Until early 2007, Omniture limited the opportunity to receive implementation certification only to its partners. Now it offers certification to customers–at an extra fee, as is the Omniture way. Only two participants in my class stayed for the certification exam, which takes place the morning after the day three class. Nearly all of the rest hadn’t heard of the offer ahead of time, for which the instructor apologized.

You’d think Omniture would encourage certification. After all, it’s in the company’s interest to establish Omniture as the standard for web analytics. The small amount of money the firm takes in administering the test pales to the revenue brought in from just one additional customer. And more new customers are likely to sign up if they can count on finding qualified implementers out there to overcome the perceived complexity and cost of implementing Omniture.

Sure, it takes some effort for Omniture to grade and administer the test, but even that is surmountable. The test is divided into a multiple-choice section followed by a sample site implementation exercise. The Q&A could be automated and those who didn’t pass a minimum threshold wouldn’t have their more labor-intensive implementation exercise evaluated.

When I’ve interviewed candidates for web analytics positions, having Omniture certification stands out. Without that designation, candidates can say they’ve worked with Omniture but the range of experience varies as much as website designs. Upon further probing, I’ve seen a candidate’s experience range from merely looking at an Omniture report that was automatically generated by someone else, to creating the reports themselves all the way up to JavaScript programming that fully integrates with a CRM.

If you’re thinking about adding Omniture certification to your resume–which is a very good idea–here are some study guide questions to help you prepare. These are gleaned from tips mentioned during the class, best guesses and a few ideas from someone who took the exam.

• Know the variable length limits for the major variables
• Understand when to use an evar, a prop, hierarchy and an event variable
• Remember the sequence:
  s.products=”Category;Product;Qty;TotalPrice;Incrementor;Merchandising”
  (and remember not to use “category” but to lead with a semi-colon before “product”

For the implementation exercise, be prepared to fill out your own Solutions Design Document based on a template given to you for a fictitious ecommerce merchandise site.

• Use traffic variables, hierarchy variables, and commerce variables and events, but not necessarily all of them all of the time.
• Make sure to complete:
  • s.pageName
  • s.channel
  • s.pageType
  • s.prop1 and s.prop2
  • s.hier1
  • s.campaign
  • s.eVar1-3
• Implement standard cart events (add to cart, check out, etc)
• Add a few custom events
• Know which functions require a call to Omniture; note these on your form
• Know how to use the debugger

Good luck! If you’ve taken the exam and want to share any tips here, feel free to share. This is intended as a study aid not a cheat sheet to help all of us become better web statistics analysts whether or not you’re working on a church website.

Building on what we learned during day one and day two of the Omniture advanced implementation class, day three covered new material and finished with an exercise implementing the basics on a sample site. At this point, we felt we knew how to ask the right business questions to be answered by web analytics and that we had the know-how to implement the right solutions.

It’s worth staying for the entire session until 5 PM. Toward the end of the day a senior implementation consultant came in to answer questions that came up during the three days that the instructor couldn’t answer. Some participants left early to catch flights, but I’m glad I stayed until the end even though that meant taking the red eye home.

Here are my takeaways from the final day of class for those of you already familiar with Omniture.

• If you are rolling up multiple report suites to a higher, global report suite, each suite must use the same time zones for the data to work properly. For example, if you have sites in Los Angeles, Chicago and New York–each with its own report suite–and you want to roll up the data to an overall report suite (total organizational performance), you must select the same time zone for all the reports; if that’s Eastern Time because your most important stakeholders are located there then your LA report is going to be in Eastern Time.
• The SAINT application uses the same database as the campaign manager, although the two appear to be separate.
• How do you use clickmaps overlays on older pages where the design has significantly changed? Save a local version of the web page to your hard drive as an archive. You can then turn on the overlay to see what happened during that time period. Better yet, use your content management system to archive the changes or keep back copies on your server. For instance, imagine that your home page one month has a three-column layout, but that you later change it to a two-column layout. Your clickmap overlay will no longer “see” some of the links that are no longer there in the new design. By going back in time to your archived copy–and changing the reporting period accordingly–you can once again see which area’s of that page were getting the most clicks. Make sure your JS file uses an absolute, not relative, link for this to work.
• Use S_Object ID to track rotating ads that use separate URLs through clickmaps. For example, if your home page always has an ad in the corner, but it points to a different message and URL throughout the day, this variable will let you count how many clicks you’re getting in the corner. Now when it’s time to track internal campaigns, you’ll use a separate evar to determine which of those ads were the most effective. This one-two combination lets you track position (are visitors clicking in the corner?) and individual ads (which ones work?).
• Import your old log files or previous analytics package data into a separate report suite. You can’t mix and match with your new Omniture data, but the reports will be familiar when you want to look at historical data. So, if you moved from the WebTrends logfile analyzer to Omniture at the start of the year, then your December and earlier data from WebTrends could be imported to a separate report suite. Your day-to-day reporting would take place in your main report suite, but you’d still have access to your older data.
• If you have multiple domains or add your tracking code to third-party sites that do some of your hosting (such as for a promotional campaign, fundraiser or contest), make sure to specify these domains as internal within the Admin tab under Edit/General/Internal. If not, supposedly you’ll have some page names show up as “other” in the reports–in which case you’ll have to call Omniture to find out what those pages are. However, I had a case where I hadn’t identified a third party as an internal domain, yet the data came through fine; the senior implementation consultant who came to the class couldn’t explain this. Nevertheless, I’ll set the internal domains going forward.
• The Omniture debugger does not work on custom link tracking; use Ethereal, an open source packet sniffer, instead.

This class also included a hard sell for Discover 2, an expensive (even by Omniture standards) option that gives you real time access to the full, non-normalized database of your analytics data. It’s like Omniture’s Data Warehouse product without the multi-day wait for data. For performance reasons your data is normally chunked up and only the relationships and correlations you requested/configured ahead of time are available. If you think of some relationships you want to explore down the road, those have to be set up in SiteCatalyst and are available only going forward. Discover 2, however, gives you every combination right away.

Next: Optional day 4: certification

With the foundation set during day one of the Omniture implementation class, we turned to more hands on activities and working through the code. Here are a few points that stood out for me.

• What’s the difference between and evar and an event? Evar = who or what triggered an event; events = how many times. So, an evar would indicate what made someone sign up for your newsletter, with the successful registration counted by your event variable.
• Don’t use “categories”; leave blank (or, technically, use a semi-colon); once a relationship is set, you can never turn back; use classifications instead to show categories
• No commas in product names; that’s the delimiter for the next product if you’re putting multiple items in a product variable
• Use purchase ID to dedupe events on final purchase page; otherwise, you could inflate totals if someone reloads or later returns to that page. For example, you complete a purchase at home; the next day at work you go back to that page to print out the receipt for reimbursement–the page and visitor count has increased even though the purchase is final
• Call Omniture to turn on cross-category merchandising — that’s when you have products that fall into multiple areas; for example, you might sell a children’s Bible under Gifts, Kids and Bibles; this lets you set which area of your site gets credit for the sale of an item that appears in multiple lcoations.
• Internal campaigns — another case where you must call mniture to turn on this feature so thatit gets a separate line item in the commerce/conversion report
• For A/B testing, use an evar to check conversion effectiveness; for example, evar4=”Home Page Design 1″ or “home Page Design 2”; since this isn’t multivariate testing, you’ll need to test but one item at a time
• To track Flash, use ActionSource
• Get QueryParam can support multiple campaign indicators — you can choose CID and PID and SRC, etc.
• If you want to track the success of your forms, call Omniture to turn on Form Analysis Plugin; it is not set by default