Like many digital marketers, I’ve spent some time living in Google Analytics (GA for short). I understand that muttering obscenities under one’s breath is a key aspect of the marketing analytics experience.
(Full disclosure: we use GA internally for web analytics at my business. That said, I don’t intend to be either a cheerleader for Google or a cranky Amazon reviewer.)
GA has evolved into a “one-size-fits-all” platform for small business web analytics thanks to one simple fact: It’s free. Up to a point, of course, but still free.
If you’ve been around the block with marketing analytics, however, you might have this reaction to one-size-fits-all solutions:
So, we’ll focus here on the limits of GA when it comes to understanding the impact of your marketing efforts.
GA offers numerous features for solving some of the thornier problems in marketing analytics. We’ll examine how these features can help you get started with advanced marketing analytics, but we’ll also explore cases in which you should probably look to third-party tools.
Here’s what we’ll cover:
GA Better for Data Collection Than Data Analysis
Core Limitations of GA for Digital Marketing Analytics
Integrating Data Sources Is Complex and Time-Consuming
Attribution Modeling Remains an Art, Despite the Comparison Tool
GA Better for Data Collection Than Data Analysis
To some extent, Google Analytics isn’t quite the right name for Google’s platform, or at least for the ways many organizations are using GA.
Frequently, GA is employed to gather data about web traffic, but other tools are brought in to analyze the traffic.
In other words, you’d do better to think of GA as a cloud-based data collection platform than a data analysis platform.
The following diagram shows how GA works as a data collection platform by gathering data from multiple sources and feeding it into analytics interfaces:
GA falls short, however, when it comes to analyzing the web data you’ve collected for marketing purposes.
As the above diagram shows, data can be fed from GA via .CSV exports or API connections into third-party analytics tools, such as spreadsheet tools (Microsoft Excel and Google Sheets) or BI platforms for visual analytics and dashboards (Tableau and Sisense).
Marketing analytics is of course a far broader field than web analytics, so feeding data from GA into third-party tools is inevitable in many cases.
Let’s turn now to the challenges in digital marketing analytics that are bigger than GA.
Core Limitations of GA for Digital Marketing Analytics
Despite GA’s robust functionality in the area of data collection, you’re still going to face the following challenges in analyzing the data you’ve collected.
Integrating Data Sources Is Complex and Time-Consuming
This is the big one for me. While Google is an excellent interface for analyzing web data, it’s a poor interface for blending web data with other types of data (accounting data, product data, customer data etc.) on the fly.
Thus, in order to understand the impact of our marketing efforts, I typically combine GA data from custom reports with data from other web analytics tools, revenue data etc. using Microsoft Excel.
Yes, this can be a huge pain, but it’s also easy, and most marketers are familiar with Excel.
Spreadsheets are one way to solve the data integration problem. Another way is to use the Google Analytics API to push data about web traffic into a cloud-based BI system.
Marketers can then blend GA data with other types of data using a visual analytics interface to discover new insights. Blends of GA data with revenue data, customer data etc. can also be displayed in dashboards that can be rolled out to the entire organization.
GA also supports so-called “universal analytics,” i.e., sending Google other kinds of data besides data about web traffic, such as data from your CRM system. This can be done by configuring custom dimensions and metrics.
The question, however, is whether you want to analyze all of the data you send to GA in GA.
In many cases, you’ll be sending GA data from your CRM system, accounting system etc. to integrate it with data about website traffic, before sending this new data set to an external analytics tool (a spreadsheet, a dashboard, a visual analytics tool etc.) for further analysis.
Custom dimensions and metrics are useful when you regularly need to combine data from a CRM system or another transactional database with GA data. For one-off analyses, you can simply blend the data in a BI tool or Excel.
Recommendation: GA is evolving into a data collection and integration platform. Despite the name, it makes more sense to view Google’s “universal analytics” as an expansion of the platform’s data integration capabilities than its data analysis capabilities.
Attribution Modeling Remains an Art, Despite the Comparison Tool
Chances are pretty good that you’ve witnessed a bitter battle or two over attribution in your marketing department.
For those of you who are new to digital marketing, attribution is basically who or what gets the credit for a conversion (a Facebook ad, an organically ranking blog entry, a mobile app etc.).
Attribution is absolutely fundamental to digital marketing, since it shows you which channels are driving revenue and thus how to allocate marketing spend.
The problem with attribution is that most visitors interact with your digital channels numerous times and in a variety of ways before converting—whether you’re defining a conversion as making a purchase, creating an account or signing up for an email newsletter.
So if a visitor finally converts via retargeting ads (i.e., ads that display based on cookies), but had already been reading blog articles on your site for months, should the content team or the advertising team get credit for the conversion?
The answer is that it depends on your attribution model, i.e., the rules you use to determine how different channels (social media, partner sites, blogs etc.) contribute to a conversion.
So what’s the catch? Gartner, a leading advisory firm that researches IT options for enterprise businesses, explains in Understand Attribution and Marketing Mix Modeling that “to date, there is no fancy software tool or quick consulting process that will meet the needs of any but the simplest campaign” (full content available to Gartner clients).
In other words, if you expect your software to foolproof the process of attribution modeling, you’re in for a disappointment.
And yet Google offers this nifty Attribution Model Comparison Tool:
So what gives? Why can’t you just do attribution modeling in GA?
Well, as Gartner explains, building mathematical models of your marketing efforts ain’t easy:
“For example, consider a marketer who wants to improve the performance of her display and search advertising efforts. To isolate the true dollar impact of display and search and to determine the absolute impact—she would have to build a model that incorporates every marketing tactic used as well as any other factor (such as seasonality) that could affect sales.”
Martin Kihn, Gartner Analyst
The catch with GA’s model comparison tool is hidden in the “any other factor” caveat. No attribution model exactly represents how visitors engage your digital channels, though some models may be closer to reality than others.
GA’s model comparison tool is an excellent way to understand the pathmarks along your customers’ digital journeys toward conversion (social media, the different sections of your website, targeted advertising etc.) by comparing different models.
However, it simply isn’t flexible enough to factor in endless variables such as seasonal sales bumps, supply shortages, store closures and openings etc. To do this kind of modeling quickly, you’ll need to use a BI tool.
Or, as Gartner recommends in Market Guide for Attribution and Marketing Mix Modeling (available to Gartner clients), you can simply outsource the process of building models to a consulting firm that specializes in this niche area of analytics, since “most marketing leaders can only execute simple projects in-house.”
Recommendation: GA’s comparison tool can get you started with attribution modeling and help you allocate spend on digital channels, but don’t start slashing headcount on your advertising team solely on the basis of what it’s telling you.
Further Recommendations
We’ve seen that data integration and attribution are major challenges facing digital marketers, and that GA won’t fully solve them for you. Here are some steps toward a fuller analytics ecosystem for your digital marketing efforts:
- Leverage GA as a data collection and integration platform for web data sources. If you’re under Google’s 10 million hit per month limit, GA is free. It’s the most cost-effective and robust platform out there for gathering data about web traffic, mobile app usage etc.
- Use custom dimensions and metrics primarily for data integration. Yes, you can now send CRM data to GA, along with pretty much any other kind of data. The primary reason to send this data, however, is to integrate it with data about how these customers are behaving on your site, not to analyze it. You’ll still ultimately need to perform the analysis by exporting the data from GA to a .CSV or connecting BI tools to GA via Google’s API.
- Create dashboards that further integrate GA data with other data sources. In the first part of our series on GA, we explain how third-party dashboards can expand how you visualize your GA data. Integrations that can’t be handled via custom dimensions and metrics can be solved in a dashboard tool. In other words, you don’t need to send the data to GA first if that proves to be too technically challenging.
- Use the attribution model comparison tool to begin understanding attribution, not as the final word on attribution. Yes, you do need to start analyzing the return on digital channels to allocate your marketing budget. However, the model comparison tool in GA is too limited to scientifically allocate marketing spend. Before you make big decisions such as killing off Facebook advertising efforts, make sure you’ve created a full model of how your marketing efforts interact with the rest of your business operations (sales, manufacturing etc.) in a BI tool outside GA. If you lack the in-house talent to build advanced mathematical models, consider outsourcing this task.
In order to fully leverage GA for digital marketing, you’ll need to begin identifying analytics tools that can be used alongside it. If you’re sick of spreadsheets, we offer a comprehensive listing of visual analytics tools and dashboard platforms that integrate with GA data.
