No matter what the task is—creating an editorial calendar, tracking site metrics, auditing content, or planning a migration—here at Tendo we always start with the data. How much content do you have? Where is it? How much traffic? There are scores of data points available, and with the growth of self-service analytics and business intelligence apps, marketers have a wealth of tools to gather and analyze data.
However, it’s critical to make sure your raw data is clear and that you’re reading it correctly before you start basing assumptions and making recommendations on it. Here are five issues to keep in mind when you’re turning your information into insights.
1. Understand the Role of Canonical URLs
Improper tracking of canonical URLs is a primary cause of misleading data. There are many ways to address the same web page, but the canonical URL is the single authoritative address. For example, these three URLs point to the same page:
Without a canonical URL, search engines would track these as separate pages, making it harder to have a clear picture of the traffic to this page. The SEO page ranking would suffer as well.
The answer to this problem is the rel=canonical meta tag, which is defined in the <head> section of the page. Every page should have one to ensure that search engines and metrics tools properly count all your traffic and assign it to the right source.
Take a look at the source of your own pages to make sure they include this metatag. And review the data stored in the tag so you know how canonical URLs are being used on your site.
Being aware of all the noncanonical URLs is equally important, particularly when auditing, planning a migration, and tracking metrics on each URL. There may be dozens of variations of noncanonical URLs with tracking codes, such as https://tendocom.com/?sample-tracking-code. Each one would be accounted for as a separate element in an audit to track the discrete traffic, but only one file would be migrated. When you replicate that for dozens of URLs on a site, it may make migration seem like a bigger task than it really is.
2. Define Your Data Points
Even if your data is accurate, when you have a lot of similar data points it’s easy to misinterpret them. Consider pageviews vs. unique views vs. sessions. Are you reading each of them correctly?
Every time you fire up a new browser window to visit a site, that is a session. When you first opened this page during your current session, that was a pageview. If you refresh the page or visit another page and then come back to this page, those will be logged as additional pageviews.
As long as you kept the same browser window open, all those pageviews would only count as one unique view. In Google Analytics, even if you open this page in a new window up to 30 minutes later, it will still count as part of the original unique view.
So which metric is the most relevant? The answer depends on the content of the page and its intended purpose for your audience. If your site is full of paid ads, all those redundant refreshes that count as paid impressions are a good thing. But if your goal is to gauge the true size of your audience, then a unique pageview is usually more valuable.
Sometimes, the right metric is a combination of data points. A video hub, for example, may draw the same visitor back many times in a session to engage with different content. In that case, the number of unique views plus the number of total pageviews would tell you not just how popular the page is, but also how useful.
3. Pay Close Attention to Video Stats
Videos and other rich media present a unique set of pitfalls for data analysis. Even the Internet giants struggle to get it right—in September, Facebook apologized for miscalculating video views. YouTube’s algorithms err on the side of not counting views that appear to be coming from bot traffic, at least until the views can be determined to represent real people.
Some video stats are pretty clear: Video started? Check. Video played to the end? Check. Was someone actually watching? That’s harder to tell. It’s important to distinguish automatic functions from user-driven behavior.
For example, a video that autoplays when the page opens is not the same as someone engaging with the video by turning the sound on or pressing play. Similarly, if the video player on your page is configured to play additional content automatically at the end of a video, is that really a view you want to count?
4. Don’t Let Data Get Lost in Translation
Be careful when exporting or moving data from one tool to another or when summarizing your raw data. Even simple calculations can introduce errors. For instance, if you meant to total the SUM of all impression reports for every day of the year, but got the COUNT of all reports instead, you’d end up with “365” as the number of impressions. It happens.
If you’re relying on the IT admin team to run reports for you, it’s doubly important to check the process used to extract and transform that information. Here again, having a clear understanding of what the various metrics represent will help you ask the right questions to get the right answers.
And whenever you’re planning an audit or any new metrics project, involve the IT team from the start. They should be aware of your plan, how long the project will take, and which areas of the site you will be reviewing. That will eliminate any surprises if they are modifying resources or directories.
5. Be Aware of Under and Overreporting
Most larger sites will recognize and ignore IP addresses from within the company. Smaller sites, however, may count internal traffic along with customer traffic. It’s important to set your analytics tools to skip internal visits, particularly when testing new pages, since you may be reloading pages frequently.
You can also add any URL that you are working on to your robots.txt file. Doing that tells Google and Bing not to look at your page while you’re testing.
Don’t forget to update your settings once testing is complete—if your pages don’t show up in the traffic data, it might be because the search bots never checked.
We’re Here for You
These five tips should help you avoid misleading metrics, but even if everything looks correct, it’s wise to have a second set of eyes on your reports. It’s easy to start making assumptions if you’ve been staring at the same stats for too long.
Have any data sob stories to share? Hit us up on Twitter so we can commiserate.