Visits in Google Analytics and Adobe Analytics (Omniture SiteCatalyst)

Author:
 

Have you ever created an advanced segment in SiteCatalyst 15, and noticed that additional traffic sources besides what you filtered for are still appearing in reports? For example, let’s say I create a segment in SiteCatalyst 15 to filter for visits that contain “facebook” in the referrer:

Then you go to look at your Referrers report under the Traffic Sources section, expecting to see only traffic from Facebook, and lo and behold – you can see traffic from all sorts of other referring domains! In the example below, the Referring Domains report shows direct traffic, Yahoo! traffic, and traffic from shortstack.com as well as facebook.com, even though we have a Facebook traffic advanced segment (shown above) turned on. (Click the image to enlarge.)

So why does this happen? If you’re asking this question, I have to assume that you’re used to the way traffic from Google Analytics works. I’ll explain the difference between Google Analytics and SiteCatalyst on this below, but first we need to have a conversation about how web analytics tools “see” your traffic on the web.

Every time you load a page on the web, it’s kind of like scanning your food at the grocery store. Bear with me. Imagine you are surfing the web, and you land on Zappos.com’s homepage. You now have a set of cookies associated with that page view, that has data like where you came from, a time stamp  which page you viewed, and so on. This is just like how when you scan a can of pineapple at the store, the SKU tells the register all about the food you are buying. How many ounces the can is, whether it’s sliced or diced pineapple, which brand… you get the idea.

So you land on Zappos.com’s homepage. Boop! Then you click on the link for women’s shoes and hit that page. Boop! Then you go to the “heels” page. Boop! Then the amazing Giuseppe Zanotti pair that you haven’t seen yet, must be part of their spring line. Boop! (I’m a size 8, in case anyone was wondering. Hint hint.)

Here’s what that looks like from the point of view of your “register” in this metaphor, your web analytics tool:

So imagine each of the above web pages is a “product” that you’re checking out at the grocery store. Seems fairly simple, right? Well in reality we’re leaving out a few steps here.

It happens all the time when you’re at the grocery store – you finish your shopping, and then you realize five minutes later that you forgot to buy something crucial (usually just as you’ve finished loading all the groceries in the car). Then you have to run back in and get it, and have a silly little embarrassing chat with the cashier.

Web users do this too – on any given night I might surf Zappos.com looking at cute shoes, write a blog entry for work, check Facebook, and then go back to surfing Zappos.com some more after clicking an ad on Facebook. So in reality my data looks like this (you can click for a larger image):

I removed the time stamp piece from this example, just for now, to make the above easier to read. (I have another blog entry about time stamps and their effect on Google Analytics traffic).

You’ll also notice I colored two of the squares above red. If I work at Zappos and am looking at Zappos.com’s analytics (which I’m not in reality, they are not a client of mine – just a favorite example), then I won’t have access to the data for facebook.com or location3.com, regardless of whether I’m using Google Analytics or SiteCatalyst. However they are important to note. Notice how the traffic sources for the zappos.com data are now Google Organic and Facebook.

So where does the difference come in? I’m getting there! Both Google Analytics and Adobe SiteCatalyst have a 30 minute timeout rule for measuring visits. So for example, in the scenario below, both tools would see the blue squares as separate visits.

However, what if only five minutes pass between the two blue page views? Like below:

Google Analytics sees the above scenario as two visits, because you have left the site. Google’s official documentation on this states that a visit in Google Analytics ends after 30 minutes of inactivity, after midnight (which is an odd one to me), or after “you arrive at the site via one campaign, leave, and then come back to the site via another campaign.” However this isn’t entirely factual – even if you leave and return via the same campaign, Google Analytics will see the returning page view as a new visit – I tested to make sure!

SiteCatalyst is very different in this scenario. Even though the user has left the site, according to SiteCatalyst, they are still within the 30 minute window so the second page view is seen as part of the same visit. Because of this, visits in SiteCatalyst may have two or more referring sources.

SiteCatalyst does offer data on the original referring domain, which would be only Google in the example above. That’s another thing – Google Analytics would not consider Google organic to be a “referrer,” while SiteCatalyst sees it as both a referrer and a search engine.

For some web analysts, SiteCatalyst’s method may be considered to be less intuitive. Meanwhile, others find it to be more logical, with the thought being that you are having a continued engagement with that web brand and experience, regardless of a short “break” between page views. My take on it is that both tools can be powerful, but you need to make sure you understand them before you use them.

Any questions? Drop ’em below and we’ll get to them ASAP!

 

25 thoughts on “Visits in Google Analytics and Adobe Analytics (Omniture SiteCatalyst)

  1. Excellent explanation!
    My concern is there are actually two parts to this equation. Yes, the most important distinction *if you’re more familiar with Google Analytics than Adobe Analytics* is the way a visit is “cut off” in GA versus the definition of Visits for Adobe, because this is how they are fundamentally different in mechanism and terminology… Adobe Analytics, by the way, last time I checked, defines a visit as a series of image requests until 30 minutes of inactivity, or 2500 successive image requests (think: bots), or 12 successive hours of uninterrupted sequential page views without a 30 minute break.

    However, even experienced SiteCatalyst users sometimes forget it is also important to understand the nature of a container under the context of the Adobe Segment Canvas: A container signifies that the segmentation engine is going to sift through all of the data from your site from the specified time period, looking for individual instances which match a given criteria, and WHEN IT FIND ONE THAT MATCHES your criteria, it will include ALL of the image requests that belong to the same scope as the specified container. And the Exclude canvas works similarly, but as a kind of “second pass” where the unwanted stuff is carved out of whatever you have already included.
    This is possibly a better way to answer the question, “I expected ONLY facebook. How come I’m not seeing just facebook as the referrer when I know that’s what I segmented for?”
    It means that lines 2, 3, and 4 of your example screenshot above are actually what you could call ‘slop over’ – these referrers ALSO occurred within the scope of the same visit during which, at some point, there was a hit that matched your segment criteria.
    I know, those values obviously don’t match your segment criteria, but they are guilty by association. And if you think about it that way, maybe there’s something you can glean from the fact that sometimes your visitors come to the site from both facebook AND from shortstack.com (either before or after – the segment doesn’t care about the order of qualifying events). I don’t know what this may signify to you, but it is not necessarily a bad thing that you now know what ELSE may have contributed as a referrer to the eventual conversions that are clearly traceable to your facebook segment. Although you can feel free to ignore the slop, or go to the exclude canvas and get rid of it. In a way, more careful segmentation would be a lot like forcing the Adobe visit to (practically) act like the Google Analytics-style visit, although there are several ways in which this is not directly equivalent.

    I hope this helps.
    Jorgen Sorensen

    1. Hey Jorgen,

      Glad you enjoyed the article! You could remove the “slop over” and force SiteCatalyst to act like Google Analytics by using the exclude canvas, but I always recommend internally that users don’t do that because it goes against the way SiteCatalyst was meant to report. In a way I feel like if you’re going to use SiteCatalyst, use it well, and use it the way its developers intended. I would also be concerned about using the exclude canvas in that way because it may not make sense to the client or other 3rd parties that are looking at the same data and are used to looking at it the “default” way.

      I also enjoy the “accidental attribution” aspect of the way that SiteCatalyst works. It’s sort of like you get to see what referring websites helped “cross sell” with the one you’re looking at. 🙂

  2. Great article. Thanks. We are trying to answer the same question: why would organic search visits, especially visits as % of all traffic is different in GA compared to site catalyst. Based on the example you provided above if they navigate and arrive from different campaigns (or same campaign.) within 30 minutes they will have 2 visits associated with them. That would mean visits in GA should be more than what we have in Omniture or at least be equal. It’s actually the other way around. What could be causing this? Thanks in advance.

    1. Hi Sudheer,

      I will reach out to you directly to see if I can help answer your question. If we solve another mystery I’ll come back to this post and add that as well. 🙂

  3. Great article, Tara.
    I often explain the difference between the two tools as:

    Site Catalyst is more “literal” than Google Analytics. It gives you what it thinks is the right answer even if you don’t like it.

    Google tends to smooth over some of the bumps to provide data that is what you expect, but may hide some of the truth.

    We use Site Catalyst because we can’t get what we need from Google Analytics, however in many cases I like how Google deals with the data. As an example, look at Site Catalyst’s medium called “internally referred”. Site Catalyst defines this as a visit that was referred by your own site, just after 30 minutes of inactivity so it’s counted as a new visit. My understanding is that Google treats this as a direct (typed/bookmarked) visit. Site Catalyst is more accurate but it is a constant question in meetings. Google is “close to the truth” but is way easier to understand and raises no eyebrows. From my perspective I would prefer the tools to report on the most previous referrer, but that would break the “visit length” rule.

    I have a question. You mentioned in this article that Site Catalyst provides an “original referring domain” value and this would match only Google in your example. We’re trying to make a similar report in Site Catalyst to show “original entry page” but we’re getting lots of additional entry pages other than what was provided by the segment. Have you noticed this issue?

    Thanks – keep up the great work. It’s difficult to find good info about the differences between Google and Adobe analytics.

    Keith

    1. Thanks Keith, glad you liked the article!

      I would agree that Google Analytics makes more assumptions about how you’d like to see the data. It results in a really nice analysis experience but when you want fine-tuned or complex details then it gets a bit tricky.

      I will reach out to you personally to see if we can resolve your question; the blog comments here are a limited forum for troubleshooting.

      -Tara

      1. Hi Tara,

        Thanks for the response. No need to reach out, I just got confirmation from Adobe that “original entry page” actually has a bug and isn’t working as intended. Hopefully it will be fixed soon. Thanks for the offer of assistance!

        Keith

  4. Hi Tara,
    Really good article, sounds like you really know your stuff. I’ve run into a situation where I created a segment and used the container visits but it was showing a different value for number of units sold. Basically, it was totaling more than it should have (some kind of duplication of some numbers). When I changed the container of the segment to pageviews, we got the correct number of units. Its working now but just not sure why or can’t explain what the difference is. Would you happen to have an idea?

    1. Hey David,

      I would need to know more about the type of segment you’re creating and the specific report you’re looking at. I’ll reach out directly to resolve there.

      -Tara

        1. Hi Ann,

          Sorry for the delay. This is a little bit of a custom situation. But typically the issue is related to this: if you make a segment for visits from Facebook, it may include sessions from other sources. i.e. if a user comes to a site from Facebook and then Google organic within 5 minutes of each other, the entire “visit” contains the Facebook hit. Whereas when you change your segments to just “pageviews” or “hits” from Facebook, it excludes traffic from the Google organic hit. Hopefully that makes sense. If you still have questions let me know and we can discuss one-on-one.

          -Tara

  5. I was wondering if I could clarify something. In the example you gave with the 5 minute browsing session, are you suggesting that GA is attributing the 2 Zappos page views as 2 visits that came from Google Organic or as 2 visits , 1 Google organic and 1 Facebook referral?

    1. Hey Liz,

      Sorry for the confusion! I meant to say that GA is counting two visits, 1 Google organic and 1 Facebook referral. SiteCatalyst in this situation would show 1 visit, with the two sources in the same visit – hopefully that makes sense!

      Best,
      -Tara

  6. Hi, Tara. Nice article. I’m curious to know which campaign you tested when you say that you returned via the same campaign within the 30 minute session and it counted as two separate visits. If it’s a paid search campaign, for example, it will generate two different gclid values, according to Google, and thus open a new visit and count as two visits.

    1. Wow, sorry for the super late answer! Our blogs got a bit shuffled in a redesign and this wasn’t assigned to me for a while as a result. You are correct that a new gclid would result in a new visit, but let’s say you came from a Bing ad or a click on a the same Facebook ad. Then it would count as the same visit.

  7. How do you break down visits into different channels like natural, paid, typed/bookmarked, etc? Can you explain what advanced segments do we require for the above breakdown?

  8. Hi,

    This is a fantastic article, that has really helped me as a Site Catalyst/Adobe Analytics’s newcomer. My question is would using the “Hit” container solve this issue? I’m curious how the hit container works for referrers and tracking codes, etc.

    I want to see a breakdown of the pages users visit after using a tracking code and how many page views each page has received. Will using hit container for a referrer in a pages report pull the total number of times that each page has been viewed through that tracking code, or do I need a Visit container for this (they seem to pull different page view figures)?

    Thanks

    Toby

    1. Hi Toby, apologies for the delayed answer. Consider this: a hit to your website from Facebook, which views /home and /about-us, and a hit to your website from Google which views /about-us and /services, both of which occur by the same user during the same visit, according to Adobe Analytics. If you segment based on hits from Facebook, when looking at content reports you will just see the first two pages. But when looking at reports that contain data on visits, you will still see data from the Google hit. This is because you’re segmenting the visit that contained the Facebook hit. Hopefully that makes sense!

      1. Thanks Tara,

        That’s really helpful and clears things up perfectly. One last question: Do you know which method would be considered best practice to use, as surely hit risks underestimating the performance of a referrer and visit risks over-estimating?

        Thanks

        Tobz

        1. Hey Toby (Tobz? 🙂 ),

          From my experience, it depends on the client/user and what they would consider to be a visit. For instance, a client based on content consumption (The Onion, Buzzfeed, Imgur, etc.) would probably prefer to segment/analyze by hit or clickthrough, because that quick interaction between sources is a little bit more essential to the nature of that client. Someone who wants a bigger picture of “what % of my traffic is social and how well is that impacting conversions” might be satisfied with a visit or even user-level scope. A combination of those, or at least a good education to the client on the differences, is really important.

          Hopefully that adds some value. There isn’t really a solid “always do it exactly this way” answer. 🙂

          Thanks,
          -Tara

  9. Thanks for the post. I’ve recently started working with a client who uses SC and realized there aren’t a lot of good resources out there – this was very helpful.

Leave a Comment

Your email address will not be published. Required fields are marked *