One of the biggest advantages to improving online marketing is the fact that data is instantly available to inform decisions. However, sometimes data gets oversimplified and provides false “insight”.
Recently a client expressed concern that the referring links from their social media platforms was too low. It was true that only a small fraction (about 2%) of their traffic was generated from their social profiles according to the website analytics. The requested action was to get more posts on the social profiles to generate more clicks.
On the surface that seems like a reasonable response. More activity will result in more traffic. However, it was actually a misassignment of data and a lack of critically analyzing the data. Most of their social posts either had no links or were linking to a blog not hosted on their website. Increasing posts would likely have no effect on links back to the website because the posts were not set up to link to the website.
So instead of arbitrarily making posts, we focused on linking content back to the website by ensuring that at least half of the social posts included a link to the website. At the end of the month, we had a more realistic figure on social engagement with about 15% of traffic coming from social platforms.
Misaligning web analytics is where “best practices” become a liability. For example, a common best practice is that bounces are bad and they should be as low as possible. That typically is true. But what if a page is promoting a social media contest with a link to your LinkedIn page? If the page is effective, most of that pages traffic will show as a bounce. So rather than looking at the page and saying, “it’s performing poorly”, some tracking from the site to the LinkedIn post need applied to find how many people moved on to the offer and how many left.
Avoid oversimplifying your web analytics. A lot of “best practices” regularly get applied across the board with no critical thinking. This typically results in wasted effort or negative impacts to marketing campaigns, or both.