Google Analytics is a fantastic tool. It allows you to get great insights, not only about your site traffic, but also about your users. It also has some really nice monetization monitoring right out of the box…assuming you’re an ecommerce site. But what if you’re not ecommerce focused? What if your site is monetized through ads or sponsorships? That’s a question we’ve had to answer here at WhitePages, and I believe we’ve come up with some great best practices.
Our primary means of monetization are two fold: through our endemic partners, and traditional display ads. Both of these are currently being served through Doubleclick for Publishers, or DFP. DFP, Google Analytics and DFA (Doubleclick for Advertisers) are all owned by Google. Unfortunately, however, only DFA and GA are able to communicate with each other, which leaves Publishers without an easy solution to tie together site traffic with revenue.
Speaking a Common Language
It’s always difficult to try to tie together data from two very different systems. The first step to doing so is to try to have the two speak the same language wherever possible. When first planning for our GA rollout, I took a look at the Site and Page Naming structure that we were currently using in DFP, and used that as a template. This way, I could make sure that when comparing reports, it would be an easy 1:1 comparison.
In order to do this, we make extensive use of Custom Variables. Our DFP page naming structure works on a Section.Subsection basis. So, I worked with our developers to have our first three Custom Variables pull in that data from our DFP ad calls. In Custom Variable 1-3, we’re passing the Section, Subsection, and Full Page Name (using the Section.Subsection structure). This allows me to see an aggregate of each section, each subsection, as well as the more granular page, and easily compare it with the revenue data I pull from DFP.
Related Resources from B2C
» Free Webcast: Build Better Products by Identifying and Validating Your Riskiest Assumptions
Evaluating Ad Performance with Traffic
This only allows us to see a part of the puzzle. Though it’s great to be able to see how each individual page is monetizing, as we make layout decisions, and decide to add or remove ads, we need to see how each individual ad on each page performs. In order to do this, I need to create a report in DFP that will show me the full page name (so I can link it up to the Visits, Pageviews and Unique Visitors data), but also each individual ad position, the revenue that position generates, and the Click Through Rate of each ad.
By having this report handy, I can easily tie the data back into any GA report that features our Custom Variables. I can look at our various traffic sources, and see how various users move through our site. Comparing this against the Ad specific report from DFP, I can get some insights to see if users from a certain source are clicking more often on ads, and therefore are more valuable users to our advertisers. With this data, we can not only make overall design decisions, but can also go to our advertisers and make specific recommendations about where they should be running on our site to get the best performance, and capture the best audience.
Accounting for the Shift to Mobile and Tablet
There are, of course, some significant roadblocks in this process. Like many sites, we’re seeing more and more of our audience interact with us through mobile devices and tablets. This can throw a wrench into the analysis. Though GA gives a lot of great options for reporting by device, DFP is currently severely lacking in that department. As a result, we cannot easily see how our revenue is impacted by this shift. Are these users more or less likely to click on ads? Do our ads load more slowly on these devices, leading to fewer overall impressions?
To begin down the path to answering these questions, there are a couple of viable methods, but all of them require you to make some very large assumptions. The first step is to take a close look at the data in GA. Are there major differences in the traffic patterns of Tablet/Mobile users? Do they tend to visit certain pages more than desktop users? Using this data, you can look at the DFP info, and take a close look at the pages that are more popular for our tablet/mobile users. Comparing the ad revenue and click through rate on those pages to the overall average should help you to get some insight as to how those users are interacting with our ads. If you see a lower level of impressions than you would expect, you can make the assumption that mobile and tablet users are not actually staying on the page long enough to see an ad. Conversely, if you see a higher than average click through rate, you can make the assumption that those users are more likely to click on an ad than a standard desktop user.
If you don’t see significant differences in the mobile/tablet user vs the desktop user, then you have to take a step back and look at the big picture. Look back at the past year, and see how the percentage of mobile/tablet users have increased or decreased. Then, look at the same time period in DFP, and look at if the average impressions, revenue and click through rate have increased or decreased. You’ll obviously have to make adjustments for expected changes, but if you notice that both Mobile/Tablet users and average CTR have been slowly growing over time, you can theoretically make the assumption that the two are linked, as long as you’re able to discount most other potential sources of the change.
Putting Data into Action
The last, and frankly most important part of this whole process is making sure the information we get from Google Analytics and DFP is distributed out to the Executive team, so it can help to drive discussions and ultimately action. We currently have a monthly meeting set up, where a full traffic and revenue breakdown is presented to the decision makers in the company. We can then discuss what the information means, what changes (if any) we should make, and what metrics we need to look for going forward.
It’s not a perfect process, and requires some manual effort, but it’s worked well for us. Hopefully these pointers will help others get the most out of GA and DFP, and we would love to hear any feedback you may have!