A little while ago, a client pointed out that the traffic to their blog had dropped significantly since launching a new site in September of 2016. The initial thought was that maybe something in the new site design had a negative impact on SEO. We decided to do a little investigating.
There definitely was a drop to overall sessions at around the end of September, with traffic coming back steadily. We decided to look further back to see if we could find any patterns.
The traffic seemed to be following a normal pattern. Traffic reached its lowest point at December of each year. So, what was the issue then? We decided to check out the blog landing page itself.
Woah. The amount of sessions and pageviews for the blog landing page had definitely fallen off. However, the site redesign happened in September of 2016, whereas the sessions and pageviews started to drop at around March. We wanted to see what the cause of the drop was if it wasn’t the site redesign.
Since the blog post URLs start with dates, we decided to search through all of the posts in 2000-2017 to see how pageviews looked. Viewing results for the individual blog post pages revealed that there was no dramatic drop in sessions or pageviews. In fact, it seemed to follow the pattern we observed for total sessions and pageviews. So, what was up with the blog landing page then?
We decided to check total referral traffic from Google to see if it was responsible for the drop.
Referral traffic from Google followed the same normal pattern, reaching its lowest points in December. So, what could the culprit be?
Direct/none traffic seemed to be the issue. It took a nose dive at the end of March.
Direct/none traffic encompasses all those users who typed in your website directly into their browser or arrived at your site from an unknown source.
We compared March 2016 to May 2016 to see actual metrics.
Although the Google traffic may seem small, that is OK. Most traffic from Google actually lands on the blog posts themselves and not the main blog landing page. The traffic to the individual blog pages was healthy. Direct traffic to the blog landing page, however, had dropped by 98.88%.
We looked at the same date range, but this time for all pages on the site. We observed that search engine traffic had risen substantially. There was a 8.42% rise in Google traffic and 2.65% rise in Bing traffic.
So, the question then became: why did sessions from the direct/none source suddenly disappear?
We had a hunch, so we took a look at ISP source for traffic:
Woah, check out those two. Nearly 100% new visitors and almost 100% bounce rate?! The average total amount spent on page was also less than one second. Something smelled fishy.
We dug further into each.
Bingo. The anomalous traffic patterns revealed themselves.
So, we have traffic that:
• Bounces 100% of the time
• Is comprised of 100% new users
• Spends an average of less than 1 second on the site
• Has dropped off the face of the earth very quickly
What does this mean? Bots.
We are willing to bet that these were some sort of bots that were scraping the blog. The drop in sessions and pageviews ended up being the result of these guys not visiting the site anymore.
The lesson learned here is to always exclude such traffic from your Analytics. Otherwise, they will have a misleading impact on your reporting and can cause havoc since they are unpredictable.