This isn’t necessarily a bad thing, implementation has to be completed for your organization to have access to all the rich data you need to start building a data driven culture. However, what many companies fail to realize is that implementation is not limited to a technology task. Implementation also involves integrating your analytics practice into the fabric of your organization.

How many times have you heard something like this, “I’m so frustrated, our product team launched a new feature this morning and the CEO is demanding an analysis. We don’t even have the correct tracking on the new feature, I just found out about it this morning!!!”?

In our excitement to deploy our shiny new analytics solution, we rushed into the build without reading the instructions first. Had we slowed down at bit, we would have realized that one of the critical steps in the process was to ensure that the solution was integrated into our existing processes.

When I was running the analytics practice for 28 corporate brands at Spark Networks, one of the first things I did was to identify the process we followed for managing product enhancements. Internally, we used a solutions called JIRA to manage our software development lifecycle. It was critical that I fully understood the process for how new features were rolled out to our sites, from ideation to launch, and how our systems supported that process.

It became clear to me that I needed to integrate my analytics practice into the product request process that we managed within JIRA. The simple task of adding a few required fields to a JIRA request, made my life so much simpler and I never again had to tell the CEO, “we aren’t tracking that new feature yet.”

When a product manager submitted a feature request, they were now required to complete two additional fields in their request:

1) Is analytics required for this feature?
2) If yes, what are the requirements, if no, why not?

This simple update to the process forced product managers to stop and think about analytics as part of their process for rolling out enhancements to the site.

This example is only one place that I integrated analytics into the process and there are many, many more that I won’t go into detail on but to get you thinking, I also integrated my practice into:

• Bi-Monthly Executive Staff Meeting
• Creative Request Process
• Email Marketing Process
• Engineering Bug Tracking
• Quarterly Earnings Calls
• Product Launch Post-Mortem Reviews
• And Many, Many More

Take the time to look around your organization and identify the areas that are critical to running the business, find ways for integrating your analtyics practice into these areas and you will take a huge step forward in building a data driven culture.

Of course, to accomplish this, it takes adoption and buy-in from across the organization, so I will address the importance of user adoption in the next part of this series.

We have worked with our clients to implement a variety of tag management systems in both simple and extremely complex environments, and as part of every engagement we spend the time to make sure that the introduction of a TMS into their digital ecosystem is part of a larger governance plan.   But before we start doing any work on the planning and implementation we take the time to make sure the client understands exactly what it is they have purchased.

If you just read the marketing propaganda you will get the impression that the TMS you just signed the contract on is the greatest thing since…

…. the in car DVD players….way better than sliced bread.   Before we dive into what a Tag Management System can do for you, lets talk about what they can NOT do for you.

TMS is not Superman, it can not leap buildings in a single bound, there is no heat vision, or any other super powers either.   It also will not:

• Eliminate the need for technical resources on your team to properly implement and maintain your tagging infrastructure.
• Magically tag your site and align everything with your business goals, hopes, and dreams.
• Free you from the critical need to do exhaustive QA and validation of new and existing tags.

You have just purchased a very very nice…..wait for it…javascript deployment tool.  Sexy right?  Can’t you see people flocking to the streets clamoring for their very own javascript deployment tool?  No?  Neither can I.   But now that you have your own javascript deployment tool..er.. Tag Management System we can talk about the three main reasons that people need or want a TMS.

Speed:  You just want to go fast.  You want to reduce the length of time to not only deploy new code and tags to your site, but to gain the speed and agility to make code fixes when bad code makes it way to the site.

Reliability:  You want to bring consistency and reliability in the code,  You want the system you choose to have the highest level of availability and uptime to make sure that you are not introducing a system into the mix that is not going to stand up to the volumes of data that you will be passing through it.

Control:  Sometimes… IT is not a pleasant group to work with, and sometimes you feel like you are dealing with  “IT” and not the Technology team.  And once they figure out that they are losing control or some of their control you and your team you might have a bit of a fight on your hands.  Tag Management Systems also makes it easier, not easy, but easier to replace your analytics vendor if you choose to at some point in the future.

So what?  What does this mean to all of you out there that has purchased a Tag Management System or are planning to?  What is the real benefit of going through the effort to implement an enterprise class Tag Management System?  Here is the typical deployment cycle in a large enterprise without a TMS:

And here is the typical deployment cycle after a full TMS implmentation:

Before you say, I thought I was getting the ability to instantly push new tags to my site, these diagrams map out the entire lifecycle of the deployment of tags.  From the initial request to business approval to code,QA, and deployment.   The part of these two diagrams I want to make sure you see is that the QA section does not and can not go away.  It is true you have the ability to push code changes to your site as frequently as you would like, with amazing speed.  That also means you now have the ability to push really crappy code to your site with amazing speed.   Think about that for a minute and then think about who on your teams you trust to get the ability to with a click of the mouse the ability to bring your website to its knees with sub-standard code if proper QA and testing has not been done.   Trust me, you do not want to be this guy.

This guy skipped the QA part, pushed crappy code and was fired.   Don’t be this guy.

If you have purchased a TMS, or you are thinking about a TMS, you need to take a step back and evaluate your specific needs and business problems that you are hoping to solve.  At Keystone Solutions we have worked with a large variety of clients in all the major verticals and the various Tag Management Systems and would be able to sit down and review your plans to help you map out a solution that will not only give you the speed, reliability, and control you desire, but the peace of mind that you have thought through all of the pitfalls and common mistakes out there.

Leveraging both quantitative and qualitative data not only paints a holistic picture of your audience but also provides the insight you need to predict behavior. Ultimately, you need to be able to answer the “why?” – the root cause or reason behind the behavior.

Before starting any optimization test, the first thing you should do is take time and research who your audience is. Let’s say you run a test and the control wins, are you able to answer why the control won from a behavioral standpoint? If not, then you just wasted time and resources because the test did not provide any insight.

For example, we worked with a client who was sending Yahoo search traffic to a poorly performing, generic landing page. The quantitative data was there, but it wasn’t until we truly focused on the specific audience, that the landing page experienced a significant lift in conversions.

By collecting qualitative data from a competitive intelligence service, we were able to discover the demographics, lifestyle, and persona of the vistiors coming from Yahoo. Before, with the generic landing page, Yahoo users experienced a disconnect when coming to the page because it didn’t resonate with them, the message just wasn’t compelling. However, by incorporating the knowledge gained of visitors coming from Yahoo into the landing page experience, the landing page experience was optimized and conversion rates soared.

We were able to help answer the “why” and build tests off of a strong hypotheses from both quantitative and qualitative data. I’m not saying to drive your tests or create experiences solely off of qualitative data; rather, you should bridge quantitative with qualitative.

Take the time and know who your audience is by answering questions like:

1) What are the key pages people are visiting?

2) Once on the site, what actions are people taking?

3) What are the primary traffic sources?

4) What are the key demographics of site visitors?

5) Are there any social media trends?

6) Are visitors coming from specific geographic areas?

7) Can we develop unique personas of site visitors?

By layering qualitative data on top of your quantitative data, you can effectively show relevant content to your audience. Your audience is like a puzzle. If you try targeting with only half the puzzle completed, you may get some success, but you’re targeting half blind and it may result in wasted time, money, and resources. However, if you piece together the entire puzzle, the message you send is relevant and is more likely to move the needle. Keep in mind, there are several puzzles – don’t try to fit your audience into just one. You can’t expect what worked for one channel or vertical to work the same for all others. Traffic sources behave differently because the motivation is different.

You could produce a winning experience that works for your display channel but that doesn’t mean you will see the same results when tested against your search traffic. Why? Because visitors who come from display typically show different motivation from those vistiors coming from search. Just because an experience works for one channel doesn’t mean you should roll it out 100% – test it first.

Once you’ve solved the puzzle, you can show relevant content, answer the “why”, and predict behavior. You’ll spend less time, money, and resources to get the results you want.

Wow, Pinterest is really taking off as a valuable social traffic source for eCommerce sites. But for some reason I don’t remember seeing Pinterest listed among my top social media sites. This makes me wonder, are all of my marketing channel reports accurately showing their value?

If you are a big user of the Marketing Channel reports, at some point they will need to be updated. Specifically the social networks channel. Adobe does a good job keeping the list of known search engines pretty accurate across SiteCatalyst, but the same can’t be said about the list of social networks.

If your marketing channel reports were initially activated using the auto-setup option, they included a list of domains that are considered to be social networks by SiteCatalyst. Unfortunately that list does not get automatically updated over time like the list of search engines does, or like the list of domains used to identify the social networks in the traffic sources referrer type report does, and that’s fine. If I have edited the list of social network domains in the marketing channel rule, I don’t want it to be changed by some automatic update. But when using the auto-setup option to enable the marketing channels, for some reason the list of domains used to create the social network channel is considerably different than the list of domains used to identify a social network in the referrer type report.

The SiteCatalyst knowledge base answer #10576 contains a list of 128 different domains that are used in the Referrer Type report to identify social networks. In contrast if you take a look at the marketing channels processing rules to see which domains get pre-defined as social networks, that list is much smaller. Only 51 domains. I even went and enabled the marketing channels reports on a report suite that has never had them before today to check, and I still only saw the 51 domains included in the social network rule.

We can see that the rule to define social networks can accept up to 500 different domains, so why the default list is less than half of what is used for the referrer types report is a great question.

But wait a minute, I don’t remember seeing Pinterest in either of the lists of social networks that was used by the marketing channels reports or the referrer types reports. I decided to double check the reports themselves, so I opened up the referrer type report, broke it down by the referrer report, and in fact Pinterest was there. It is accurately considered to be a social network in the traffic sources reports. That tells me that even though the list of domains used to define a social network in KB answer #10576 was current as of November 2011 as it states, it in fact has been updated since then. Now checking the marketing channels reports manager, I see the rule which defines a social network still contains the original list of 51 domains. The social network marketing channel for my eCommerce sites has been missing out on the credit for all the conversions generated from that valuable Pinterest traffic, as well as any conversion credit from those other 77 domains considered to be social networks that were not included in that channels rule. My marketing channel reports were not providing an accurate representation of my social networks true value.

Whether you’re using the Unified Sources VISTA rule, the Channel Manager s_code.js plugin, the Marketing Channel Manager that’s built into SiteCatalyst, or some other channel identification method, make sure you are keeping it updated regularly to ensure all of your marketing channels are getting all the credit they deserve.

enjoy!

Web Analytics Wednesday (On a Tuesday) – For years now, the unofficial start to Summit has been the event put on by ObservePoint and of course, Rob Seolas. I look forward to this every year and I was not let down. They had over 300 people registered and it seemed as if most of them showed up based on the crowd. It’s a great time to network and socialize with friends and colleagues outside of the formality of the conference.

Adobe’s Opening night reception in the Partner Showcase – The showcase this year is really taking advantage of the extra space they have with the conference moving to the Salt Palace. The Exhibit hall was more open and it was easier to move through the hall and see the partners without feeling like you were being herded through a barn. Highlight of the exhibit hall so far is the Ice Cream machine in the back that if you smile and let it take a picture of you smiling it gives out free ice cream. Who does not like that?

Social Overview Report

All Google Analytics users will soon have access to a new report that allows you to quantify and visualize the social media contribution to your various site goals, including Ecommerce conversions. This report provides two separate measures of value: last interaction and assisted conversions.

A last interaction goal completion is logged any time a visitor is referred from a social media website and completes a goal during the same session. An assisted goal completion is attributed any time a goal is completed after a visitor is referred by a social media website, regardless of the visit that the completion occurred.

In most cases, the last interaction measure will understate the value of social media on your site, while the overly generous assisted attribution will overstate the contribution. This report effectively creates upper and lower bounds for quantifying social media value. The “true” value is somewhere between those bounds and will depend on the behavior of your sites’ visitors. This will come as a breath of fresh air for GA users who have struggled in the past to identify an aggregate social media valuation for their site based on an amount of retweets, likes, and +1’s.

The only requirement for this report is that you have at least one goal defined for your site. It’s important to note that this report is based on a large predefined list of social networks and is NOT dependent on the UTM link tracking that your organization is using to track traffic sources and marketing campaigns. This means that this functionality is available immediately and you won’t have to worry about making any implementation changes.

Social Conversions Report

In addition to aggregate measures, the “Conversions” report will be used to analyze conversions from individual referring social media sites. Last interaction and assisted conversions are both included along with a ratio of those two metrics. This ratio is meant to quantify whether the referring site is sending upper funnel (relatively high ratio) or direct converting traffic (relatively low ratio).

Off-Site Analytics

If your organization relies heavily on Google+ to engage with users, then you’re in luck. (If not, you probably won’t find this new feature incredibly useful.) Google has integrated off-site data from their Social Data Hub partners to provide greater insight into users’ behavior while on these participating social networks. This feature has the potential to be valuable in the future as the amount of partners increases.

There is still a long way to go to clearly define success within the realm of social media. Analysis in these channels must become more about how these networks help our sites accomplish their goals and less about ambiguous popularity metrics. It’s encouraging to see more analytics tools heading in that direction.

Here’s what has been happening behind the scenes. All signed in users are now on a secure version of Google (https), and a redirect has been added to each search results click. That redirect is to a non-secure page (http), where the referring URL is changed before the visitor arrives at the page they requested. That new referring URL value has had its keywords removed, but still contains enough information to determine it was a Google Secure search. Workarounds were created to help identify a Google Secure search in SiteCatalyst keyword reporting, as well as Omniture making a change to try and account for those searches.

Since making that change Google has determined that the additional redirect is unnecessary and potentially slowed down the users experience, so they have decided to eliminate it (unfortunately that does not mean analytics tools will be able to see those keywords again).

Today Google announced a change to the way they plan on handling referring URL’s starting in April 2012. Google has decided that they will now begin to use the referrer meta tag for browsers that will support it, as opposed to the redirect to the non secure page. Currently the only major browser that supports it is Google Chrome.

If you are not familiar with the referrer meta tag, what it does is it lets each web page decide how referrer from it should be handled. For example, here’s what a meta referrer tag looks like:

What this tag will do is that it tells the browser to never pass any referring information from the page its on. The browser should then set the referrer header value to a blank string for referrers from that page.

Fortunately Google is not going to that extreme. They have decided to use the “origin” value:

This is the referrer meta tag value that Google will begin to use in April 2012. When the change goes live, all search clicks from signed in users will now only have the referrer value of https://www.google.com/. There will be no other information in the referring URL, so no way to determine that it was specifically a Google secure search other than the URL being simply that host value. Non-secure searches, ones made from a user not logged into a Google account, will continue to function in the same way as they do now.

Currently the referrer meta tag is not currently supported in all browsers. I tried it using Chrome 17 and it is working. Testing it in Safari 5.1.4 and Firefox 11, the referrer meta tag has no impact.

So what does this mean as far as SiteCatalyst reporting? According to the Knowledge Base answer #5329, “If the domain of the referrer corresponds to that of a recognized search engine (e.g. “google.com”) and contains the recognized search keyword query parameter for the given search engine domain (e.g. for Google this is “q=”), then the referrer is considered a search engine, and the value of the keyword query parameter is taken as the search keyword.” So no search keyword query parameter, no search is counted. Currently for a Google secure search, the parameter is still there, but it’s unpopulated. Now Google plans to remove it all together.

Hopefully before Google rolls this out publicly Adobe will come up with a solution for SiteCatalyst so there are no interruptions in the Search Engine reporting. If Adobe does not get to it in time, or if Google decides to push the change out before April, then a couple of lines of code added to your s_code.js file will keep the impact to a minimum while Adobe works out a solution.

if(document.referrer=="https://www.google.com/"){
	s.referrer="https://www.google.com/?q=google%20secure%20search";
}

What this will do is look for a referrer with the exact value of https://www.google.com/ and append a q= value to it with the keyword of google secure search. If the .com version of Google is not the one most used by your visitors, then just replace it with the correct tld version applicable to your site. This snippet of code will make sure the search is still counted, and you will continue to keep the same level of reporting that you have now.

2) Innovation is alive and well at Adobe. I can not go into the details because of the NDA, but I was blown away with the special projects that the developers and product managers are working on. I for one hope that all of the sneak peeks I saw make it to the market in the very near future. But almost more important that the new projects was the fact that the company seems to be more committed to working on the projects and continuing to innovate at a rate that seems to have been missing over the past few years, and this is a great sign of things to come.

3) I miss the Grand America. The new space is nice and the event is well done, but there is just something about the Grand America that makes is feel more like an Analytics retreat than a conference. Perhaps I am just being sentimental.

Bring on Day 2