Higher Education is among the most competitive industry segments in digital marketing. This only makes sense, with 3,000+ accredited institutions of higher education competing for 12.3 million college students. The higher education market is estimated to gross $475 billion annually. Overall, education at all levels captures 9 percent of GNP in the US., making it the second largest market, second only to Healthcare.
With this level of spending, it’s not surprising to see extremely high levels of competition within digital marketing channels. This is especially true in the hyper competitive online education space. According to SpyFu.com, which tracks digital spending, advertisers are paying AdWords on average $25.03 per click for the key phrase “online nursing programs.”
Overall, colleges and universities, especially for-profit institutions, are backing their digital marketing budgets with robust budgets (source: SpyFu). University of Phoenix, a leader in the online education field, is spending $3.6 million a month on AdWords alone. Four institutions tracked by SpyFu are spending more than $1 million a month in AdWords.
For any advertiser planning to bring those kind of budgets to a digital marketing campaign, investing in a sophisticated web analytics platform like Google Analytics 360 is a no brainer. There are a myriad of integration features with other products like DART that make multi-channel funnels attribution much more accurate. My favorite GA 360 integration feature is the automatic data dump of unsampled data into BigQuery every day. While not all of the GA 360 data is imported, it currently includes 206 of the more useful dimensions and metrics.
The data in BigQuery is stored so that each “hit” (interaction with the analytics server) gets its own row of data. Examining a row of data for each hit is worth the time and effort when you have paid $25.03 to attract that initial hit. While a $25.03 click may provide substantial ROI for a $100,000-nursing program, advertisers and their agencies better be using the most advanced tracking tools to make sure the impact of that kind of spending is understood. “Hit” level tracking shows step-by-step user interaction with the website.
Typically, I started by querying BigQuery for a list of fullVisitorIds along with the number of sessions that user performed, filtered by the users that sent email requests for information. That results gives me a list of unique users and the number of times they visited the site during the timeframe (often 18 months). To learn more about the specific user journal, I then queried using a specific userid, in this case “1498883599952546297,’ requesting about 40 different data fields in order to learn more about what interactions might have influenced the desired outcome – an email request.
Among the initial observations, this request originated with a click-through from Google Analytics’ Partner page, where users can look up potential partners by state – obviously a fruitful potential lead source. The interactions are viewed from left-to-right starting with column B, with each hit producing one column of data. The email occurred in the final column, meaning that this user emailed us on his 10th interaction.
Here is a walk-through of each customer action, along with presumptions of their actions based on the data collected: The user arrived on the “home page” (hits_page_pagePath), then clicked on the top navigation link to resolute.com/google-analytics, which makes complete sense. The first takeaway from this observation is that we should ask Google to give us a link from their Partner page directly to the resolute.com/google-analytics page, thereby saving the user a step. On hits 3 and 4, the user returns to the home page (note to self –- figure out why). On hit 5, the user checks out the resolute.com/about page, thinks about launching an email on hit 6 on the resolute.com/contact page, and then returns once again to the home page (“do I really want to contact this agency?”). The session concludes with a click back to the resolute.com/google-analytics where the user submits the request for information.
I am able to demonstrate hit level analytics using a manual process because the number of inquiries we receive every day is relatively few. For a large online college spending $100,000 a day on AdWords and other digital marketing channels, doing this analysis manually would be nearly impossible.
Enter “machine learning” technologies, which are designed to figure out and optimize answers to questions like “what is the best user experience to provide on my website that results in a request for information or, better yet, an application” or “what are the most productive combinations of traffic course, geographies, demographics, and device types” for producing “applications.”
While many digital marketing managers might fear that machine learning techniques are beyond the capabilities of their existing tech teams and, while these approaches may not yet be commonplace, they certainly are becoming more common. With all the work that Google and Amazon have been doing over the past few years, it’s really just a matter to harnessing the open source tools to make it work.