Higher Ed Analytics

As you’ve probably seen (or heard) by now, a new version of Google Analytics is available to use. The new version brings many new advancements to reporting capabilities by allowing more flexible ways to look at data through a cleaner, faster user interface.

But Google didn’t stop there. They’re also adding new functionality to the tracking API. One of the new features is the ability to track page load times. Once setup, you will now have access to a sampling of page load data. It appears Google is only tracking the page loads for about 2% of visitors, so it will take quite a few pageviews to get an appropriate sample.

The Importance of Page Load Speeds

Tracking page load times can provide valuable insights. For instance, does a page with a high bounce rate also have a long page load? If so, you might have your reason for the high bounce rate. What about those users in rural America, are they experiencing a increased page load? If so, maybe it’s time to trim the unnecessary files. Maybe Internet Explorer is taking longer to load than Chrome (this is likely true). This is important to know when building new pages.

Furthermore, site speed is one of Google’s primary metrics when determining page rank. In fact, Google builds tools into Chrome and offers many more tools to help developers lower the page load times.

Installing the Tracking

This is very easy, and it’s available with one line of code (that’s the part in bold):

<script type="text/javascript">
 var _gaq = _gaq || [];
 _gaq.push(['_setAccount', 'UA-XXXXX-X']);
 _gaq.push(['_trackPageview']);
 _gaq.push(['_trackPageLoadTime']);

 (function() {
    ...
   })();
</script>

More information can be found at the Google Analytics help section.

Search Engine Optimization (SEO) is a crucial part of a successful online strategy. If we have worthwhile content, we want our users to find it. In increasing fashion, our audiences are turning the to likes of Google, Bing, Yahoo! and other search engines to help them find our content. To help us, Content Management Systems are “baking-in” SEO best-practices and helping to close the gap between what needs to be done and what can be done.

Once we’ve implemented SEO strategies, it can be cumbersome to determine success or failure. The most elementary solution is to query the search engines on a regular basis and observe if a position has moved up or down. This audit requires an understanding of keywords to focus on (defined in your SEO strategy) and time to do these searches and catalog results. Not to mention the possibility of localization which could vary your results from users across the country.

It’s time for a better solution: use Google Analytics to track keyword position when our visitors click on our result in the search listing. With a quick setup, we can track the actual item rank based on keyword from Google listings.

Continue reading →

As many of you know, I led the 2010 HighEdWeb Google Analytics for Higher Ed workshop. It was a great workshop, and we looked into a lot of great details. I’ve heard from many of the attendees since then, and I’ve helped quite a few get through some sticking points. Along the way, I realized that I never posted my presentation! I apologize for that, and am now fixing my error.

Feel free to leave comments, or contact me if you have any questions.

Avinash Kaushik, THE expert in Web Analytics, gave the opening keynote at the SIM Tech Conference last week (Oct. 21st, 2010) in Las Vegas, NV.

Avinash Kaushik at the SIM Tech 2010 Conference Photo by Michael Fienen

It’s not often that you get a chance to listen to one of the best minds in Web Analytics trying to help higher ed institutions stop making faith-based marketing decisions and switch to a more data-driven approach.

And, it was a real treat (pretty nerve-racking as well as I was scheduled to present about the Higher Ed Analytics Revolution just after his session – and he sat in the back of the room during the whole presentation we gave with Shelby Thayer).

Later at lunch, when I asked Avinash if he would answer a couple of questions via email to share some of his insights with folks who could not attend SIM Tech, he agreed right away. He actually got back to me in less than 8 hours after I emailed my questions a couple of days later.

So here’s your treat: an opportunity to hear the truth about the State of Higher Ed Websites from the Master of Web Analytics with this short interview of Avinash Kaushik.

1) Avinash, in preparation of your SIM Tech 2010 keynote you reviewed many higher ed websites. So, how is higher education doing when it comes to websites? What did strike you?

I hate to admit this but it was mostly heartbreaking. I went to at least 50 websites of education institutions ranging from the really large universities to small private colleges to truly niche outsiders. Even a cursory review would indicated that most websites engage in this type of online marketing:

1. Shout Marketing. “We don’t care what you want or consider any signals of relevance you might provide, we’ll just shout at you about what we want to “pimp”.

2. Offline Marketing. “We know how to print glossy brochures, and look how cool it is that now we can put all that online exactly as we did in the offline world.”

3. Unimaginative Marketing. “Our site was created in 1990 as soon as the web got cool. We are really working hard to figure out what has changed since then. Meanwhile here’s our 1990 site with the addition of Follow On Twitter & Facebook buttons.”

Sad right?

Let me hasten to add that lots of non-education websites do the above type of marketing, many Fortune 500 websites are still that sad.

But higher ed is where everything cool and now and imaginative and killing of the lame happens. That made the sites by universities so heart breaking.

We need a fresh infusion of people who truly get online marketing. We need a fresh infusion of people who understand the agility and awesomeness of the web. We need a fresh infusion of people who understand how to use data on higher ed sites to experiment, execute and fail faster.

2) You gave great insights and tips to the conference attendees on how to use Web Analytics 2.0 to improve their websites. Can you share your most important point for those who were not in attendance?

My singular hope was for the attendees to understand that Web Analytics is not simply analyzing clickstream data spewing out of Google Analytics / Omniture / WebTrends etc. Web Analytics means understanding the What, How Much, Why and What Else.

All that translates into an obsession with identifying Macro & Micro Conversions and computing Economic Value (yes even for Higher Ed sites!!). It means listening to the prospective students and website users by being agile and nimble in using Surveys, Online Usability, Testing and doing so at scale (UCD and HCI are integral to Web Analytics 2.0!).

And all this does not have to be hard.

Avinash Kaushik at the SIM Tech 2010 Conference Photo by Michael Fienen

Every tool you could ever want to have, from clickstream to mobile analytics to testing to surveys, can be had for free or cheap. See: Best Web Analytics 2.0 Tools.

You don’t have to do all of it overnight. You execute in small steps and in a particular order of priority (see above post). Baby steps. Lovely amazing exciting baby steps.

And you can start fixing things today. It takes less than a day to identify your top landing pages with high bounce rates, or ensure you know how horrible your internal site search engine is (takes 5 mins to configure Analytics to get that data), or start tracking three key goals, or type some of your top queries into Bing and Google related to your higher education institution and see how good or stinky your customer experience is (pull some of your hair out, it is ok).

Start today. Start small. Fix things.

Rinse & Repeat.

We’ve been asked a lot lately about how we came up with the 12 metrics, what they mean and why they’re relevant to us.

Before going into the 12 metrics, remember that any benchmarking is a guideline – something to help you get started on your analytics journey. Be sure to benchmark against your own site as well. Your own numbers will give you the best benchmark. There is a time, however, when looking at an industry benchmark can help guide you, especially when you’re just starting out. This is our goal – to get your started.

Why don’t we just use the built-in Google Analytics benchmarks? For a couple reasons, but the biggest one is segmentation. The GA benchmarks throw all sites in together – making the averages even more … average. Through our benchmarking project we aim to segment wherever possible so you can compare like sites with like sites. Google does compare sites of similar sizes, however, we want to go farther. If we get enough response, we can compare admissions sites to admissions sites and alumni sites to alumni sites. If a college website is all-inclusive, then we can compare those as well.

OK, now on to the metrics. Keep in mind that most of these metrics are certainly just that … metrics. They are not key performance indicators (KPIs), especially the first 5 or 6.

Very high level metrics – is our website even breathing?

• Total Visits: This metric is a building block. We need visits to have outcomes.
• Total Page Views: Like visits, in most cases we need page views to have outcomes – again, another building block. Page views by themselves don’t mean a lot. However, coupled with other metrics (like visits), you can get a better understanding of engagement. Be careful: page views are becoming less and less relevant on the web. Also, higher page views per visit don’t necessarily mean engagement. It can also mean your website navigation is confusing and things are hard to find.
• Average Time on Site: Time on site can be a very high level gauge for engagement as well; however, the same thing applies to ToS as does to page views per visit. A longer time on site isn’t necessarily a good thing.
• Average Bounce Rate: In the words of Avinash, this metric tells you that the visitor “came, puked, and left.”  This metric can tell you at a high level if you have quality traffic or not. The bounce rate of your top landing pages can give you key insights into the quality of those pages (or the relevancy of the page to the type of traffic that lands there – great metric to look at for campaign landing pages).
• % New Visits: Is your website good at getting new traffic or is it just the same returning visitors all the time? If you’re in the recruitment game, this metric is relevant.

Here’s where it starts to get good …

• Total Direct Visits: Again, in the words of Avinash, this metric is your BFF. This is brand engagement! How many people know you enough that either a) they’ve bookmarked your website or b) they type your URL directly? Word of caution: filter out those people you know already know you (current students, staff, etc.). Also, be sure to tag all your campaigns. Otherwise they will come in as direct traffic as well.
• Total search engine visits with *branded* search terms (or % branded search terms): Key for admissions and other websites where “branding” is may be a goal. This metric coupled with direct visits can give you brand engagement that can be used before and after campaigns.
• Visitor Loyalty: How many times does a visitor visit your website? Visitor loyalty answers that question. This is a great metric to use if you have a long buying cycle or if loyalty and engagement are key to your website success. Rarely do people convert on the first visit. They need to come back. Use this report to show you how many times they do. Use this with campaigns as well. A month after a campaign, do visitors come back or are they one-hit wonders?
• Visitor Recency: Now that you have loyal visitors, how long is it before they come back? Daily, weekly, monthly? Again, if your website has a long buying cycle, there may not be high visitor recency, it may be more medium (a few days or a week). If you update your website daily with a blog post, IT alerts, or any important daily information, then you’re looking for high recency.
• % traffic from admissions homepage to application form page: This is our stab at some kind of conversion … for admissions websites. Although this isn’t a perfect metric by any means, we need to start looking at outcomes not just traffic. If you don’t run an admission website, think about the outcome on your website. What is your conversion rate? This metric will surely evolve – possibly into simply conversion rate. Stay tuned.
• Total visits from social media sites: Social media ROI is all the rage now. Everyone wants to know “is it worth it.” Just using URL shorteners and measuring traffic isn’t enough. We need to measure outcomes. By creating a social media segment, we can make a first step toward looking at outcomes. There are definitely flaws with only creating a social media segment and looking at that. Be sure to get out of your analytics tool and use other tools to measure social media as well.
• Total visits from mobile devices: Creating mobile websites is popular now as well. Depending on your audience, this is very important. What is the usage, though? If mobile device usage is high for your audience, is your website mobile friendly? By using an advanced segment for mobile devices you can find out quickly.

This week, I’ve had the opportunity to attend HighEdWeb, the 2010 edition. During this conference, I’ve been excited to hear institutions are experimenting with and leveraging the new technologies available in HTML5. Whether it’s Christopher Schmitt’s presentation, casual conversations in the hall or even discussions with some of the vendors, HTML5 is on the minds of many developers.

While HTML5 promises to bring much-needed advancements to the web, it is still only in working draft-mode and has not been officially adopted as the new standard. Nonetheless, many browsers have already begun to implement these technologies. For example, the webkit browsers Safari and Chrome have geolocation. Firefox 3.6, webkit and even IE9 have support for the HTML5 video element.

However, none of the browsers support every HTML5 element. Furthermore, the matrix of which browsers support which elements is large and cumbersome. findmebyIP.com has done a great job of presenting this information, but it would be great if Google Analytics could track this for me.

Well, Google Analytics can track this for you! There are two ways to do this: 1) Report on individual browsers and version, compare to the findmebyIP.com tables specific to a particular HTML5 element, and determine if you’ve hit a threshold worthy of utilizing the technology; or 2) use my new script to track this inside Google Analytics for you!

trackHTML5inGA

Today, I’m releasing an easy-to-implement Javascript code which will utilize the powerful Modernizr HTML5 detection library to track specific elements inside Google Analytics. It’s a simple implementation:

1. Download the latest files (available with dual BSD & MIT license, so fork away!).
2. Make the Javascript files available on your server.
3. Add the JS code below to your pages (preferably alongside your current GA implementation). Check out the example.html file for an, well, example.
4. Change the “featuresToTrack” array to allow for the elements you want to track. A list of features is available in Readme.textile or on the github page.

<script type="text/javascript">
    featuresToTrack = ['video','audio', 'geolocation', 'localstorage']; // set this array with the features you want to track. **Note max character count is 55 inclusive
    (function(d, t) {
        var s=d.createElement(t),x=d.getElementsByTagName(t)[0];
        s.async=1;s.src='js/trackHTML5inGA.js';
        x.parentNode.insertBefore(s,x);
    })(document, 'script');
</script>

A few things to keep in mind with this code:

1. This is experimental. While I have tested in FF 3.6, Chrome 6, Safari 5, IE 7 & IE 8, and everything is working correctly, individual implementations may have issues. Let me know if something goes wrong, and I’ll do my best to help.
2. This will only work if you’re using the latest Google Analytics Asynchronous Tracking. If you’re not using Async, this script will not only fail to work, it could cause JS errors.
3. Due to character limits in Google Analytics Custom Variables, only a few items can be tracked (total character length needs to be less than 55, inclusive of all tracked features). So pick your items strategically; though, you can always change which items you want to track at any time.
4. Speaking of custom variables, this will be stored in slot #5 of GA. If you don’t know what this is, don’t worry. If you do, and you are using slot #5, go ahead and change the script.

Where are the Reports?

You will find these reports in your Visitor Tracking, available under Custom Variables (Visitor > Custom Variables). There will be an entry titled HTML5, and drilling-down farther you will have a list of possible combinations of tracked features based on browser support. From here, you can build Advanced Segments that will help you to better understand visitors to your site with certain HTML5 capabilities and how this trends over time (I see another post in the near future on this topic alone).

Comments, thoughts, suggestions

If you’re having a problem, or you have a better idea, drop in a comment. If you’ve found this helpful or have some examples of how you’re using it, let the community know also with a comment!

Web analytics provides quick and easy access to a rich amount to quantitative data. All of this data becomes power at our fingertips as we begin to understand how our content is being surfaced, how it is interacted with, and in many cases, who is interacting with it.

However, there is an inherent problem with quick analytics: it doesn’t always spell actionable data. In many cases, our web analytics environment is setup to spit out ‘averages’ for many statistics, such as “Average Time on Site” or “Average Pageviews.” So when we see a statistic that says, on average each visitor consumed 6 pages, we assume most visitors have consumed six pages.

Unfortunately, we don’t always have ‘average’ visitors to our sites. In many cases, we’re dealing with a wide-range of visitors, from prospective students to alumni, faculty and staff. Each of which use our sites for different purposes. Also, in many analytics packages Bounce Visits and Exits are excluded in calculations (for technical reasons), which can drastically change metrics.

Instead, we should be focusing on ‘median,’ or the point where exactly 50% of the data is above and 50% is below. In other words, we’re looking at the middle, instead of an area to far to one side or the other. Unfortunately, we don’t always have quick access to median, and must export data and calculate locally.

How do we get around this limitation?

My recommendations:

1. Segmentation: Drill down to one particular audience type. You’ll likely find that in many cases different audiences interact with your site differently.
2. Calculate the median: Export your data and run these calculations manually. You can even do this after segmentation. In some cases, a media-esque report is available in your analytics account (see Depth of Visit).

Let’s take a real case situation with averages. When looking through a client’s analytics data, I found a page with a very high amount of pageviews that also had a very high time on page (normally I discredit low pageviews and high time on page). This peaked my interest and we setup a few advanced segments to look into this a bit more. Long story short, we found that this page was being accessed mostly by a developer who had setup a page refresh every 10 minutes — thus negating the bounce or exit exclusion. (BTW, his reasons had nothing to do with analytics and were legitimate, albeit an unorthodox way of achieving his goal)

I bring this example up, because a quick glance on the surface would indicate a potential intervention (trying to determine why, and what [if anything] needs fixed/expanded). However, by segmenting and drilling down farther, we were able to determine no intervention was needed. In fact, when we removed the tracking from the developer, the median time on page was about 5% or the reported average.

So, take the time to get past the average and you’ll find more actionable (or non-action needed) data.

Welcome to the blog of the Higher Ed Analytics Revolution.

This is the place where higher ed professionals like you will soon share ideas, case studies, best practices and good tips about online analytics.

Want to take a more active part in the revolution?
Just email karine@higheredanalytics.com to find out if you can write a guest post for this blog.