Is Google Analytics an “enterprise” class analytics solution? That’s debatable, and in fact, it has already been debated.
In my opinion, it depends. It depends on your analytics needs.
We’ve worked with plenty of “enterprise” class organizations that were new to web analytics. They had very simple needs and GA met most of them easily. We’ve also told companies that GA is not right for them because it did not fit their core needs.
Your organization may be different. You may need a tool that integrates with ODBC data sources, something that GA doesn’t do very well. If that’s the case then you might need to go with a different tool. But again, it all depends.
But the point of this post is not to debate GA’s “enterpise-y-ness”, but to address some of the common issues that we usually see during an enterprise installation.
Issue #1. Tracking All Sites Logically
Large organizations tend to have more sites, and more sites mean more data. Collecting the data in an organized fashion, that allows room for growth and appropriate access for users, takes time and planing.
During an enterprise implementation we usually create a series of accounts and profiles that segments the data based on business logic and access needs. We create a data hierarchy that provides high level aggregate tracking across the entire online experience (i.e. roll-up reporting) and detailed tracking for each individual property.
Let’s consider the websites for Major League Baseball. Each team has their own site located on a subdomain. There is also an MLB store and different micro sites dedicated to things like the All Star Game and the World Series.
Lots of content on many different sites. While the exact implementation solution will depend on their specific needs, it probably involves collecting all the data in a single profile for roll-up reporting and then creating profiles for each team and micros site for detailed reporting.
Issue #2. Unique Visitors
Tracking lots of domains usually leads to an issue with unique visitor tracking. GA uses a first party cookie to identify each visitor. This means that if a visitor visits 3 different domains they will receive 3 different cookies and appear as three different unique visitors.
Now, I know GA has a cross domain tracking feature. But what happens if an enterprise wants to know the unique visitor count across 50 web properties? Installing cross domain tracking on that scale is a huge task. In fact, it’s a pain in the ass.
Many of the clients that I’ve worked with have compromised and ignored unique visitor tracking.
You may be different. Unique visitors may the one critical metric that you can’t live without. Could you use GA? Maybe, but you should carefully weigh the implementation needs vs. your analysis needs.
Issue #3: Page Tagging
When I first started working with GA I never thought that tagging pages would be an issue, but it is. It’s not so much a technical issue as it is an organizational issue. Big companies can have so many sites with some many nooks and crannies. It can take a lot of work to identify every site, find an owner and then get the tags placed in the appropriate place.
And let’s not forget non-HTML pages. Tracking non-HTML content with Google Analytics can be a huge challenge. You can’t slap a JavaScript Tag on a PDF. When we work with large organizations we usually help then develop an automated click tracking script. This takes more time and more effort and doesn’t always work (usually due to page rendering delays).
Issue #4. URL Structure
This is probably one of the most difficult challenges we face when working with large sites that have hundreds of thousands of pages. GA will only track 50,000 unique URLs per day. While this is completely adequate for most sites “enterprise” sites can exceed this limit, especially if the site is content based (think about a some of today’s largest community sites, they have forums, blogs, and tons of user generated content).
What happens when you fill GA with 50k unique URLs in a day? You start to see ‘(other)’ in your content reports and you can no longer identify which pages visitors are viewing on your site.
To resolve this issue we usually need to create some type of bucketing strategy to ‘roll up’ pageview data into different content categories. This is normally done by matching requested URL patterns at the server level, and then generating a ‘virtual’ pageview in GA.
Sometimes we segment the data into different profiles, thus giving us more ‘buckets’ to store the data.
Again, the exact solution depends on many different factors, but this issue can be mitigated with some effort.
Issue #5. Campaign Tracking
This is a problem for everyone! I find very few clients whoa are diligent about tracking their marketing campaigns using link tagging. A general rule of thumb, the bigger the client the more challenging it is to track all online campaigns. Why?
Big organizations have different people running different campaigns. Many times they’re using one or more agencies to help run their campaigns. Getting everyone to use a cohesive link tagging strategy is a lot of work due to the sheer number of people that are involved. This is more of a training/process issue rather than a technical issue.
Wrapping Up
If you’re an enterprise organization, or consider yourself an enterprise organization, don’t discount GA without taking a hard look at your real analytics experience and your needs. GA might just work for you.
If you do decide to use GA don’t expect to slap the tags on your site and finish the configuration in a week. Like every tool out there, it takes time and planning to get things right.
Do you have experience with GA in a large, “enterprise” environment? Leave a comment and share your thoughts.
We’re using Google Analytics on one of Germany’s biggest parenting sites and I agree primarily with the URL structure problem that is definetely one we’re facing and we haven’t solved it yet to our satisfaction.
We could easily create content buckets in different profiles, (e. g. for community and content areas), but one of our analytics aims is to get a feeling how users accept the combination of community and content
I’d be very interested in reading more about your strategies on tracking the URL pattern on the server and creating a virtual page view. I don’t remember reading sth about that in your eBook ;-)
Great article Justin – every web analytics manager / director who works at an agency should immediately relate to all of this :) Communication is the key.
Joe: Thanks Joe. I completely agree. Without any communication any project is doomed for failure. It’s critical that any GA implementation projects has open communication between the client analysts and the implementation team.
Bjoern: Most of our content bucketing strategies utulize some type of custom code. This code is either clients side (JS) or server side. The client side code is some type of URL lookup, where we compare the current page URL to a ‘library’ using a reg ex and then manipulate trackPageview() based on the result.
The server side code is more or less the same. The only difference is that it is written in the client’s programming language. We create some type of script, like ga.aspx and pass it the current page url, like this:
script src=ga.aspx?url=www.foo.com/forum/?id=32
ga.aspx would then take the url variable and compare it to a library. Then it would output the appropriate pageTracker() call.
Hope that helps.
Thanks to both of you for the comments.
Justin
Although I probably haven’t dealt with enterprise organizations of your size, I recognize almost everything you are talking about. Enterprise organizations can often be synonyms to the Three T’s (Things Take Time).
I just want to mention that Campaign Tracking using link tagging isn’t only problematic in Google Analytics. My experience is that Enterprise organizations that have invested in expencive web analytics tools, are more likely to want to use the same tool to track for example banner advertising using link tagging (since they are paying for an enterprise tool some tend to think the tool is suited for everything).
If you are tracking and optimizing a banner campaign across several domains and sections, with several banner types and messages – you really don’t want to do that using link tagging. Use a tool suited for the task like DoubleClick or a similar adserving tool.
However, I would add some general link tagging to the advertised URL to avoid that the traffic is reported as direct traffic (which may happen with Flash ads).
Justin, I LOVE what you are hitting on here man! After just having an “enterprise” web analytics provider call us (one of the big 5) and try to explain to us what we are missing out on by not using their service over Google Analytics. The problem is that I could tell this was a stock response. It wasn’t in context to how WE at our company use analytics. If he took the time to be honest he would say…”If that is all you need, GA works”.
The enterprise guys would engender more trust if they would actually admit that based on USE GA might be fine.
And while the enterprise tools have a lot of great features – most companies are still not taking action on anything other than traffic segmented by referring type / ad campaign.
Kudos for so clearly illustrating the case when enterprise analytics makes sense. Thank you!
I’ve implemented analytics for a number of websites and as you say, GA more than meets most organisations’ requirements. You also list the ways in which GA can be stretched and what can be done in those scenarios which is very useful for me. Because I’m not that familiar with enterprise analytics solutions, I wouldn’t mind knowing how they handle these scenarios. Omniture for example.
Elvind: I completely agree, that it is critical to track pre-click information when running any type of campaign (email, banner, cpc, etc.) But a good analytics tool that can tie conversions to source, and provide insight into what the visitors due while on the site is very important when tracking a campaign.
Will: Thanks for the vote of confidence. I must admit, that I am not the first to propose that people use a tool that fits their need. Avinash came up with that idea. My goal is o help people understand that if they do decide to go with GA that they should not under estimate the potential complexity of the implementation.
Nick: Glad you find the “workarounds” useful. If there is one thing I’ve learned it’s that every implementation will require some type of customization due to the unique setup of the client.
Wish I had more information about other web enterprise analytics apps, but I’m using mostly small free apps here and on other sites.
Thanks to all of you the thoughtful feedback.
While there are definitely GA-only issues here, i find that a lot of these issues can be applied to many analytics app. As you mention it, in the end it is more an organizational challenge than a technical one.
There is one feature that i do miss on many projects: segmentation on the fly. I find that it requires a lot (and i mean, a whole lot!) of planning and imagination in order to get all the slice’n dicing possibilities required to get the whole picture.
But in the end, GA always manage to meet my expectations.
Paul,
I totally agree. If you want to do Nth level segmentation with GA it takes planning. Creating all of those profiles is a lot of work. I’ve been asking Google for a profile ‘copy’ feature for a long time. It would make the whole process of creating ‘segment’ profiles much easier.
Thanks for the comment,
Justin
Hey Justin/Paul –
I definitely agree with the whole Profile copy recommendation. I can’t tell you how many times I’ve wanted to just copy an existing profile and make modifications.
It’d also be nice to be able to copy/use filters outside of actual accounts … but we digress …
Jeremy,
Thanks for stopping by, great to hear from you.
A ‘global’ filter library would be awesome, I never thought of that.
Thanks for the idea,
Justin
Hi Justin,
great article, you hit the nail on the head. I found GA useful for some ecommerce projects, after managing the tags for shop data.
A big disadvantage in my opinion is the limited csv/tsv export. You cannot get a complete list of all pages for example. And this makes using GA really hard when you want to use Excel.
Limited Segmentation is the other issue. Sometimes a client wants a special report for a single section of his site, but I cannot provide him history data in a new profile :-(
Markus,
I totally agree. The export in options are too limited. However, I think Google is really pushing us to use GA for more analysis. Hence the email reporting. Do your analysis in GA and then email a report instantly to a coworker. But there are just some times that I want to use Excel…
I also agree on the segmentation. Using profiles and funnels is such a buzz kill…
Thanks for the feedback.
Justin
Justin, does the 50,000 URL limit apply if you have GA reg ex filters setup that group all URLs viewed to LESS than 50,000? In other words, can one effectively use GA itself for their bucketing strategy instead of doing it via virtual page views?
If GA can provide a filter by landing page, that would be great. We always use some URL parameters to distinguish campaigns to the same page, but when you wanna build up profiles for each of them, you just can’t do it.
I know with utm_source,utm_medium…,this can be done. But my case is a little special, I’m dealing with all pop-up ads, millions of them and the result of using utm parameters is I can track only 1/10 to 1/5 of the actual visits.
I gusee it is a challenge to GA’s data processing ability and if there is a paid GA which has more hardware resource equipped, the problem should be resolved.
What’s your opinion, Justin?
Hi Clark,
You can now create an advanced segment to segment your data by landing page. I think the thing you may want to try is using advanced segments in conjunction with an advanced link tagging strategy.
Justin