 |
|
 |
 |
I just thought I would fire off a quick post and share my experiences with "Advanced Segments" functionality in Google Analytics.
Imagine this scenario, you are granted "read" access to a client's GA account. Yes, "Read" access will allow you to view all reports. Missing within "read" access (amongst a several other capabilities) is the ability to filter data - in this case, traffic associated with your physical location as well as your client's location. You're stuck, and must ask the admin of this account to grant you "admin" access, ehh?
Enter "Advanced Segments". Using this functionality it is extremely simple to segment (or in this case, parse) out unneeded metrics.
At the left, you will notice "dimensions" and "metrics". If you are not familiar with the two data types, "dimensions" are the types of data you are looking to analyze while "metrics" are the values you would find within these dimensions.
 For this example, assuming you are met with the same situation I have mentioned above, log into your account and click "advanced segments" within the navigation on the left.
First, choose a dimension or metric to segement. For this example, choose Visitors > City. Drag the green "City graphic" into the dotted line area that says "dimension or metric".
Next, choose a condition. Since we are looking to segment out test data derived from a particular location or locations, our condition needs to be set to "Does not match exactly". The "Value" drop down will then be equal to whichever City you do not wish to record within your reports.
Repeat this with as many cities that you are wishing to remove from your reports. Just make you you follow the above-mentioned instructions and do not drop your metric/dimension within the "or" but rather the "and" location on the list.
Lastly, choose a name for this custom segment and click "Save Segment". That's it!
Back on the main page of your GA profile, you simply need to enable the segment. Click on the "Advanced Segments" button in the upper right of your screen. Next select the segment's name. Lastly, click "Apply."
Pretty simple, huh?Labels: Advanced Segmentation, Google Analytics, metrics, web analytics
I am so very tired of hearing that social media is not trackable. This is simply not true. Whenever you post a link to your own site, whether it's to highlight products, news, events, etc you should track the inbound data.
Because Google Analytics is so readily available, I will use this program as an example and provide instructions how to measure inbound links to your site using Facebook.
The easiest way to create a link that is trackable via Google Analytics is to visit Google's URL Builder page and fill in the various spaces. I would advise being consistent for all of your posted links that you track. Otherwise, you will have a difficult time measuring your various web campaigns effectively.
Follow these Steps:
Step 1:
- Enter in the URL for the page you wish to share
Step 2:
- Enter in the place where this link will be shared (in this particular instance, I chose "Facebook")
- Enter in a "Campaign Medium". This will be the type of data that is being shared (in this particular instance, I chose "Text Link")
- Enter in a "Campaign Name." What is this link? How would you identify it easily within Google?
Step 3:
Step 4:
- Many social media sites (including Facebook, although it is more generous at 421 chars) offering "Status Updates" require character limits on any posts. Therefore consider shortening your URLs with a "Short URL" app. Depending on where I am posting my update, I typically use Twitpwr.com or tinyurl.com.
- Copy your "Short URL"
Step 5:
- Paste the link onto your social media page.Viola, that's it!
The following day (again assuming you used Google Analytics to track your metrics), log in and go to "Traffic Sources" > "Campaigns". Mid-way down the page, look for the "Dimension" drop down menu. You can cycle through various filters such as "Source", "Campaign", "Medium" as well as other dimensions.
Labels: Facebook, Link Tracking, social media, web analytics
The team responsible for the development of Google Analytics (GA) has developed an extremely robust application and it's always improving. With the recent addition of it's custom reporting and advanced segment reporting, it even meets many of my needs.) That being said, there are several items that I would like to see implemented in the near future.
#1: Customizable dashboard metrics:
I am familiar with Coremetrics and feel it's dashboard customization capabilities are a wonderful complement to the application as a whole. Collected data can be viewed in any number of customizable outputs so it can provide a snapshot to the user.
With regards to GA, some enhancements would allow persistent views of customized data reports to be built in as well. Imagine having the ability to compile a series of customized metrics and view these within the dashboard.
Additionally, I would imagine that within various reports, a new button could be added globally called "add this view to my dashboard". By clicking on this hypothetical button, a snapshot of the particular report would then be saved to a user's dashboard.
Ideally, these customized dashboards would be attributed to an individual's account regardless of user type (administrator vs. user). I would imagine that various users would have separate preferences on snapshots of data, much like Coremetrics' allows as well. Much like the sharing capabilities found within "Advanced Segment" reports, these custom dashboards could also be shared.
Lastly, whereas GA only allows a maximum of (4) dashboards to appear, users could display an infinite number of dashboards modules (assuming performance issues would not be a factor). All dashboards, whether standard or custom, could then be sorted through a drag and drop mechanism, as is currently provided within the application.
#2: Compare metrics from other profiles:
Various profiles can share (1) account, primarily for administrative reasons. GA allows the notion of comparing various metrics as long as the data resides within (1) profile. It would be useful to be able to compare metrics from (2) or more profiles. As a result, individuals would then be able to compare data from subdomains, microsites, even filters.
Naturally, comparisons would be constrained to profiles within (1) GA account as all profiles within an account share the same Account ID. Only administrators of a given account would be able to set up this comparison. Users would be able to view this comparison report only if they have access to each compared domain within a particular account.
#3: Comparing non-sequential date ranges:
Currently GA allows you to choose a sequential date range and even compare it to another date range in the past. Another wonderful enhancement would allow a user the ability to select non-sequential days in a month or times within a given day. This is currently only possible through the "Custom Reporting" tool. I would imagine this sort of within the calendar selection tool (top-right of the application) as an "advanced date selection".
#4: Account migration and/or removal: 
This enhancement is more or less utilitarian in nature. If a user is an "administrator" they are, in essence, the owner of an account. An administrator can add other administrators or users however they can not transfer or remove access rights for themselves under an account they have created.
Imagine an administrator sets up a GA account through their own account. They then leave a job or are no longer responsible for administering an account. They can not "pass the baton" to another administrator if they so choose - they are in essence "stuck" with an account unless the account is deleted altogether.
Another related scenario, lies within a "user's" inability to be removed from an account. "Administrators" can grant account access to users and can delete them. Why can't user's delete their own access to a given account?
#5: Integration of Greasemonkey's "Business Notes" Feature:
Routinely, more than one individual reviews metric data. Since the creation of GA's Custom Reports feature, I have found that a little bit of explanation can go a long way in terms of deciphering a report. By merely providing a notes feature (bonus points for integrating this feature with Google Docs), individuals can review customized reports and make sense of the data by reading any explanation I provide for them.
Additional Recommendations:
- Integration of "Google Website Optimizer"
- Integration of Heatmaps
- Enhancements with Real-time data collection
- Enhanced support for off-site data collection
- Collection of offline data
Labels: A/B Testing, Coremetrics, Google Analytics, Link Tracking, Omniture, web analytics, Website Analysis
Let me start off by saying that having many friends on Facebook, Twitter, and/or LinkedIn is important. I am on all three of these sites and use them very actively for different reasons. When you have an audience, chances are greater that your message will be heard.
The problem with this philosophy is also that without properly considering your audience and hand-selecting who you are engaging with, you may also be wasting your resources - let me explain.
Say you work for a retail establishment. Your primary focus is bringing customers in-store to make a purchase. The website may only be a supporting agent to this objective.
By performing a social outreach campaign you attempt to reach out to anyone that will listen. By the end of your endeavors you may even have a formidible friend list.
Yet when looking at your analytic data you see less than 1% are converting. In looking back you may have even targeted the proper demographic audience. In this particular instance, if you did not geo-target this campaign you are likely wasting your energy.
Of course this is an example and it seems so plainly obvious. Yet why are many marketers still considering a successful campaign by have 1,000+ friends on Facebook or Twitter when an extremely miniscule amount of in-store traffic is resulting from these efforts?Labels: common sense, social media marketing, web analytics
Routinely I field questions from clients regarding the metrics of their site. The questions are typically related to how many visitors came to the site over the past month, how many pages were viewed on average. Inevitably the next question I am faced with is "is this a good number?" Much to the chagrin of the person asking the question, I inevitably respond with a simple "it depends."
At the end of the day, does it matter how many people are visiting a website or how many page views they see during a given session if these people are not performing your desired tasks.
The more that I analyze web data for clients and companies, the more I am realizing that these individuals that base their ROI models on the above metrics are really missing the boat. They are the same individuals that become frustrated that the reports are not telling them anything in particular. And they are correct, but it's also correct that they are not asking the correct questions.
Look, at the end of the day it truly depends on how your individual web presence is delivering for your customer. Just like you would be foolish to run a business without a business plan, so is running a website without a web strategy. By outlining clear goals and objectives you can begin to ask solid business questions and hone in in your desired response.
Questions such as:
- Have we been able to increase site utilization while decreasing reliance on our phone representatives?
- Are we increasing our revenue by posting full-featured demonstrations of our products online?
- What kind of impact are our online sales channels having on our offline sales efforts?
- How are we building brand value on the website?
- What specific pages appear to be posing a problem to our visitors?
As a stakeholder, certainly you can see by being able to answer questions you can begin to properly determine a furthered strategy. This is as opposed to merely saying on X date, Y amount of people came to my site, which does not tell you what they are doing once they are on your site.
Once you develop your questions and subsequently have them answered; develop additional questions. You will find that gauging success or failure on the web becomes more obvious.Labels: business questions, web analytics, web strategy
"What's the difference between hits and visits?" I took a deep breath before attempting to explain the answer to this million dollar question to a group of board members during a client pitch this past week.
Explaining web metrics in a clear and concise way is nothing new to me. Especially since many clients do not take the time to consider analyzing their own data and instead focus on throwing money away on endless paid-per-click search campaigns without any sort of knowledge. I learned early on in my career, if you could explain things in an easy method, breaking down information into bite-sized chunks, chances are much greater that you could close a sale. Any salesperson will tell you that a client buys into you more than the product or service you are selling. Therefore I always made it my number one goal to educate my clientele instead of trying to pull the wool over their eyes.
Pre-dot com bust everyone focused on getting their business online and paid no attention to any web reporting (despite the fact that reporting was also in it's infancy.) Fast forward to today and I question if really anything has changed.
Clients often times clamor to get online without first considering web metrics, and in turn their own audience at all. Fortune 500 clients still do not employ web metrics into their routine. Those that do, are often times paying hand over fist to simply know the amount of hits they receive from a website. The problem that any analyst will tell you is that the number of web hits mean extremely little when it comes to metrics.
I will go one step further and suggest that a number of visits to a given website can mean very little as well. In the case of individuals within a niche marketplace it is acceptable to believe that the desired demographic is going to be more finite than a website that can pander to an extremely wide-ranging audience. Therefore it's important to pay more attention to time spent on the site, page views and bounce rates.
For example, I would much rather 100 people spending 6 minutes or more on my website, thumbing through every page vs 1,000 people spending 10 seconds and going through 1 page (or even leaving directly from the homepage.)
So back to my introductory statement and the answer I gave the group of board members for a luxury retail client... "There is a huge difference between hits and visits, but neither of which should be of a primary concern to you..."Labels: metrics, web analytics
In the old days of the internet (approximately 2 years for you traditional marketing folks), data was measured in terms of one or two primary metrics. If you said "hits" please exit my site immediately. Thank you, now that I have your attention, let's continue...
Primary success metrics of a given online campaign were routinely measured by leveraging total visits with the average number of pageviews per visit. This was a gold mine. I could create seemingly limitless reports based on this most elementary data. Of course that was in the days of synchronously loading page data - before scripting methodologies such as AJAX had to complicate matters.
Fast forward to today, page views are still a main stay within analytics programs such as Google Analytics however I question whether or not they should be. This belief can still be summed up within a few short accronyms and one syllable words: AJAX, PDFs, and Flash.
This method of scripting, a document type and multimedia application have been causing havoc for web analysts for some time now. And Adobe, you still have not convinced me that Flash is completely queriable. Especially given the fact that so many people still use Flash within a timeline. That is instead of it's one frame object-oriented routines with simple "getURL" linking structures found within this said frame. (And this is just one example of Flash 'issues' with regards to analytics, I have several more). Like text that is broken apart or part of a graphic instead of dynamic text instances.
The point is creating 'fake' pages that only have the responsibility of being logged within analytics programs (to aid with AJAX and PDFs) is a tedious process.
So... What metrics do I particularly like to use when providing reports to clients? Every vertical marketplace is going to be different however if I had to pick two to replace the above two metrics in terms of equal importance it would most likely be "Bounce Rate" and "Average Time Spent on Site".
AJAX can have many calls to varying pages occurring within the same HTML based document. PDFs can have text that has been converted as an outline (and is essentially speaking, merely a graphic). Neither of my above-mentioned two metrics are clouded by AJAX, PDFs, and Flash.
With "Bounce Rate" I can see definitively that someone came to a given site and left immediately thereafter. If this rate is unusually high I can then monitor that cluster of users against a series of other metrics (visitor loyalty, and traffic sources) more effectively than I could have done with a "pageviews" metric on a site predominantly coded using a series of AJAX calls.
With "Average Time Spent on Site" I can still see that a given user spent (X) amount of time on a given site. Since I can correlate this data across a series of given days, I can analyze why time on site may have been so very much higher on Monday, June 23, 2008.
At the end of the day, the one quote I will always repeat is 'the only thing constant is change.' Within the area of internet technology this could not be more true as I have seen entire methodologies completely change online within 12-18 months.
That being said, as the fundamentals of a more engaging Web 2.0 are starting to be realized it becomes more important than ever that web analysts provide clean and accurate data to web strategists so that they can react with a proper course of action for a given web campaign.Labels: metrics, web analytics, web marketing
|
| Return
To Top |
|
|
