Web Analytics: How to Handle Direct Traffic, Not Provided, Dark Social and Do Not Track
For marketers, measurement is no longer an option, but a requirement. The C-Suite (which many marketers have now proudly joined) is looking for more than warm fuzzies, and is instead looking for reporting, results, and ROI.
A number of developments over the past couple of years—not surprisingly, led by Google—have shaped the way we measure, analyze and optimize digital marketing tactics. This is a good thing, as long as we understand what we can track—and what we can’t.
Google’s (Not Provided) Mystery
Take organic search keywords as an example. Despite it being just over a year since Google encrypted keywords and implemented secure search for anyone logged into any Google product when searching, many people still don’t understand the impact of this (not provided) result on their analytics. Search Engine Land explained the (not provided) issue well when released, but essentially what it means is that if people are logged into a Google product (think Gmail, YouTube, Google+, etc.) and conduct a search using Google, the referring keyword data is not passed through to your site. This is also commonly misunderstood as an issue with Google Analytics; in fact it doesn’t matter which analytics program you are using because Google search is holding the data and never passing it through to your website, regardless of whether you are using Google Analytics, Omniture, or any other application. For a significant number of the clients that I work with, (not provided) has become one of the top 3 keyword sources in organic keyword reports. This means that a significant amount of keyword data isn’t being captured; it’s just gone.
We can only control what we can control, but knowing what you are seeing in your keyword data—and just as importantly what you aren’t—is critical to your evaluation of organic search strategy and results.
Can We Shine a Light on “Dark Social?”
Similarly, an article in The Atlantic last month, Dark Social: We Have the Whole History of the Web Wrong, has sparked a good bit of conversation on how to view social media metrics within your web analytics platform. Alexis Madrigal uncovers an issue that many of us who know analytics have looked at for a while – that “Direct” traffic in your analytics reports is a big catch-all bucket, and often doesn’t mean the traffic was direct at all. Instead, a good chunk of it can come from email links, instant message sharing, and clicks from third party social media apps. In essence, anything that can’t capture a referring or tracking URL inaccurately shows as “Direct” and is not measured—thus the phrase “dark.” While it’s well worth reading the whole article, the main points are the following:
1. The sharing you see on sites like Facebook and Twitter is the tip of the ‘social’ iceberg. We are impressed by its scale because it’s easy to measure.
2. But most sharing is done via dark social means like email and IM that are difficult to measure.
3. According to new data on many media sites, 69% of social referrals came from dark social. 20% came from Facebook.
4. Facebook and Twitter do shift the paradigm from private sharing to public publishing. They structure, archive, and monetize your publications.
The phrase dark social comes from the fact that it is difficult to track and manage, so there’s not a clear prescription. However, it is still important to recognize that dark social is a factor as you look at your analytics, and you should consider looking at it in the same way as The Atlantic.
Maybe the greatest threat to your analytics and targeting data is what has been categorized as “do-not-track” (DNT) browser standards. By default, Internet Explorer 10 will be set to encourage publishers to ignore third-party tracking cookies, and other browsers will follow suit on the privacy bandwagon. The Interactive Advertising Bureau (IAB) has recently come out in opposition to this, saying it limits visitor choices and options. The eye-opening stat the IAB provided was that they estimate within a year 50% of browsers could be using DNT, meaning you may not get any analytics data on at least 50% of your visitors. This could be a fundamental shift in the way we analyze data, and is worth watching.
What Does it All Mean to Me?
While there are no “magic bullet” solutions to remedying these issues, for those of us that rely on web analytics it’s important to recognize what the data is telling you, and where you may be overvaluing or undervaluing data and potentially drawing incorrect conclusions. Understanding what your analytics program is telling you, and what it might be missing, will allow you to factor it into your decision-making process.