Quick links:

• GOTCHA #1: Visits per Page (or what happens when GA trusts us to make our own reports)

• GOTCHA #2:  Direct Traffic (or how someone did not actually type in a URL 280 characters long)

• GOTCHA #3:  GA Attribution (or how someone came to your site today from that email you sent out 4 months ago)

• GOTCHA #4: Bounce rate vs. exit rate

• GOTCHA #5:  Mobile Traffic (or why you may not need an instant site redesign)

GOTCHA #1: Visits per Page 

Visits are a standard metric throughout Google Analytics.  But one place you don’t see them is in the Content > Pages report.    Here, instead, you’ll see Pageviews and Unique Pageviews.

Many enterprising users therefore try to get around this by creating their own custom report containing Page as the dimension and Visits as the metric.

While Google Analytics will allow you to create this report, the combination doesn’t actually work.   In Google Analytics’ data model, the visit count increments only on the first page of a visit.  Therefore when you force Visits and Page together into a report, it is actually showing you visits per landing page, and the counts will be understated. Don’t do this.

If you would like to see how many visits contained a page view of a particular page, use the Unique Pageviews metric or create a segment on the Page you’re interested in and view the Audience > Overview report.

GOTCHA #2:  Direct Traffic (or how someone did not actually type in a URL 280 characters long)

Here’s a question I got that represents much of the confusion around direct traffic.

In my direct traffic report I’m seeing many deep page links receiving visits from direct traffic.  I heard that people bookmarking the page and using that URL to visit the website will measured as direct traffic, that’s fine. But here I’m sure typing and bookmarking all the internal pages is not possible.  So can you please explain me how these URLs come into the direct traffic report.

The reason he’s seeing this is because traffic credited as “direct” or “none” is often explained as reflecting typed or bookmarked traffic, but there’s really more to it than that.   In reality, direct traffic is anything where there’s no referral info coming in from a browser. Some examples include:

• typing in the URL/ bookmarking the URL and clicking on it  (this is the standard explanation)
• untagged emails that get clicked on from email clients like Outlook
• clicks from instant messaging apps like Skype or MSN (if your users are sending links to their friends in chat)
• Twitter apps
• mobile apps
• tracking issues that strip out or overwrite the referral information (for example with certain kinds of redirects)

GOTCHA #3:  GA Attribution (or how someone came to your site today from that email you sent out 4 months ago)

Users visit your site from a variety of sources over time. Maybe they discover your site via search, sign up for your email newsletter, then return to your site after clicking on the emails.  Distributing credit to your different traffic sources is a question of attribution.  Most frequently, companies use last-click attribution, where the most recent traffic source gets credit for the visit.  That’s also the model that makes the most intuitive sense – when you ask “what traffic sources drove today’s visits” you’re usually not asking “how did they come to the site 2 months ago”.

Google Analytics, however, uses a default attribution model that is a variant of last-click.    If a user comes in directly (as explained above), GA will NOT credit that user as direct if they ever came to the site from some other traffic source in the past 6 months (assuming they didn’t delete their cookies).  Essentially GA is trying to make the “none” bucket as small as possible.

In practice this means every traffic source except direct/none is overstated if you’re looking at it from a last-touch perspective. This is also why you’ll see old emails and outdated links in your reports.

GOTCHA #4: Bounce rate vs. exit rate 

Bounce rates are the percent of single page visits, where the user arrives and doesn’t trigger any events or visit any other page.  Exit rates are the percent exits from a page, regardless of whether the user viewed any other pages.  Bounces are actually  just a specific type of exits.

This sometimes leads to the question, “If bounces are just one type of exits, how can bounce rates be larger than exit rates for my page?”

The answer to this lies in looking at the specific equations that make up each metric.

So it’s important to note that the denominator is different, and the two metrics are independent.  This report helps illustrate the difference (thanks to GA customer support for originally sharing it with me): https://www.google.com/analytics/web/template?uid=wKrFy131QdmtB8X6qBP2_w

GOTCHA #5:  Mobile Traffic (or why you may not need an instant site redesign)

Check your Mobile > Overview report to see how many of your users are checking your site from a mobile device.

When I first did this for a client site, I thought holy crap! Nearly a quarter of their traffic is coming in from mobile devices and they don’t even have a mobile site!

But drilling down into device showed the story was a little different.    Yes, it says it in the label, but it’s worth repeating:  Mobile (including tablet) does, in fact, include tablets :).  You’ll likely find that a large portion of your “mobile” traffic actually consists of iPads, Nexuses, and other devices with a screen much more similar to a computer than a phone.   The story of mobile traffic may be different than you think.

Vanity URLs

A vanity URL is typically a short and memorable URL taking the format www.site.com/promo. Then, when the user types in that URL, it should immediately redirect them to a page like www.site.com?utm_medium=redirect&utm_source=promo. That example contains tracking parameters for Google Analytics, but you could just as easily apply cm_mmc parameters for Coremetrics, or your custom tracking parameters for Omniture.

Redirect Implementation

To implement the redirect, there are many ways and it will depend on your setup. However if you’re doing this yourself using a CMS like WordPress or Drupal, I believe the easiest is as follows:

1) Open and edit your .htaccess file (you can just create a new text file named .htaccess if there isn’t one there already).

2) Create a backup of the existing .htaccess file. Don’t skip this part – it’s actually very easy to bork your site if you make a mistake in your .htaccess file.

3) Add the following text anywhere:

redirect 301 /promo http://www.site.com/?utm_source=promo&utm_medium=redirect&utm_campaign=Feb_promo
[GA example]

OR

redirect 301 /promo http://www.site.com/?cm_mmc=promo-_-redirect-_-Feb_promo-_-XYZ
[Coremetrics example]

Substitute in your page names and desired tracking parameters where the text is blue.

In this example, /promo refers to the page you want to redirect, and / refers to the page you’re redirecting to. If you wanted to redirect to a page other than a homepage, for example to /newpage you’d implement the redirect as follows: redirect 301 /promo http://www.site.com/newpage?utm_source=promo&utm_medium=redirect&utm_campaign=Feb_promo

Tracking for social media has both strategic and technical elements, so this post will go deeply into both, along with a strong dose of practical considerations.

Quick links:

• Part 1: How to Identify Your Social Media Sites of Interest

• Part 2: How to Find and Isolate Your Social Media Data in the Reports

• Part 3: Pitfalls and Problems (or Why You Should Tag Your Social Links for Google Analytics)

• Part 4: HOW to Tag Your Social Links for Google Analytics

• Part 5: Technical P.S. For Google Analytics Nerds

PART 1: IDENTIFY YOUR SOCIAL MEDIA TRAFFIC

Find traffic from your social media sources by navigating to Traffic Sources > Sources > Referrals.  This will bring up a list like this:

Look through this list to identify your site’s social media traffic.  In addition to this, think through the sites your company has a social presence on, or sites where it makes sense for you to foster some kind of community.  There are two important points to keep in mind:

1) You can see from this example that social media sites are represented by a few different URLs. m.facebook.com is Facebook’s mobile site, and t.co is actually Twitter’s shortened URL service.

2)  YOU decide yourself what constitutes a “social media” source.  For example, the above list contains online fashion community polyvore.com. You probably don’t need to track this for your electronics site, but if you have a fashion-related site, then it’s probably quite relevant.  It’s up to you to decide what you want to identify as social media traffic, though a “standard” list would probably include Twitter, Facebook, Google+, Pinterest, LinkedIn, YouTube, Reddit, Tumblr, and a few more that have come out by the time you’re reading this :).

PART 2: FIND YOUR DATA

OK, you’ve made your list of social media sites you want to track. Now what?

The first thing you can do is click on the Advanced search box above the Traffic Sources > Sources > Referrals report we were already looking at.

This will bring up a dialog box as follows, where you can add in the right criteria to isolate just the social media sites you’re interested in (be careful with t.co since that will match a lot of sites if you’re not careful).

You’ll probably notice that this method will get tiresome after more than a few sites have been entered, so you can instead take a shortcut by choosing Include Source Matching RegExp and then copying and pasting the following (no spaces): (facebook.com|linkedin.com|lnkd.in|pinterest.com|plus.url.google|tumblr.com|twitter.com|^t.co$|youtube.com)

In RegExp syntax, a pipe (“|”) means “or” so if you have additional social media sites you want to track, you can append them by adding a pipe and then the site URL.

Now your report will be filtered for just your social media sites.

Out of the box, you’ll be able to see visits, pages/visit, and other key site usage metrics. If you’ve enabled ecommerce or goal tracking, you can switch to the right metric set for those for those by clicking to the appropriate tab at the top of the page.  This is actually just one of a few ways to view the data, but it should give you a good way to jump into comparing and understanding the traffic quality of your various social media sites.

PART 3: PITFALLS AND PROBLEMS (or WHY YOU SHOULD TAG YOUR LINKS FOR GOOGLE ANALYTICS)

If you are only interested in analyzing social traffic that arrives at your site because your fans (or perhaps enemies) are posting links back to your site, you can check your data as described above and stop reading now.  However, for everyone else actively managing social media accounts on Twitter, Facebook, Google+, and others, I’m sorry, but it’s not quite as easy.  Note — if you are already tagging your links, you can skip ahead right  now to Part 4: HOW to Tag Your Links for Google Analytics .  If, instead, you’re posting on social media sites but not putting special tags on your links, or if you didn’t know what a “tag” was before now, keep reading as this will apply to you :)

As you can see from the filtered report we created above, you can get a wealth of information on social media traffic and performance out of Google Analytics, all without adding a single tag.   There’s only one catch – social media is by definition social. Your social media team may be out there posting links (“come check out our Winter promotion”), but so are other people.  It takes a lot of time, money, and effort to sustain an active social media presence, so at some point you will want to understand how much social media traffic is coming from the links YOU post on social media sites versus links that get shared by your site fans and visitors.   Let’s call the first type “active social” and the second type “organic social”.

A tag is Google Analytics’ way of allowing you to control the way traffic to your site shows up in the reports – more on that shortly.  In short, if you don’t tag your links, you can’t tell active and organic social traffic apart. It all gets rolled up together.

This also means you frequently can’t identify what specific posts and status updates drove your social traffic.  You’ll see you have a big spike in traffic from Facebook, but won’t be able to drill down further than that because Facebook doesn’t supply a deeper level of granularity.

These issues are resolved by tagging your social media links.  Which brings us to….

PART 4: RECOMMENDATIONS FOR TAGGING SOCIAL MEDIA TRAFFIC

If you’re already posting on social media sites and adding tags for Google Analytics, congratulations :) As stated before, tags let you explicitly identify how traffic will show up in your GA traffic source reports.

In brief, there are 5 slots you can use to add information about every link you post out in the world.  When users click on these links and arrive at your site, any original values will be overwritten by those 5 values, which can then be looked up in the reports.  The 5 slots are utm_medium, utm_source, utm_content, utm_campaign, and utm_term, and directly correspond to the Medium, Source, Ad Content, Campaign, and Keyword fields in GA.

Every time you or your social media masters are posting links that link back to the site, they can append detailed data into these slots.    This also means you can overwrite default values.  For example, by default, social traffic comes in with a medium of “Referral” in the Traffic Sources report.  However, if you are tagging your links, you get to choose what medium name to use.

In the above case, I see many companies choose “Social”.    This means you’re out there carefully appending ?utm_medium=social&utm_source=twitter.com&utm_campaign=todays-awesome-post to the daily link, so that anyone who clicks on it gets stamped with Medium = Social.  Meanwhile your site visitors don’t know what a tag is (except MY site visitors :D) so they’re out there in the world, posting naked links back to your site that get the default stamp of Medium = Referral.

There is nothing inherently wrong with this, but it causes a split in the data that you need to be cautious about.  If you follow this strategy, make sure anyone looking at the data is aware that social data now appears under two separate mediums, and must be summed for an accurate count of overall traffic.

In practice, I find that it is easier to avoid data misinterpretations if you tag your social links as “referral” so it all rolls up together.  You can then differentiate active from organic social by adding a tag into another slot like “utm_content”.

Also be aware of maintenance.  You need to tag every post you make, and ideally with tags that change for every post or update so you can identify the traffic. This can be a significant overhead in terms of time and effort.  Reduce the time and effort associated with producing tags by creating a centralized tagging document (your company intranet or a Google doc should work well) that everyone is on board with.  You may also benefit from some automated system for creating the tags, either built internally or via some software designed for this purpose.  Whatever system you use, make sure there is in fact a SYSTEM.

PART 5: A NOT TOTALLY FINAL P.S. FOR GOOGLE ANALYTICS NERDS

There is one remaining subtle tracking issue related to how Google Analytics works.  Ironically, when you tag your links, you can actually get LESS information than what you had before!  This is because, by default, Google Analytics gets the full referrer URL (out of the document.referrer property), and populates the Source and Referral Path fields with the info.   However, when you add campaign tracking tags, you only get the information you placed in the tags.

As an example, here are the full referral paths for untagged traffic coming from Twitter.  These are the original referring URLs, meaning we can click all the way through to see the exact status update that drove the traffic.

However, when you add campaign tracking tags, you overwrite this original referrer information and no longer have access to this extremely granular information.

Despite this, I still believe the drawback of potential data loss is outweighed by the benefit of tagged links.  This is especially true since Facebook doesn’t provide full links anyway.

How do you tag and track social media for Google Analytics? Any favorite custom reports to share?

• Unilytics gives their perspective on how three major web analytics tools stack up (http://unilytics.com/blog/uncategorized/web-olympics-champion).  Contains a great product comparison grid and is a nice counterpoint to my own article comparing Coremetrics, Omniture, and Google Analytics.

• Search Engine Land investigates another lost source of search term data for analytics: http://searchengineland.com/ios-6-removes-all-google-search-referer-data-134560

• Dorcas Alexander at Lunametrics wrote a great article on how to use the goal flow report in Google Analytics (http://www.lunametrics.com/blog/2012/08/06/goal-flow-how-visitors-really-move-through-funnel).  Lunametrics consistently puts out some of the most practical, detailed posts on web analytics so just subscribe there, now :).

• Andrew Anderson at Omniture has a very thought-provoking article on how we all have the bad habit of looking at data and creating stories to “fill in the gap,” even when there’s no way to know that the story is accurate and the story should have no impact on the decision anyway.  Yikes, I think I do this every day…(http://blogs.adobe.com/digitalmarketing/personalization/conversion-optimization/why-we-do-what-we-do-separating-fact-from-fiction-the-narrative-fallacy)

• Mariel Bacci at FutureNow wrote a very useful article comparing the new Google Content Experiments vs the old Google Website Optimizer for A/B and multivariate testing (  http://www.grokdotcom.com/2012/08/22/google-content-experiments-the-good-the-bad-and-the-alternatives/)

• Jim Gianoglio at Lunametrics explains how not all “direct” visits are actually direct.  The article talks about GA but actually applies to all web analytics tools. (http://www.lunametrics.com/blog/2012/09/17/direct-visits-in-google-analytics).

• Cool, clear overview from Brent Dykes on ways to get data in and out of Omniture: http://blogs.adobe.com/digitalmarketing/analytics/how-to-get-data-in-and-out-of-sitecatalyst-part-i and http://blogs.adobe.com/digitalmarketing/analytics/how-to-get-data-in-and-out-of-sitecatalyst-part-ii.

• Sean Carlos at Antezeta tells us to relax about Google having too much data on us if we use Google Analytics… because they are already tracking our every move through a myriad other ways.  Not sure whether to feel relaxed or alarmed!?  http://antezeta.com/news/how-google-tracks-us

Hope everyone is having a good weekend!

Read any good analytics-related articles recently? Post them in the comments so I can take a look.

Profile Filters: What’s Changing

Google Analytics profile filters previously allowed you to include or exclude traffic to certain sections of the site, but didn’t apply to ecommerce stats.   Now these filters apply to ecommerce stats as well.

Example:

For example, imagine you had an ecommerce site selling various household goods and just wanted to see traffic to the mug section. To do this you could set up the following filter:

Before, this filter would not apply to the ecommerce data. Ecommerce data was not associated with any particular page, so filtering data for a subdirectory didn’t affect it. Instead, filtered profiles continued to show full ecommerce revenue, transactions, etc as in the non-filtered profiles, and filtering ecommerce data had to be done separately, via one of the Ecommerce filters designed for this purpose:

Under the new logic, only transactions taking place on the /mugs sub-directory will be counted.

What to Know

1) If you’re filtering to show just traffic that occurs on a subdirectory (say, /mugs) of your site, most likely you will no longer see any transaction data at all in your profile.  This is because checkout usually occurs on sub-directory like /checkout, not on /mugs.

2) To see transaction data, you will need to add in a condition to the content filter. For example, you could alter the original filter to look like this:

Note that this changed from the “predefined filter” to the “custom filter” with an include condition. This is because only the custom filters allow regex, which is what enables you to insert an “or” clause via a pipe (“|”). This way I can specify that I want to include traffic that visits either the mugs section OR the checkout section.

With this second method, you will be back to seeing all ecommerce transaction data, so if this isn’t what you want you will need to insert additional ecommerce filters.   Note that if you just want to see ecommerce data for users who viewed a certain section of the site, one easy solution is to  create an advanced segment for that condition and apply it to the main unfiltered profile. Then view the data in the ecommerce reports.

3) If you’re filtering out internal traffic with an IP or hostname filter, this filter will now apply to transaction data as well. This is a very welcome change, since previously these filters filtered out the internal traffic but not the internal revenue, resulting in some very skewed data from any QA or test orders.

More Info

You can view the official notice of this change at http://analytics.blogspot.com/2012/09/ecommerce-filtering-and-custom-variables.html