People who work in web, mobile and online marketing analysis come from a variety of backgrounds. Here are a few areas which can help a career in the space.

An analytical degree – I’ve met and worked with good analysts with a variety of undergraduate and graduate degrees, but generally I think a degree in something math and analytics heavy is only a plus. This could be statistics, computer science, a hard science, economics or other areas.

An understanding of web technology – You might not need to code every day in an analyst role but you should understand the basics of the web such as HTML, Javascript/AJAX, URL structures and redirection, and the mobile web. There are a lot of resources available to learn about some of these topics. Being technical in an digital analytics role might not be absolutely necessary but can only help.

An understanding of database structures – Even if you never have to interact directly with a database, learning how databases are laid out and how to query them (SQL) provides a solid foundation for thinking about sets of data and manipulating data with various tools. Additionally, some of the most interesting analysis is also done by combining web data with data stored elsewhere and sometimes you do this work in a database.

An understanding of online marketing – Online marketing is getting increasingly complex and specialized but understanding the basics of paid and organic search, affiliates, social media, display ads, email and other channels is necessary. There are various blogs and resources in each area worth following and learning from.

Excel – For better or worse a lot of time spent by any analyst in almost any domain is spent in Excel. Getting comfortable with fairly complex spreadsheets is essential.

Exposure to web analytics tools – Start with Google Analytics. It is the most widely adopted tool and you can use it for free. I highly recommend the Google Analytics training videos, additionally there are some good books about GA available. Some of the enterprise tools such as SiteCatalyst are harder to get exposure to because only large corporations use them.

Certifications – These aren’t essential but if you don’t have a huge amount of experience, taking the Google Analytics IQ exam can differentiate you from other people that have logged into GA and pulled a report once and now feel they know GA well enough to list it on their resume. Additionally the Digital Analytics Association has a certified web analyst exam for more advanced analysts.

Network – Various conferences, seminars and other places are a good way to meet other people in the space. Check out the Web Analytics Wednesday events put on occasionally by Web Analytics Demystified.

Analysis Exchange – Web Analytics Demystified also has an effort to provide mentorship to aspiring analysts and services to non profits. Probably worth checking out.

Once this is setup, marketers can then build segments and pull lists of users who visited a certain page or completed a certain action. This list is then joined to client data and targeted emails or direct mail can be sent to the users based on their previous web browsing. With some additional work, impressive automated marketing campaigns can be built and deployed.

Analytics provider Mixpanel recently launched a targeting system which tracks individual users. I believe their product goes a step beyond just capturing a unique id or an email hash and actually captures the email address itself.

Apart from email, Google Remarketing recently launched their big product integration with Google Analytics which allows you to build remarketing lists by using the advanced segmentation feature in GA.

The same skills and tools used to analyze web traffic are increasingly not just being used to measure and optimize sites and marketing, but are now being used to target audiences away from the sites they are visiting.

This is part of why the Do Not Track proposal feels so overreaching and ambiguous to me right now. Personally, I wish the proposal was named Do Not Target, and morphs into a proposal that would not ask for the prevention of measurement completely, just the prevention of the utilization of captured data for targeted ads. I’m interested to see what comes of the proposal and IE10′s decision to keep the setting on by default.

I think if you visit a website you are allowing yourself to be counted, the same way retail stores are allowed to video tape their own premises. But, if a retailer started using your photo and facial recognition software to display ads to you elsewhere… yeah people might find it creepy.

Visit Omniture’s blog post about their DigitalPulse Debugger

Bookmark this page.

Copy the javascript bookmarklet.

Edit the bookmark you made.

Select all the blog post URL and paste the javascript bookmarklet in its place.

Save the bookmarklet. Then on any page with SiteCatalyst, click this bookmark to launch the digitalpulse debugger.

Mobile is changing the web. A good analyst will track the success coming from mobile devices, mobile optimized platforms and mobile marketing channels. One trend they may not be keeping tabs on though, is the percentage of total web traffic coming from mobile networks, regardless of device type.

Cisco recently released their report on mobile data traffic

The Mobile Network in 2010 and 2011
Global mobile data traffic grew 2.6-fold in 2010, nearly tripling for the third year in a row. The 2010 mobile data traffic growth rate was higher than anticipated. Last year’s forecast projected that the growth rate would be 149 percent. This year’s estimate is that global mobile data traffic grew 159 percent in 2010.

An increasing share of web views are coming through mobile 3G and now 4G networks. More devices are being sold mobile network enabled.

This is bad news for the usefulness of IP based geo reports. In the past, you may have seen geo web analytics data that had an AOL segment. This was because AOL routed all their dialup traffic through a central gateway. Mobile networks also exhibit similar behavior. When someone is using a mobile network, the IP based geolocation ends up being the location of the carrier’s gateway.

Tonight I opened up my android phone here in San Francisco and used one geolocation service that identified my location as Huntington Beach over 400 miles away.

HTML5 offers the GeoLocation API, which examples a number of ways to determine location (GPS being the best), but because of privacy concerns, each site that wants to tap into a browser’s location has to be allowed permission.

The recent debate around Bing’s use of referring URLs in their search engine algorithm, specifically Google’s referring URL, has gotten me thinking about the availability of keyword data. There is one way for Google to prevent Bing and others from using this data in their relevancy, simply mask the referring URL. Bing could potentially still capture the Google search and the destination page, but reverse engineering the input into a site’s search box would potentially be more litigious than using every referring URL.

Search engine keywords are visible on destination pages through referring URLs. This is how web analytics, marketing tracking and other optimization tools capture the external search engine queries. When Google decided to roll out SSL for search, they decided that letting web sites capture this data was less valuable than letting users use encrypted search. Their SSL search strip referring URL information by default. Google turned on SSL for all of Gmail recently, would they turn it on for all of search?

Sites like Facebook already mask referring URLs and all Facebook traffic ends up flowing through a redirect script (facebook.com/l.php). This protects privacy, so if your profile page links to an external site and others click the link, the external site will not know that you are the individual or profile driving the traffic.

Search engine keyword data is primarily used for SEO purposes. Search engines already have a somewhat adversarial relationship with SEO. One could make the argument that preventing search keyword tracking would improve Google’s search quality by having their index be less reactionary.

Google could also strip the referring keywords and still provide keyword data to website owners. They could provide the data through Google Webmaster Tools, and potentially through an integration with Google Analytics. Hopefully if Google goes this route, they open up the integration to all web analytics vendors and don’t use their competitive advantage with Google Analytics.

Even if Google declines from stripping keyword data from URLs, web browser development, a major web browser such as Firefox or Google Chrome, could drive the disappearance of external keyword data. If a web browser decided to default search to Google SSL, a significant portion of keyword data would quickly disappear.

The big themes this year were social and mobile, which are also themes at my day job. I personally attended sessions on landing page optimization, advanced sitecatalyst plugins, marketing attribution, offline data integration, sitecatalyst data insertion/APIs and thought I got value out of all of them. John Battelle, Seth Godin’s keynotes were enjoyable, as was Facebook’s presentation. The music act was also great this year, The Killers.

This is the first conference since Adobe’s acquisition and Adobe used the forum to try to sell the acquisition both to Omniture customers and (likely) the investment community at large. Overall I’m still not completely sold on the acquisition, but if Adobe uses their capital to invest MORE in the Omniture Business Unit than Omniture could previously afford by itself, and if Adobe doesn’t try to squeeze dollars, the acquisition might have long term potential.

Apple and Google are clearly the main threats to Adobe and Omniture now. I’m not much of an Apple fanboy and I laughed when the iPad was called the iPod Touch Gigantor. I am a bit of a Google fanboy though, so I hope Omniture realizes they need to compete with SiteCatalyst instead of pitching Discover or Insight for some of the features that Google Analytics provides for free. Maybe I’m just upset because I’m not a Discover customer at my day job. I do think SiteCatalyst is the most advanced tool for capturing data and hopefully their new idea exchange will further motivate them to add features like advanced segmentation. Whoever convinced them to create the idea exchange should get a raise.

While I do think investment in SEO often has the highest ROI of any online marketing spend and many businesses of all sizes underspend and under-utilize search optimization, the graphs used in Rand’s post (and shown below) don’t tell the whole story.

Someone once told me to always beware of pie-charts, and I have to agree. First off, lets call the search channels what they actually are natural or paid search. Not every click on an unpaid listing in a SERP is a result of SEO efforts. (Also not every paid search click is necessarily pay-per-click. CPA is now making its way making its way into paid search results.)

A study in 2008 implied that around 10% of searches are navigational, 80% are informational and another 10% are transactional. You can find definitions of these keyword types from the leaked Google Quality Rater guidelines [PDF]. Lets look at these ‘keyword spaces’ across both the paid and natural search channels.

Navigational – 10% of search clicks
If a brand isn’t ranking for its own navigational searches, they are already beyond help. The vast majority of navigational searches for a brand/website will result to that brand/website without any optimizations. Navigational SEO will likely have low incremental yields unless the website is already royally screwed up to begin with. Brands should probably run paid search ads for these terms, but only because they are cheap clicks and basically forced to defensively buy these ads by the search engines. Optimization around these paid search ads will also not create success and mostly just cannibalize natural search.

Informational – 80% of search clicks
SEO is great for information searches. It’s really fucking great. Look at About.com, Mahalo.com, Demand Media and plenty of other content portals that have a business model driven solely from aggressive SEO on informational queries. Of course, Wikipedia is the dominant player, likely getting around 2% of ALL Google downstream traffic (2007 data). Informational searches are less competitive and have more words per search which create more variation in word use and order. There are a ton of these searches! There’s still plenty of space to play in informational search.

So what’s the problem with informational search? It’s great and easy to measure if you’re selling ad space, but if you are all about customer transactions, it’s more difficult to measure success. The problem with gauging the success is not in the search tools but with the internal client analytics and CRM. Can you attribute customer acquisition and lifetime value to a informational search that occurred days, weeks or months ago? Can informational searches claim they retained a customer that would have otherwise gone inactive?

Transactional – 10% of search clicks
The search engines themselves have provided simple and effective tools to judge success of transactional searches with paid search tracking pixels and Google Analytics. Because judging success is more straightforward with these keywords, paid search spend skyrockets to the limits of what is deemed effective ROI, and the majority of paid search spend is placed on these keywords. In less competitive areas, this is where SEO is a huge win. In competitve areas, this is the space where SEO is hard and brand strength(domain authority, ‘trustrank’) matters more than your clever linkbuilding campaign and perfect on-page optimizations.

Is SEO a missed opportunity? Yes, but so is all of the informational search space. Will this change? Honestly, I’m not too optimistic, but I don’t think it’s SEO tools, training or awareness that is missing. The fundamentals of SEO are increasingly just proper website usability/architecture, viral marketing, brand building and marketing analytics- not some secret skillset that only SEO experts have.

What is needed is stronger client analytics with a more rigorous definition of success beyond the immediate event or purchase. This will help justify any ‘informational search’ spend and also the trendy social media budget.

The usual method most people use to track their WordPress site search is the WordPress Search Meter Plugin. This is an easy way to get some feedback on how your users are utilizing your site search, but don’t you know that Google Analytics can track site search for you?

As great as Search Meter is, Google Analytics is the ideal place to track your onsite search. Not only can you capture the search phrases used, but I will show you how to capture the number of results in the search result set. This will let you identify all the null search terms used on your site and also see the number of results for popular phrases, and which phrases may be returning too many results.

The first step to tracking site search in Google Analytics is to edit your website profile information and and select Do Track Site Search, and put in s as the Query Parameter.

The Query Parameter input is the only thing necessary for Google Analytics to start tracking your WordPress search results, but notice we also added a Category Parameter called numResults and set that parameter to be stripped out from the URL.

Next we are going to add a javascript variable near the top of the page in the global header template.

<script type="text/javascript">
var pageType =  "default";
</script>

Then we are going to edit the template code where the search results are displayed. Some themes have a search results template, some display search results from a main template, but this code block should work inserted into either.

<?php if (is_search()) {
$numResults = $wp_query->found_posts; ?>
<script type="text/javascript">
pageType = "search results";
var numResults = <?php echo $numResults ?>;
var searchString = '<?php the_search_query() ?>';
</script> <?php } ?>

The outputted javascript should be 3 variables, pageType, numResults and searchString (We don’t use searchString in this example, but I wanted to show you all how to make it visible in JS).

Finally, you need to edit your Google Analytics code usually placed in your footer template. Make sure you don’t paste this in verbatim, as you need to maintain your unique GA id.

We are going to add a switch statement, and send a different request if the page is a search results page.

<script type="text/javascript">
try {
var pageTracker = _gat._getTracker("UA-99999-1");
switch (pageType) {
case 'search results':
pageTracker._trackPageview(document.location.href.toLowerCase() + "&numResults=" + numResults);
break;
default:
pageTracker._trackPageview();
}
} catch(err) {}
</script>

This bit of javascript appends the numResults variable to the pageview in Google Analytics, which Google Analytics then interprets as a Search Category. It also forces the URL and search string to lowercase, which normalizes search phrases with different cases.

Once Google Analytics starts reporting on your search results, you can break down search terms by number of results or number of results by search terms.

You can then click on the different number of results to see the search terms used. In a future post, I’ll show you how you can use this same GA switch statement for 404 page tracking.

I recently installed the plugin on one of my blogs with some of my own tweaks.

Replacing the images is easy enough, I decided to go with the Aquaticus icons. You can find a few lists of social bookmarking icons, here and here. Make sure to edit the CSS so the images show up the right size. I recommend including the sociable CSS in your primary CSS file.

Once I had stylish icons on the blog, I wanted to be able to tell if visitors were using the links. Of course you can track the referring site traffic from social sites easily enough with Google Analytics, but I thought I could do better than that.

I wanted to track clicking on the links as an event, so I could evaluate which services my visitors were most interested in sharing content on.  To do this, I added some javascript and edited the sociable.php file a small amount. I used Erik Vasilick’s best practices to tracking outbound links and the Google Analytics Event Tracking Guide.

The javascript function I added looked like:

<script type="text/javascript">
function trackSociable(url,description) {
try {
pageTracker._trackEvent('social', description, document.location.pathname);
setTimeout('document.location = "' + url + '"', 100);
} catch(err){}
}
</script>

I also edited sociable.php, in the php where the href tag was created for the social bookmarking links.

$link .= " href=\"javascript:window.location='".urlencode($url)."';\" title=\"$description\" onClick=\"trackSociable('".$url."','$description');return false;\">";

The event is recorded in Google Analytics with these attributes
Action = social
Category = site (Facebook, Myspace, StumbleUpon)
label = article’s relative path

Here’s an example of one of some of the data in Google Analytics:

You can also correlate social site clicks with article paths or article paths with social site clicks.   Alternatively you could record the article’s page title by substituting document.title in the javascript function instead of using document.pathname.

Overall I had a great time- especially networking with people and meetingwith venders and consultants.I felt like I left with a better understanding of the metrics marketing landscape. I think the most direct benefits I had were the discussions with other analysts, especially from other retail sites. The keynotes were entertaining but nothing too mindblowing. The breakout sessions were hit and miss for me.

I think that my three medium to long term goals have been clarified.

• Data integration with external tools, specifically email, paid search, endeca

• Understanding and building channel attribution models, based upon a combination of first touch/last touch and time spans

• Data-driven decisions and eventually a multivariant testing tool

Realistically Test and Target is about the only other Omniture product that is sparking my interest.