000 or whatever it is." And OK, so what does that mean? You need to put it in a trend. You need to say what's the difference between going in there and saying "We've got 5000 visits today" or "We increased our visits 60% over yesterday". That's a much more actionable and much more tangible thing. You really want to put it in the trends.

Now that's not to say that hard numbers are bad. Hard numbers are good for context. So if you're running a marketing campaign or if you're not running a marketing campaign, and let's say you've sent out an email, you got clickthroughs to a landing page and people bounced immediately OK? So you have that campaign. You have a 40% bounce rate. So remember bounce is when somebody comes in one page and they leave immediately. That's a bounce rate.

So campaign rate has, let's up it, let's say 60% bounce rate, pretty bad right? Well campaign B has 10% bounce rate. That's pretty good right? But you bring in the hard numbers for context. So if campaign B with the 10% bounce rate only got 4 clickthroughs and campaign A with the 60% got 500 clickthroughs, that really brings in the context. This really isn't statistically significant. So you can't really say that this is a great campaign. So that's where hard numbers are great.

So optimizing marketing efforts. And I'm going to go through a couple of case studies that we've actually done internally. One thing that I really like and one thing I really write on my blog and everything is going beyond the clickthrough. A lot of times everybody just looks at clickthrough rate. We got this many clicks from this email campaign. If you run Google AdWords, we got this many clicks from our Google AdWords, our paid pay-per-click words. And that really doesn't tell you the whole story. You can get a million clickthroughs. But if they're not doing anything on your site, then it doesn't matter.

So about a year and a half ago we launched a Geospatial Intelligence certificate. And as part of that launch, we obviously launched a marketing campaign to coincide with that. And part of the marketing campaign was Google AdWords. We did keyword research. We launched a Google AdWords campaign. And a few of the keywords specifically were bringing in a ton of traffic. We're talking more than 4 or 5000 clickthroughs per week. So this was a ton of traffic. So we were really excited.

But again, going beyond the clickthrough, we were seeing that there was a huge bounce rate. And unfortunately there was a big sag on conversions. So that's not very good. So what we did was we came up with some assumptions. Our number one assumption was that the landing page was terrible. So they're going to the landing page. And the landing page must not maybe the Apply Now button isn't in the right place or whatever. So our assumption was that this landing page was not good.

But what we did was we dug even deeper than that. And we looked at internal site keywords. So we segmented out just those people coming in from that campaign. Remember this is Geospatial Intelligence with a few ad words that were bringing in a ton of traffic. We segmented out just those keywords from that campaign. And then we looked to see if people were actually searching in our internal site keywords and what they were searching for. And this is just a very small subset of what they were searching for.

So as you can see by this, our assumptions were completely wrong. This was actually very poor low-quality traffic and we didn't want them. Our landing page was fine. Other things were fine. This was not our audience. So what did we do? Obviously we stopped those Google AdWords for those specific keywords.

But what if we didn't go beyond that clickthrough? First of all, we would have poured way more money into those ads because we were getting a ton of traffic from those ads. And that's hard costs.

But what if we didn't actually go beyond the bounce rate and the simple conversion? What if we didn't segment out those people and see what else they were doing? We would have continued to tweak our landing page, which is soft costs, the design time, the copywriting time, the development time. So we would have lost all of these time and effort into basically a lost cause because this was not our audience. So if you're not in marketing, you can also use this to optimize the user experience.

So can users do what they want to do and how do you know? So let's look at fallout. And if you were in Karlyn's session right before this, she had an example of this as well. So fallout is basically you have a link to a forum. You have the forum and then you have a Thank You page. It could be 10 pages long. It could be 3 pages long.

So if your visitors are linking to the form and then they go away, your fallout is on the forum and likewise. So you see where your visitors are going away and that's your fallout.

So a while ago we had a conference that was really well attended. And we decided to after the conference put up some MP3s of some sessions for a small price so other people could get the sessions. And we put them up on the website. It was a really niche group but it was a very popular conference. So we thought it was a good idea. And they were very cheap. They were like 99 cents or whatnot.

So we put this up and we saw what happened. Now after about a couple of weeks, we saw that we only got two people that actually bought these podcasts from the conference. And we thought, "Wow! OK, well maybe we shouldn't have charged them. Maybe it was a bad product they weren't really interested in."

But what we saw was that more than 100 people actually clicked a link to the download but only two people actually got through the little checkout process for downloading, which is the 98% fallout rate. Yeah, that wasn't too good.

But that was good because we went back to that shopping cart process. And we saw that it was very cumbersome. It asked for shipping information. They're downloading an MP3. [Laughter] But it gave us actionable steps. "OK, we need to fix this right here." So that's what we did.

So let's talk quickly about usability testing. Everybody does usability testing. It's fantastic; everybody should. But how do you know if it worked? You measure before and after. You measure, again, using your objectives, using your key performance indicators. You measure what's happening before and what's happening after.

About a year ago we re-skinned our homepage. Our objective on re-skinning the homepage, well, it was two things. Specifically for navigation, we were seeing very low clickthrough rates on the left-hand side. You see they were, the links were within the paragraphs. We wanted to make those more scannable, readable. And the second is with the lion, we wanted to instead a lion have a picture of a student and give it an actual face.

So this is what we came up with. And we found that just from bulleting the left-hand navigation, our clickthrough rates skyrocketed, double, triple and in some cases even quadruple the quick-through rates just from bulleting.

But again, we wouldn't have known that if we didn't measure before and after. But what was interesting is we also saw that our update section which on the old page was like up here, our update section, their clickthrough rates were plummeting. So we decide to simply flip our update section on the right and the clickthrough rate went back up. And again, we wouldn't have known that if we didn't measure before and measure after.

So really quickly, before I hand it over to Joshua, free usability tests that I love are referring keywords, internal site search keywords and if you use a knowledge-based, those keywords. So referring keywords, keywords coming from your search engines. You can see what you're relevant for on those search engines. And if you're not seeing the words that you want to be relevant for, well then maybe you want to beef up your search engine optimization efforts.

Secondly, your internal site search keywords are fantastic to get those terms that actual users are using. So if users are using pricier costs and not tuition, maybe you should change that term. I think in higher education we come up with a degree name that's 15 words long that users don't know what the heck it means. Maybe in our H1 tags we have to not use the actual name. Maybe we just use the name in the lower so they can see it.

Same thing with our knowledge base. If you have a knowledge base, you can get the exact same insights about that. And remember that this isn't just for marketing. If you're an IT and you want to decrease viruses across campus, if you're running an internal marketing campaign to your campus, you can actually get conversion rate to your download of your antivirus software or anti-spyware. And that's measurable. You can see how many people are downloading that, what's the percentage of your campus of those viruses. You can get that. That's measurable, attainable.

Another thing, if one of your objectives is just for brand awareness, a lot of people say, "Oh, we can't measure brand awareness." Well you can with this because with brand awareness, you want to look at your referring keywords as your branded keywords. Are the rate of users using your branded keywords actually increasing over time? Or if you're running a brand campaign, measure before, measure after. Branded keywords and also direct referrals. People typing in your URL or bookmarking. Those are all branded things. So you can measure brand campaigns this way.

So I'm going to hand it over to Joshua. And he's going to tell you how to do it.

Joshua Ellis: All right. Thanks, Shelby. So I'm going to spend about 15, maybe 20 minutes, going over a few of the features of Google Analytics that are really focused on getting the type of information that Shelby was just talking about. Specifically I'm going to look at segmentation, referring keywords, the internal site search keywords, the site overlay or what sometimes is referred to as the heat map that lets you see where people are clicking on a Web page. We're going to talk about events and then also funnels.

And we're not going into how you implement Google Analytics on your site. If you're already getting reports on Google Analytics, here's how you can use some of the more advanced features to get the actionable information.

So I'm going to actually sit down for part of this because I'm going to try not to make this look too much like a software demo but it's what it is. OK, I'm going to flip over to Google Analytics here.

OK, we're going to look at segmentation first. So this is your Google Analytics dashboard. And this is actually the information for the last month on our Outreach Internet. About six months ago we launched a new Internet for Outreach. It has got all the cool wiki social media-type features. We really like it.

One of the things that we're using as a performance indicator on this to see if people are actually using it and how useful it is, is how much time are they spending on the site when they visit the site. And we're looking for trending on this. So this is the average time on site right down here. We watch for trends on this.

So we had this number almost six minutes average time on site. That's a useful statistic. I'm actually going to jump back up here and go back to the first month of our launch, which is February. OK, so it's about six minutes. So we have this number that we can look at trends on.

What segmentation allows us to do is actually drilling it and get some information that might be hidden. Right up here in the corner of almost all of your reports in Google Analytics is this Advanced Segments box. And this allows you to pick some segments. There's some really good default segments. New visitors, returning visitors, paid traffic, whatever.

I'm going to look at new visitors versus returning visitors. And then I'm also going to grab iPhones and we're going to apply this.

Now if we go back and look at this average time on site, we actually see some information that was hidden, looking at the big aggregate number. And that is the first time someone visits the site, they're only sticking around for about a minute.

There's a lot of stuff on the homepage. There's a huge menu. It can be a little overwhelming. Once our people get used to the site and they know how to use it, we're seeing about a 10-minute average visit. This 10-minute is very different than the 6 minutes we were seeing. And we can watch those trends. And then we also would hope to see that if we're doing a good job with user education, the length of time on site for our new visitors would increase.

Shelby alluded to the iPhone issue. So one of the things that came up in the discussion of the launch for our Internet was how friendly do we want to be for iPhones? How much effort do we want to put into it?

So the average visit length for iPhone users was 17 seconds, which is basically enough time to log in to our Internet and realize it is completely unusable with an iPhone and then log back out. So this is something that could be alarming if this was something you really wanted to concentrate on. But again, this is where some of the exact numbers provide context. That was 17 seconds but it was only seven people over the month who tried hitting our Internet with their iPhone.

So while I think this is something we will want to address eventually. For seven people in one month, we probably can spend in investor effort more wisely elsewhere. So this is some of the out-of-the-box ones. Google Analytics also allows you to create custom segments. I am going over to Advanced Segments over here and I am going to actually delete the ones that I created before and show you how this works.

Yes, yeah, I'll just blow one of my jokes later.

So I am going to create a new custom segment. And the custom segment builder gives you all sorts of things along the left-hand side here that you can measure and then you drag them over and then you can create comparisons or filters and you can make this quite complicated.

One of the things that we care about is internal versus external traffic. So I am going to create a segment for internal traffic. So for us this would be network location matches the Pennsylvania State University. �I'm going to call this "internal traffic" I have to make sure this works. OK, so I create this segment. I am going to add the opposite segment then. So I am going to say network location does not match the Pennsylvania State University. I'll call this "external traffic".

OK, yes?

Audience 1: Quick question. Did you have to do something to sustain the IP?

Joshua Ellis: No, that is one of the nice things with these segments here. So it will actually use your existing analytics data to take a lot of that guess work out. So if I go into network location, based on the reverse lookup information it already has, it will give me that information there. So here's all are local cable providers and the thing for our hotel, those sorts of things. So that is all there for you.

So I am going to go back to my dashboard here. And I am going to change the segments I'm looking at here from the new one and returning to internal versus external traffic. So this is our employee Internet. And we are seeing that may be a little better than a quarter. Maybe 25, 30% of our traffic is actually coming from outside our hardwired Penn State network.

Now we know some of this is from sick twisted individuals who do work while they are on vacation or at conferences. But for our particular organization, we also have a lot of county extension agents in every county around the state who will be interacting with our Internet via whatever network connection they have.

So let us go back to that so-what question. Well, because we know that 25% to 30% of our traffic to the Internet is coming from outside our hardwired Penn State network, that has implications for the types of content we can have on there from say FERPA or a financial perspective, if this information is being transferred over an unsecured channel. �If we are going to start having a lot of sensitive information, we may look at this and say, "OK, if you're going to hit the Internet, you have to be VPN it."

This also has some implications especially for our folks out in the field in terms of the amount of things that require high bandwidth because we know we have folks coming in not with dialup but maybe just a little bit better with shared DSL lines, those sorts of things. So this is the type of information that you can get with segmentation that you can't get just by looking at your raw numbers.

So some other possible segments. The segments that you want to look for really depend on your website objectives and how you're going to measure those. But some ones that we have found useful for ourselves are internal versus external traffic, in-state versus out-of-state traffic and looking at the behaviors there, depending on our marketing campaigns and the people we're trying to reach for our programs.

Search terms, which I'll get into in a little bit. But like Shelby talked about with the geospatial intelligence example, we created the segment based on people who were coming in through certain terms or certain campaigns. And we are able to look at their behavior.

Frequent flyers is another one. If you have a site that you want to encourage people to come back, that you want the site to be sticky, you can set up segments that would say, "Let me look at the behavior of people who have been to the site more than 10 times in the last 30 days versus the people who have been to the site less than 10 times in the last 30 days." Where are these two groups of people? What are they doing in the website that is different? Where are they going that is different? What are they looking for that is different? What is causing these set of frequent flyers to spend more time in the site versus these other folks?

So that's segmentation real quick.

OK. So the next thing I'm going to look at is the site overlay or the heat map. And this is a really good tool but like Shelby talked about with that before and after for the site redesign. And I'm going to this particular site. This tool in Google Analytics is great. It is not perfect. And if you have a very AJAXy-type site with lots of Javascript, this tool breaks sometimes. But if you have a landing page or a fairly static page, this is great.

So I'm going to go to the site overlay for this particular site. And my department takes no credit for the design of this site. So the site overlay allows you to get a real quick visual representation of where people are clicking on your sites. So I am going to hide this over here. This is what the site looks like. We got links everywhere. You can see the overlay and you get these little bars that tell you where people are clicking.

Where this can be useful is a page like this which is a staff directory. We look at where people are clicking. And almost 40% of the people are clicking right away to the All Staff link. Now this is listed in an alphabetical order but we have links on here where nobody is clicking on them whatsoever.

So this might be an opportunity where we could maybe group some of these links together. We might change the order, instead of going alphabetical maybe go by topic area, and see if more people are going to a specific category to get a listing of people as opposed to hitting that entire employee list right away to find the person they are looking for.

Another place that can be useful is if you have a specific purpose for your page. Now this is a news and publications page. And if the purpose of this page was to host a bunch of annual reports that nobody ever downloads, we have succeeded admirably.

But the other piece of information we have on here is a listing of job openings and that is the most clicked link on this page. Sadly enough we have more people using the breadcrumbs to get back to the home page then we have downloading the annual report. So this is somewhere we may want to rethink the page and create a page where it is easier for people to find what they're looking for.

The second most clicked link here is request assistance. In a redesign, this may be an opportunity to actually move those in the primary navigation. Here is the job listings. Here is request assistance. And if you want the annual reports, which apparently no one does, you can still get them here in this page. All right, so that's the site overlay.

I'm going to go into some of the keyword reports now. So your keyword reports are the traffic sources and the keywords that are bringing people into your site. Your content targeting, that's your paid advertising. So that's your pay-per-click, your Google AdWords, your banner ads, those sorts of things. And then that's compared to the actual keywords that people are typing into your search engines that bring them to the site.

So this is actually an interesting comparison if we look at average time on site and in some of the statistics for our content targeting. Now this is traffic that we are paying to get to our site here. They spend an average of 48 seconds on the site. We have a 71% bounce rate. If we look at some of our conversion rates, we have two-tenths of 1% conversion rate from the traffic we're paying for to get to the site.

If I go back and compared that to our branded keywords, this is statistics for our management development office and they offered like project management trading and those sorts of things. Our branded keywords will be like management development, Penn State management development.

Look at the numbers we get for our branded keywords here. They're spending 7 minutes on the site. Our bounce rate is down to 14%. If we look at our conversions, we are getting about a 1% conversion rate here, much higher. If we look at the full name Penn State management development, look at our conversion rate, we are getting almost a 16% conversion rate. So the people who know what they are looking for are coming to the site and they are taking the actions that we want.

Now the sad thing is in terms of our pay-per-click traffic, we have been looking at this recently. We have some months where we are paying more for pay-per-click traffic than those people who we pay to bring to the site or actually generating a revenue for that particular month.

So this is somewhere where we can take action on this, the so-what. The so-what on this is we probably would be better spending a little bit more time investing in our search engine optimization, getting our search engine ranks up higher. We come in I think around 25 on the keyword management development but we know that is good traffic. So that is a so-what there.

Related to this, our internal site search keywords. I'm going to go back to our Internet here. Internal site search is really useful for telling you what people are looking for on their site that they can't find obviously.

So a quick history on our Internet. We used to have this old one called My Outreach. And then we created this new one, we called it Our Outreach. It was really cool. The reason we replaced our old one is after all the effort that was going into creating this nice customizable portal, the only thing people used it for was getting to webmail, that's all.

[Laughter]

So one of the things we needed to make sure that we didn't break was the ability of people to get ��to webmail from this new Internet. So this is something we watched for, our internal site search key terms. This is actually getting much better. But the number one thing people are searching for on the site is email. The second one is My Outreach, the name of the old Internet which has been gone for six months but that is where people knew they wanted to get to to email.

Number two is Yuengling, which I think is a type of beer. �So I am not sure what's going on with that. But this is something where if we wanted to look at the behavior of people who were searching for that, we can go back into our advanced segments. We can create a new custom segment. And we can create a segment where people are searching for the keyword Yuengling and we'll call these beer seekers.

And now on any of our reports, we can compare the behavior of beer seekers to the rest of our traffic to see what they are doing. Apparently people who seek beer spent a lot more time on our site than the general public. Yeah, I am not sure what action we want to take on that one but that is how that works.

I am going to go real quick into goals and funnels. So each Google Analytics profile can have up to four goals. And those are expressed in terms of a page. You have like a thank you page or something. You set it up as goal and you want to measure how many people go to that page.

On our Internet we have a goal that is set up. One of the things we want people to do is update their profile, put their picture on it. We've got 1,500 people all over the state. They don't meet each other that often. So we measure how often people are updating their profile.

I am going to go back to the month of August real quick. We had an event for our continuing education staff in early August. And one of the things that we had was that people came in all from across the state. We had a photographer there to take pictures. We emailed the pictures out and we were wondering, "Was this worth it?" Well in a week and a half following that, we had about 95 people updating and using their profile. So that was something to us that said this actually worked.

I'm almost out of time here. I got two things to cover. You can also expand the goal to a funnel. And I will show a quick funnel visualization that looks like this. So I am going to go to this. So we define the funnel. These are really good for measuring process and seeing how successful they are or where people fall out.

So we set up this funnel that we want people on our management development site to come to the home page, go to this program page and register for a course. This is actually a funnel that works pretty well because we have 21% of the people who actually gets that page register for the course. So our action there may be to drive more people to that page in there.

The last thing I'm going to cover, because I'm almost out of time here, is events. There are other things that you want to track on your site besides just what pages people are looking at. And event tracking is something that allows you to do this. We'll have it on the website. There's this cool plugin called the Wasp, Google Analytics plugin. It's a one-liner Javascript code you add to your page and it automatically tags all of your links and unclick things as events.

Someone asked us the question, "I have this page on my site. It's a lot of length. There was a library site going out to other sites. I've got like a 95% bounce rate on this site. I don't know if this page is working or not." This allows you to see where people are going, what external sites people are going to when they click on things.

So this is our Internet. Number one of course is webmail. Number three another way to get to webmail and then number two is our employee monthly attendance tracking. So if you have a site that then sends traffic out to other sites that you can't get analytics from, you at least can see where they're going and look at the quality of traffic that's leaving.

So that's all the time. Do we have any time for questions? Yes?

Audience 2: When you have the page dollars, obviously you send PDFs.

Joshua Ellis: Yeah.

Audience 2: One thing I learned from watching things on Sunday was the way things are. They only faxed the number of kids who had gotten to that page, they can't track it, some PDFs. We had to do something special to make it know that you are going to a PDF like a question mark or something else?

Joshua Ellis: Yeah.

Audience 2: Are you doing that?

Joshua Ellis: Not on that particular one. But we know from event tracking that no one is still downloading the annual report, though. Yes?

Audience 3: One more. When we set up out analytics, we put the filters in this little internal drive cause they didn't want to show it on the main site.

Joshua Ellis: Yes. Because if you put a filter on that, the record of that traffic is gone. So my recommendation is always to capture everything you can and then use segmentation to split it out because that way you can ignore it if you don't want it. But if you would ever want to get that internal traffic back, if you filtered it out, you can't; it's gone.

I guess that's it. Thank you.

[Applause]