15. Benefits of Tag Management System (TMS)
• Almost independent of development teams
• Faster
• Less communication issues
• Scalability
16. Challenges
• Complexities common in large organizations
• Learning the Tag Management System
• Access for external agency
• Transitioning from agency to in-house
• Changing technology
17. Results
• Correlate original source, campaign & engagement to revenue
• Greater marketing accountability
• Correlate optimization testing with revenue
• Blend full user journey data with CRM
• Better forecasting & prediction models
Original
source
Franchise
Site
Store
34. Blending GA & CRM Data
GA Data
• EA Customer ID
• Download Clicks
• Only Origin Purchases
CRM Data
• EA Customer ID
• Completed Downloads
• All Purchases (offline, consoles)
40. Results
• Download completions went from 10% to 47%
• Purchase conversion rate for downloaders 3x the site average
• More homepage real estate
41. Takeaways
• Align across all teams who are involved/affected in the funnel
• Remember to visit your sites and interact with features
• There’s usually more than meets the eye
• Find a way to get data for the entire funnel
Good morning! Has everyone here heard of EA, EA Sports or Electronic Arts? Anyone play games or have friends or family that play?
So EA is one of the largest video game development and publishing companies in the world and with that we have over 35 websites for marketing our games, each supporting a dozen or more countries and we also have an online store called Origin. We have several web development teams globally that build and maintain these sites. Now most of these websites are on their own domain, like www.battlefield.com or dragonage.com and Origin.com. And that creates some interesting issues that we have had to solve for. So that is the landscape from which I’m going to share some examples of how we at EA have measured the value of content:
There’s 3 areas that I’m going to cover: First of all, start with Cross Domain Tracking & Revenue Attribution in this multi-site context as it is the foundation for the rest of the measurement we do.
Then I’m going to share an example of how we have measured fun on one of our sites, and end up with how blending data sources gave us valuable insight on what was hindering conversion rates.
How many of you have multiples sites that you are responsible for and that are part of the same ecosystem?
Let’s review a basic description of what cross-domain tracking accomplishes for those of you who aren’t familiar with it.
Here is a typical user journey for someone looking to buy an EA game. A user starts their journey from somewhere say a Google search result or a facebook post, and goes to one of our game sites and then decides to buy the game, that takes them over to the Origin store which is on a different domain than the game site.
The problem, from a typical data collection standpoint, is that when the user leaves one domain, the session is over. You have no visibility into what the user does or what site they go to next.
Here is why it is so important from a business analysis standpoint: if I own or am responsible for the game website, I can see where all my visitors are coming from – paid search, display ads, social and I know their activity on my website but as soon as they leave my website, I have no visibility what happens after that. I have no way of knowing how the activity up to this point affects the ecommerce part of the journey.
Now the Store can see the website that is referring traffic, but the initial source – the post, or ad from where the user started their journey is gone.
The solution to tie the entire session together is cross-domain tracking. If you only have a couple of sites where you need to implement cross-domain tracking, it is pretty straightforward. But when you have a network of sites like EA does with multiple development teams, and hundreds of stakeholders wanting straightforward access to the data, things get far more challenging.
I should mention that it was good timing to find a clean solution as we were also moving the company from Site Catalyst to Google Analytics Premium so we could start from a clean slate.
As I mentioned, cross-domain tracking works by extending the visitor’s session from one domain to another and this is done using cookies. When a visitor lands anywhere on an EA site, in this example it’s Battlefield.com, GA creates a set of cookies storing source / medium / campaign that brought the user there.
When the Visitor clicks on a link pointing to the store , data stored in those GA cookies is passed along to the store and new cookies created there inherit the values of the cookies passed to them.
Now the Store also wanted to know which EA site was referring the traffic, in addition to the original source & medium, so we put that in a custom variable.
Session activity for Battlefield.com is recorded in BF’s Google Analytics account and session activity for Origin is recorded in Origin’s account. That makes sense, but here’s the cool part.
But here’s where it really gets cool we also added logic that said if the referring site is within the EA network, fire a second tracker to record the session and any transaction data and send it to the referring site’s GA account as well.
Battlefield.com traffic data, event tracking and custom variables
Origin Store traffic, ecommerce data and custom variables if visit came from battlefield.com
Source, Medium, Campaign revenue attribution
They don’t need to go into multiple accounts to see the data they care about – it’s all there in one place.
And here’s an example of what the owners of Origin.com in a GA report: they see all traffic,interaction, transactions data for the store regardless of where the traffic came from, and they can attribute that back to the original source as well as to the referring site within the EA network.
This could have been a nightmare to implement this code on 35+ sites with multiple web development teams across North America but thankfully, we had a tool that helped. Can anyone guess?
Yes, thankfully, we had recently purchased a tag management system, Tealium, and we used the TMS to deliver the solution throughout the EA network. The benefit to this was that our team’s javascript developer could implement the whole thing himself. Instead of working with each web dev team individually, most of them with little analytics implementation experience, we had full control to do the implementation and testing on our schedule. This saved an enormous amount of time and communication issues.
We did have our share of hurdles but they mostly were due to the complexities of a large organization – multiple stakeholders, communication issues.
-First we had to learn the Tag Management System which was still pretty new to market and had some idiosyncrasies.
-We had hired an agency to come up with code but they were only able to take us so far. It was impossible to get them access to our internal servers to test the code with a variety of different ecommerce flows including a pop up window that was being updated intermittently
-We had to bring in-house a specialized JS developer to tweak the logic depending on the site do the actual rollout to all the sites.
But the results were worth it - we now have a full view of user journey - this gives us a richer data set as now we can correlate source & campaign data with engagement data and revenue data which leads to greater marketing accountability. Optimization testing can now be correlated to revenue as well as better forecasting & prediction models.
Now I’m going to show you some practical examples of how we used this tied together data to optimize content on one of our sites.
One of the primary objectives of our game websites is to engage customers – the idea behind that is that through providing an exciting and fun experience even while they’re learning about the game, we’re building relationships and that’s part of the reason they buy our games.
The problem is, apart from a survey, it’s hard to measure the level of fun users are having, and that’s what we were tasked to do.
For those of you who aren’t familiar with the Sims, it is a virtual reality game where players create Sims (virtual people) and construct a world around them, their houses, their work, their families, even their moods etc. Last year the Sims marketing team decided to build a 10 question quiz for the site “What’s your Sims personality?”, with the primary objective to engage the customers.
One of the practices that my team works hard to instill with marketing, is to match their their online tactics to their business objectives, and to set measurable targets for those tactics. Here is an example of the model we follow. So the first objective up there is “Engage fans” The Key Performance Indicators that we felt would measure the effectiveness of this tactic towards reach this objective were “The Take Quiz Click Thru Rate” which was the start of the rate and Quiz Completion Rate. And then we like to crystallize why this KPI makes sense under the Success Column. The Take Quiz click, indicates the initial interest users have with the quiz, and Completion rate indicates how long we’re able to keep that interest. Note in the next column, that we list the targets for the Key Performance Indicators, if we don’t meet those targets, we make changes to the tactic. What I’m highlighting here is that by going through this process, we are priming everyone that the data we collect needs to lead to action, not just, oh, that’s interesting.
And then, we start to flesh out how we’re going to track the KPIs with the analytics tool, this can affect how the web development team builds the feature.
For instance, in order to track how many start and completed the quiz, and where the drop off points were, we asked that each of the 10 questions in the quiz have a separate url. This allowed us to set up a funnel in GA, to track the progression of users moving from question 1 through to question 10 to get a quiz completion rate.
One of the other KPIs was the Share Rate - to see how many players would think this was fun enough experience to share their personality type on FB when they were done the quiz. So, as you can see I ended up Spaced-out Artist, I guess that’s what I am in my spare time, so we set up event tracking so that we could track it as a goal in Google Analytics.
There were several different personality types that the quiz generated based on how you answered the questions, Happy Romantic, Bored-Creative, and our retargeting team had a cool idea. They asked us to grab the personality type of each user and store it in a custom variable. We then had our tag management system take that custom variable and surface it in the data layer to so that they could retarget users based on their sims personality. So for example, all Spaced out Artists would see different retargeting ads than Happy Romantics.
We did one more thing with the social tracking, and that was to see the strength of the quiz in creating a viral effect. We added campaign parameters to the share url and that enabled us to see in GA, if people were coming to the site, because they saw the quiz in their friends post.
Here’s a report from GA that shows not only the sessions generated from the fans’ facebook posts, it shows us which personality types these referrals came from because we gave each personality a separate campaign id. Like the angry-romantic down there at #5 they drove over 1000 referrals to the site during the time period. I mean do angry-romantics even have friends?!
You can see, the average number of pages shows that these referrals also completed the quiz.
So now for the results. How effective was the quiz in engaging fans in a fun way?
Well, because we took the time to really break down what marketing was trying to accomplish and put all the proper tracking in place, we were able to measure it! That’s really exciting because often it’s such a rush to get new features out the door that tracking and measurement is an afterthought. That’s happening less and less as we become more data-driven, but still it’s something my teams works very hard on evangelizing.
The quiz was overwhelmingly engaging – there was an 80% completion rate and over 8% shared their result on Facebook. Now that might not sound like a high % of shares, but it surpassed our targets and set a new bar for what good likes like for social interactive features.
We acquired a bunch of new users because a high % of friends who saw posts on FB came to the site and completed the quiz as well.
As a result of the quiz success, a quiz template was developed to easily add this functionality to any site. We just launched another quiz on the Sims and we are seeing the same level of engagement & Social Sharing.
And incidentally, the objective of the quiz wasn’t necessarily to drive purchases, but I was curious to see if those that completed a quiz had a higher propensity to buy the game. What I found with the first quiz was there was no difference between those that took the quiz or not. For this reason, I suggested that the Share button change to a Purchase CTA after someone shared their personality. And so when we launched the next quiz “Which career is right for you?” we added this.
Now 60% of purchases on the store are from Quiz Completers! We’ve permanently added a Purchase CTA to the last page and we’ve done A/B testing on the call to action text. Instead of “Learn More,” turns out “Get it Now!” wins by a big margin!
Okay, moving on now to an example where the entire customer journey was NOT visible to us in Google Analytics. The journey started on our website, but only the CRM captured if someone completed it because part of the journey involved registering a product. The potential problem we faced here was making assumptions about the success of a strategy without knowing the full story.
The “Create a Sim Demo” was an idea from marketing to generate purchase interest in the Sims 4. The tactic was a demo that you downloaded from the site and then experience how easy and fun it was to create a Sim, a virtual human being.
Now, full disclosure, I am not a gamer and I never have been, but I actually wanted to try this thing. I got a chance to try it while before it went live and I thought – this is so cool! I actually shared it on FB, which is something Ix reserve for only the novelist ideas. I thought for sure this thing was going generate purchase interest in the game.
Turns out people really liked the idea of a demo. In fact, 67% of people who saw there was demo, clicked to download it.
But did those people purchase the game at a higher rate? PAUSE….What do you think? NO! Google Analytics showed that the segment of visitors who downloaded the demo had a lower purchase conversion rate than those who didn’t download!
At first glance, it appeared the demo was a flop. But I was really curious so I decided to take a closer look at what was happening after people clicked “Get it Now.”
I was shocked, it turned out you had to work really hard to actually complete the download process. When I had tried the demo it wasn’t live yet, so I had no idea. But there were 5 steps to actually get the thing!
Step 1 was clicking download on the Sims site, this took you to the store site, Origin.com Then step 3 click another download but if you clicked the wrong button you would actually be adding the game to your cart. Then, step 4 you add to either sign in or create an Origin account. And finally step 5 download the Origin client which would load the demo. Yikes! It became apparent that just because someone clicked to download on our site, didn’t mean they actually completed the demo.
The only good thing about the journey was step 4 which involved logging in or creating an EA account. This step automatically assigns an EA Customer ID which we collect in a customer variable in Google Analytics and that is what would allow us to cross-reference with our CRM data.
(Side note with a prize on the line- Does anyone know how we could collect Customer IDs in the Sim’s GA account even though that happened on Origin.com and not the Sim’s website? That’s right! Going back to the first presentation, cross-domain tracking, we extended the session from thesims.com to origin.com, so all of the custom variable data was available not only in the Origin Store’s GA account but also in the Sim’s GA account.)
We took our data from GA - the Customer IDs, DownloadClick Events (which was the start of the download process) and Origin Store Purchases. We blended that with the CRM data – Customer ID, Completed demo downloads, and all purchases of the game made on Origin or through other avenues of buying such as offline in retail outlets or through consoles.
Here’s what we found. Remember I said 67% of users clicked to download the demo? How many do you think actually completed it?
Only 10% of users actually completed it!
And here’s the kicker - the purchase CR for users who completed the download was 3x higher than the site average.
Bottom line, it was clear that the demo itself was not hindering purchase conversion this was actually a great conversion tool. But the 5 step journey was clearly impacting people from completing the download process.
I put together a presentation of this data for marketing and the Origin store and together with our U/X team, they made some serious improvements to remove the points of friction.
They reduced the steps from 5 steps to 3 and took away the need for the user to leave the Sims site by adding an iframe.
I was shocked, it turned out you had to work really hard to actually complete the download process. When I had tried the demo it wasn’t live yet, so I had no idea. But there were 5 steps to actually get the thing!
Step 1 was clicking download on the Sims site, this took you to the store site, Origin.com Then step 3 click another download but if you clicked the wrong button you would actually be adding the game to your cart. Then, step 4 you add to either sign in or create an Origin account. And finally step 5 download the Origin client which would load the demo. Yikes! It became apparent that just because someone clicked to download on our site, didn’t mean they actually completed the demo.
So some key takeaways from this:
Align across all teams who are involved or affected in the journey. I mean even if we weren’t able to make all the improvements at the beginning, we could have done significantly better than what we ended up with just by having conversations with the folks in the Origin Store.
Remember to visit your sites and interact with features – go through all the steps an end user will have to go through so that you can be another set of eyes spotting potential points of friction. And then work with U/X to make iterative optimizations.
There’s usually more than meets the eye – if something seems fishy about the data, there’s a good chance there is. Try to look at it from different angles before you make any conclusions.
And lastly, find a way to get data for the entire user journey so that you can connect the dots across to uncover insights that you can act upon!
In Summary, I hope I’ve highlighted the advantages of being able to see the entire user journey to optimize each step along the way.
-Cross-domain tracking in a complex, multi-site environment is within reach and leveraging a tag management system will make this significantly more manageable
-To obtain actionable insights takes up front effort in working with the marketing team to identify key performance indicators & targets and working with the dev team to make sure those things are measureable and then following through with testing
-And finally, the importance of getting all of the data involved in a customer journey in order to uncover opportunities or points of friction.
Thanks very much for having me.