But before all that, some context.
A recent wonderful experience for me was participating in a analytics webinar hosted by NTEN (Nonprofit Technology Network). I am very fond of NTEN, they do amazing work (plus Holly and her team are awesome!).
This was my second outing with NTEN, and this time round the idea was simple. They ask questions about Google Analytics and I try to answer as many of them as quickly as I can (with a modicum of intelligence).
It was a lot of fun. Often we live in our high and mighty towers (everyone except me of course!) and it is nice to get a dose of reality. Real people with real problems.
Of course there were waaaay more questions than I could answer in the available time. So I promised that I would write a blog post to answer them. They get what they want, and I make up for the fact that in two years I have written two posts exclusively about GA!
These questions were in the context of Google Analytics. But I think they'll apply to many different web analytics tools, in as much those of you that use other tools might also find them to be of value.
Sit back, relax and enjoy the ride!
Google Analytics: "Frequently" Asked Questions.
How can nonprofits on content management systems best use Google Analytics when they don't have access to html code on the backend?
The nice thing is that almost all websites have something like a footer.html (or php or jhtml or whatever) that gets automatically added to all the pages to the site. If you can just add your .js tag to that file then you are set. So it is not a lot of work to add the base tag a site, and you don't need access to all the pages.
If are in the "out of luck" category, then you can opt to get access to your website's server logs and then use a log file based web analytics solution like Urchin from Google or ClickTracks (or in the honor of Dr. Stephen Turner: Analog!).
Remember the important part is that you don't have to live without data. : )
Here's a specific one:
I had a question about Exit Pages: if a visitor uses the 'Back' button to exit out of a page, what is counted as the exit page? Is it the last page they got to before they started backing out, or is it the last page the hit 'Back' on?
Exit pages are typically the last page that gets recorded in a website visitors session.
Let answer the question in layers.
Specific to back buttons the answer will depend on which type of solution you are using.
Google Analytics will report the Exit Page as the Home Page (the last page recorded in the session).
Navigation summary question – why is previous and next page often the same as the page you are viewing?
This is a good one, I was confused about it a while back as well.
The explanation of this one is really cool, it might sound technical but stick with me and you'll see what I mean by cool. . . . .
The report won't fit here so here is a badly butchered picture to show as an example (click here for a higher resolution picture):
The report is for my Web Analytics 2.0 post. And you'll notice that 15.90% of the referrers of traffic to that post is the post itself and the next page seen on the website by 15.90% of the visitors is that same post.
Cari explained it to me in September (when I was confused!). . . . .
It turns out that this page (web analytics 2.0) has close-ups of graphics (click on this image for higher resolution etc), and if a person reads the page, they will likely view the image / graphic, then hit the back button, causing a second hit.
This means that it will be its own "previous" page and its own "next" page.
Take this session history as an example:
Visitor Action One (view): /avinash/2007/09/rethink-web-analytics-introducing-web-analytics-20.html
Visitor Action Two (click): http://www.kaushik.net/avinash/wp-content/uploads/2007/09/web_analytics_1.0.png
Visitor Action Three (back): /avinash/2007/09/rethink-web-analytics-introducing-web-analytics-20.html
Visitor Action Four (click): http://www.kaushik.net/avinash/wp-content/uploads/2007/09/web_analytics_2.0.png
Visitor Action Five (back): /avinash/2007/09/rethink-web-analytics-introducing-web-analytics-20.html
To Google Analytics (or any other Analytics tool), it will look like this:
If you look at this in GA, you will see that the page was viewed 3 times, and that on 2 occasions, it was its own previous and on 2 occasions, it was its own next (!!). If this is the case, the previous and next values should always match (which it does in my case).
Amazed? Surprised? A] Cari is very good at what she does. B] Web Analytics is fun because it is such cool detective work!
Your navigational report might show the same page as the referrer because your Visitors might click Reload a bunch of times (generating "self referring" hits), and other such delightful behavior.
Benchmarking – how are the categories assigned or determined?
One of the features in Google Analytics is that those clients that opt-in their anonymous data can avail of features like Industry Benchmarking (Important : You have to make the choice to do this, you can choose to opt you data in anonymously or keep it private – not provide google permission to look at it. See: Data Sharing Setting FAQ.)
Here is the official answer to your question: When benchmarking is enabled, Google crawls the websites in the account then categorizes them by vertical and the amount of visits. The data is then made anonymous through aggregation.
In English: Google indexes and categorizes websites and uses that categorization to bucket your website.
But here is the nice thing:
Click on that link that says Open Category List, and you'll see this:
You can choose your own category, if you don't like the one you were placed in automatically by Google. Win-Win.
You get three levels of categories, in this case: Computer & Electronics -> Software -> Operating Systems. And as it says at the bottom, as more data is available more categories will be available to allow you some very granular comparisons.
For more delightful details see: Using the benchmarking service.
How valuable/accurate is the benchmark data in GA?
Let's break it into two pieces.
The data is valuable.
Here's a helpful analogy: If you only look at your web analytics data it is like sitting in a car with the dashboard in front of you. It tells you that you are going at 75 mph etc. But its as if the windshield and windows are blacked out, you have no idea what's going on around you. Using competitive intelligence data is like scraping that black paint from the windows so you can look out – and see that you are in a race and everyone else is driving at 200 mph. Context is king.
In terms of accuracy, when it comes to competitive intelligence that is such a tough word (for any tool / data provider). Here is how the team works hard on your behalf:
1) Google will only show you benchmarking data if it has a critical mass of websites in your category. This is to ensure that you get good data.
2) Additionally the team applies mathematical algorithms to ensure significance and that any outliers are smoothed out to ensure you have the best possible benchmark.
3) You will only see yourself compared to "like sized" websites to ensure you are looking a relatively apple to apple comparison. Your blog won't be compared to www.cnn.com , as an example.
Compare trends, ensure your category chosen is correct and you'll be set. Over time as more sites opt in and more data is available it will only strengthen the benchmark.
Our debate here-Do we have very specific niche topical blogs or have more general topic blogs so we "have bigger numbers"?
This is not a Analytics question, more like a blogging strategy question. IMHO unless you have something absolutely unique to offer (big number of bloggers, famous people to write, astounding offline reach etc etc etc) it is much hard to break through with a general topic blog.
You are probably very good at one thing, pick that topic and go. If you are good your unique audience will find you and Lady Success will be yours. Here's a helpful post: Ten More Blogging Tips From A Novice Blogger.
Funnels – ?? just a bit confused about how these things work!
Simple. You probably have a "structured experience" on your website. The simplest example of this is a cart and checkout experience. Or filling out a multi step application (credit card or a adwords account etc). Or sending a money order from a bank site.
Essentially anything that goes: Start -> Step One -> Step Two -> Step Three -> Done.
On your site perhaps: Add to Cart – > Start Checkout -> Billing Information -> Shipping Information -> Submit Order.
When you set up a goal in GA you have the opportunity to input the steps to get to that goal. In the above case the goal would be Submit Order and you can add the page urls from add to cart to shipping info. Then GA will automatically show you:
1) How many people enter at each step of the funnel.
2) How many people abandon at each step.
3) How many people make it to the next stage.
4) How many people make it all the way through.
Here is a funnel:
It tells you that you have a major problem on the first step, 76% abandonment.
Most people come to that first stage from mifotonatura.php but a good bunch also come from the home page.
Of the 25k that exit about 10k exit your site! They are upset! :)
The second step is doing better, 58% abandonment. And your overall conversion is 9.98%.
You know what to improve and fix.
Funnels are very good at understanding where structured experiences on your websites are failing. They are typically the easiest things to fix on your website, so please use 'em.
Sorry – can you repeat again how to set up funnel and goal?
Here are a couple of helpful links:
Funnel Question: Does funnel visualization parse old data or just start tracking new visitors after it is set up?
Only the data from the day that you set up the funnel (starting an hour after you set up the funnel).
This is a common "feature" in pretty much all web analytics tools, you have no (or severely limited) ability to "reprocess" data. Some of the analytics tools comes with options where you can purchase an add on data warehouse component that will allow you to "reprocess" data. But most off the shelf ASP based solutions don't allow reprocessing of data.
Also if you set up a funnel today and a month later you remove a step, or add a step, then that change is only going forward. It is important to make a note of it.
How can I DELETE an existing goal?
You can't actually delete a goal once it's been created.
But you can turn it off.
In Analytics Settings > Profile Settings > Goal Settings click the 'Active Goal Off' radio button.
Does it make sense or is it fair to compare data among a couple or more analytics solutions? For example Google Analytics and HBX?
Nothing in life is fair. :)
It seems like a good thing to have more than one tool on your website, just in case. And I should not complain, I have eight tools on this blog.
But I have often recommended on this blog, and in Web Analytics: An Hour A Day, that it is optimal that you end up with one tool. Life is hard enough in terms of making sense of one tool and getting it to provide actionable insights. So find the one that is optimal for you and then stick to it. This is clearly a case where less is more.
It is ok to start off with multiple tools, and rather than whittling down your choice on a RFP or a vendor dog and pony, pick from your experience in a production environment. But strive to get one.
If you do use more than one tool, say GA and HBX, then realize that numbers between different tools will never tie, it is simply impossible. They will tend to be in a narrow band and tend to flow in the same general direction. If you have more than one tool and if they are approximately 8 – 10% close to each other than thank your lucky stars and move on.
If they are beyond 10% different (and you want to use both tools!) then see this post on a recommended process for reconciling between different tools: Convert Data Skeptics: Document, Educate & Pick Your Poison.
Oh and good luck.
Describe the "bounce rate". It's always high.
Bounce rate is perhaps the sexiest web metric today. Period.
From a customer perspective it measures this phenomenon on your website: "I came. I puked. I left."
It measures the landing pages on your websites that stink so badly that they can't even mange to get the Visitors to make a single click. Sad.
It is also immensely actionable. When you look at the data, it is easy to know where you need to take action. So use Bounce Rate, it is your friend (if not a true BFF!).
There are obvious exceptions. Blogs. I don't use bounce rate for my overall blog traffic because many of you are returning visitors, you will read my latest post and leave. This will mean a high bounce rate, and that is quite ok.
When I look at bounce rate for the blog I segment it for my New Visitors. I don't want them to leave after one page view, I want them to read a couple posts, be convinced of the blog's greatness and sign up for the RSS feed. So in that case bounce rate can be a little helpful.
Learn a lot more about bounce rate and how to action it here: Standard Metrics Revisited: #3: Bounce Rate.
Are pageviews/uniques getting obsolete with Ajax, Flash, etc.
Yes, atleast page views (I am so sorry I don't know what was meant by uniques).
The method so far to deal with rich internet experiences was to create fake page views. Just so the web analytics tool would measure something. But this only resulted in lots of painful tagging work (which you are probably still doing) and this fake page view data polluted your real page view data resulting in a big mess.
We are now entering a new phase of evolution where we are going to use a new way to collect data, its called "Event Logging". Google Analytics uses this, and there are atleast one or two other analytics tools that use true event logging.
No more fake page views, no more six year implementation cycles. And most of all you get to define up front exactly what "business interactions" you want to capture and you are able to do that at scale and analyze them with reports uniquely created for those rich media experiences.
Pageviews as a measure of success are on their way out, but they will still be around for some time (most of the web is still static). They will measure less and less of the complete customer experience picture.
Here's way more about GA Event Tracking than you ever wanted to know: Event Tracking Overview (Beta).
Do you need to use the newer Google Analytics code to add the event logging code?
Click on the GA logo on the top left (after you are logged into the tool). You'll be on the Analytics Setting page.
Click on Edit link next to the name of the website.
(And this is silly and waaaay hidden) Click on Check Status.
Click on the New Tracking Code (ga.js) tab.
Copy the code, replace your current (urchin.js) and you are good to go.
When analyzing navigation paths on my website, it gets really challenging because all our pages are interlinked and each can be accessed from a variety of pages. So when I ask "how many people clicked donate from the homepage and actually completed the donation form?" I can't get an answer because I can't tell from the donation page who got there from the homepage and then actually went on to the confirmation page. Ideas on this kind of multi-page navigation given lots of page re-use?
Long question, and a very complex one. But, and it is magnificent, I have perfect answer for you! A real hidden gem in Google Analytics (that sadly very few people use).
The Entrance Paths report.
Go to any page you want, in your case your home page, by going to Content -> Top Content (or Content by Title). In the summary of the page (data like page views, time on page, exits etc) you'll see, on the left, a link called Entrance Paths. CLICK!
This is what you'll see:
You are looking at the Home Page and in the middle are all the possible clicks people can make on that page. Choose the link that says, in your case, Donation Form in the middle column. The report will transform to this…
Let's say the second link in the middle column is the Donation Form link, the column on the right shows where those Visitors ended up. You'll easily be able to locate the Donation Form Completion Page and see how many donated. In my picture above let's say that is the third link so it would be 246 (9.27% of those who saw the Home Page).
I love this report because it is the anti path analysis (Path Analysis: Not A Good Use of Time). This report compresses all the other pages people might have seen and paths they might have followed. It shows you very effectively if that page (your home page) is doing the job it is supposed to be doing.
Very helpful. It takes a little thinking, but remember God helps those who help themselves. :)
How well does google analytics filter out bots? Can I be assured that my results are not skewed by bots?
This is not the case with logs based solutions, for example log file parsers from WebTrends (or Urchin or ClickTracks mentioned above). It is a constant battle to filter bot traffic from server logs.
Also you should know that the Vendors, including Google, has mechanisms in place to be very proactive with their monitoring algorithms to monitor such stuff. They are not always 100% fool proof, but they are on it all the time.
Not the answer you want, but I hope it helps a bit.
Can Google Analytics be used to combine stats from 2 different web servers w/2 different sub-domains?
Here's Alex with the answer. . . .
Unlike traditional log analysis packages, Google Analytics doesn't rely on your web server(s) access logs. Since it's a fully hosted solution, it doesn't really matter where or how your website is hosted, so long as you can add the tracking code to every single page on your website. If your site spans 2 different sub-domains and your goal is to track the two sub-domains as one site:
(1) Create a Google Analytics account for yoursite.com
(2) Modify the tracking code as stated in here: How do I track all of the sub domains for my site in one profile?
(3) Add the same tracking code to both sub-domains
How do I merge 3 or 4 websites into one set of stats?
This one can be a non trivial amount of work (in Google Analytics and many other tools). It can be done, but you are best placed in recruiting one of the many wonderful Google Analytics Authorized Consultants (GAAC's) to help you with it. They are all over the world and they offer services at all price points.
Once again, here is the GA Supreme Guru Alex. . . .
In the absence of roll-up reporting, and passing first cookie information from one domain to the next (to de-dupe unique visitors) can get tricky if you have more than 2 domains. If you're not so worried about unique visitors, you can simply add the same tracking code to all your domains to log the data into 1 "super" profile. [AK: Cheap trick! :)]
You can try linking a visitors navigation between the 3-4 websites together by tagging all outbound links that point to one another, using the _linker() method, but if a visitor doesn't click on the tagged links to get to each site, and instead navigates to each domain directly, Analytics will identify each that visitor as 4 unique visitors (one for each domain hit) in the super profile.
Not exactly straightforward.
Here's a helpful article: How do I install the tracking code if my site spans multiple domains?
If you only have 2 domains
Shortest follow up question ever!
1] ALWAYS _link() method method on any links that point from one of your domains to the other: What is _link() and how can it help me?
(2) Do as this article says: How do I install the tracking code if my site spans multiple domains?
What are the best reports to get top management to pay attention to web stats?
After two years of MBA school the one thing I learned was that you can answer any question with this: It depends. :)
All kidding aside, that is quite true in this case.
See this post for more awesome, practical, ideas: How To Excite People About Web Analytics: Five Tips.
And rather than reports might I suggest that you initially focus on reporting to them metrics that will connect with them and provide deep actionable insights.
My advice, pick metrics that pass these four filters: They are Uncomplex, Relevant, Timely and Instantly Useful.
Indepth details on strategy here: Web Metrics Demystified.
For dynamic sites, esp using AJAX, can the goal URL have a dynamic attribute such as mypage.aspx?ID=Success instead of a totally separate URL?
I am way out of my league with this, here's Alex again. . . .
I think what this question is asking is "how do I differentiate a goal page from my funnel pages if all appear to have the same URL?"
The answer is to use the _pageTracker(); method to generate fake pageviews each time a goal is "hit". See the section titled "Identical URLs across multiple steps" in this help center article.
– – –
You are a tough bunch. But that was a lot of fun, got me to share some of the knowledge I had and a couple things I feel excited about.
I hope you learned a lot as well and that this was productive.
Did we get the answers right here? Would you add / remove anything? Was something mis represented? Did I forget something? Is there a unique way you have solved one of these problems?
Please share your feedback with us, it would be greatly appreciated.
The most scalable way to get help with Google Analytics is sadly not this blog.
Google provides direct email support in 27 languages, simply click on the Contact Us link at the bottom of any Google Analytics report.
Option two is the wonderful Analytics Help website, run by Google.
Option three is the fantastic Analytics Google Groups site, run by our users (who we love and adore and cherish!).
Option four is getting Accountable Support through GAAC's – Google Analytics Authorized Consultants. This option is very affordable, the GAAC's provide full service or a la carte service at very competitive price points – worldwide!