We've been doing a lot of Google Analytics audits, debugging and configuration. It's a crucial time for ecommerce merchants to be thinking about the state of their web analytics, and the correctness of the data. Weak though the ecomony may be, the dog days are summer are quickly giving way to fall/holiday (or should we say "Fall-iday?").
Of course, back-to-school merchandising has been in full swing for a couple months. This weekend when I popped into the supermarket for some milk, I was blown away to see full displays of Halloween candy. Consumers may not have to fall and winter months in mind yet, but hopeful merchants are all over it.
If your analytics are configured incorrectly, you'll be without a proper roadmap for assessing what your current baseline looks like, and what aspects of your Holiday season campaigns and merchandising are successful and which are not. Even worse, bad data may make you draw completely wrong conclusions -- taking credit for internally-placed test orders, say, or classifying paid search traffic and sales as organic. If you can't properly tie online revenues to their source, you're bound to terminate some successful campaigns and perpetuate some dogs.
So here's our list of the most frequently seen Google Analytics configuration errors -- and how to fix them:
1.) Failing to tag paid search campaigns. Google can let you employ "autotagging" to effortlessly track whether your Google search traffic (and sales) is paid or organic. But paid campaigns at Bing, Yahoo, and third-tier search engines will look like free, organic traffic unless you use campaign tracking links to tell GA otherwise.
2.) Failing to track email. The house email is the biggest-yielding channel for most ecommerce merchants. If you don't use campaign-tracking codes, you'll have no easy way to track the total revenue contribution of your house email, or its effect on your total conversion rate. It's also imperative to track your Abandoned Cart email and other transactional emails as a separate class of email campaign. We use utm_medium = email to tag all emails, and utm_source to discriminate which list is being sent, and whether it is from our transactional or promotional email programs.
3.) Failing to filter internal traffic and test orders. Everything from conversion rate to raw revenue numbers can be rendered meaningless if your web team or IT department places a lot of test orders without setting up an exclude filter. You can still maintain an "all traffic" profile -- so that you can SEE the results of tests you need to place, for instance! -- but set up at least one profile filtering out internal IP addresses. That way you'll know what the REAL top and bottom lines are.
4.) Failing to track on-site search. Not all web platforms are engineered to display the search term in the URL of the results page, but most are. GA provides a basic reporting of top search terms, and the overall share of users who use on-site search, and how much business they do (almost always well above your site average). There's no native reporting of no-results-found, but it's a start -- and a vital insight into one of the most important merchandising tools on your site. Plus, without baselining your site's current on-site search metrics, you'll have no way to gauge whether an advanced third-party search like SLI is worth the money.
5.) Overreporting on-site search. Some website owners like to configure internal links on their site, or landing pages from paid search or email, as search results pages, rather than having to build special category pages or static landing pages. That can be smart merchandising, but if you do it, also set up an additional profile or "advanced segment" to exclude visitors by medium. Otherwise you'll be overstating your on-site search usage, and probably understating conversion rate.
6.) Overusing profiles. Now, virtually all of the functions of separate analytics profiles can be played by Advanced Segments. Setting up advanced segments is pretty easy, they're convenient to apply or unapply to any report you're working with, and they save you the hassle of setting up the same goals and other settings across a ton of different profiles.
7.) Setting up the shopping cart funnel one page too late. The shopping cart should be the first page of your conversion funnel, NOT the check-out page. When we say a 50% cart abandonment rate is about the industry average, we're talking about the opposite of funnel conversion rate, if you start at the shopping cart. The later you start your funnel, the more falsely inflated your numbers will be. The cart is a critical page to track in the funnel, because it also tells you your engagement rate (or the percent of site visitors who add anything to the cart). And it tracks the performance of the first page of the checkout -- which, along with the final step, when consumers have to actually part with their moolah -- is one of the hardest transitions in ecommerce.
8.) Failing to test. Web analytics are not just stale numbers -- they're calls to action to solve shoppers' problems, make the online experience easier, more compelling, faster, better. Anybody who has been doing ecommerce awhile knows that their best hunches and marketing instincts should be tested and quantified to see what (if any) lift they provide. So using analytics to baseline your performance, and then the Google Website Optimizer testing tool (or the A/B testing features in Adwords) enables you to prove your instincts right or wrong. We never stop learning at this game!
9.) The "Site Overlay" report. Don't worry, it's not your fault. Google Analytics' Site Overlay report just sucks. It doesn't work. I've never seen it work. It combines the stats of any duplicate links on the page, and the data is always screwy. If you want a better view of how web shoppers are actually responding to your navigational and merchandise offerings, you need to use a good, inexpensive third-party tool loke CrazyEgg or ClickTale.
6549641c-c63f-4b87-8a4d-5d16d191b65f|1|5.0