Troubleshooting Website Traffic drop with Web Analytics


Web analyst often take emotionally driven decisions in their analysis and this can often lead to unwanted issues. Identifying the exact cause of web traffic drop should be done in a highly systematic way. In order to help you troubleshoot traffic and conversion drop I want to share outcome based analysis approach.

Step#1. Analytics Intelligence (AI)

Your first step is to prepare yourself for future incidence and web traffic/conversion related issues. Google Analytics recently launched automatic intelligence alert feature. Using statistical significance analysis, the analytics system calculates and creates alerts in the “Intelligence reports” if data is out of trend.

You can also create custom alerts and setup auto email functionality when the alert is triggered. Here is a quick video on how to setup automatic and custom alerts using Google analytics intelligence.

YouTube Preview Image

Omniture Sitecatalyt offers custom alert feature with every report. Omniture’s alert feature is more advanced because it can send you email and text messages to your cell phone.

google analytics custom alerts

Step#2. Prepare control limit graph

Website traffic/conversion/bounce rate can fluctuate due to multiple uncontrollable environmental factors. You could see a spike in traffic due to a press release from top competitor or you could see a drop in conversion due to technical issues. Whatever the case is, it is your job as web analyst to identify a true trend and eliminate noise from data.

To eliminate noise, we will apply statistical control limits to visits data for the last 90 days. This will allow us to identify a true drop in traffic.

Calculate your website visits upper control limit (UCL) and lower control limit (LCL) using control chart calculator.

Sample size = 90

statistical control limits

Prepare a graph for your website visit with the control limits data. In my previous post I explained how to create the control limits graph.

The sample graph below clearly indicates that the traffic has been dropping consistently since week 3. This requires immediate segmentation of data to find out the “suspect” channel.

contro limit chart

Step#3. Create custom segments

By default, Google analytics does not offer email, banner or other campaign segments. Here is a quick video to create custom segments in Google Analytics.

YouTube Preview Image

Omniture Sitecatalyst offers paid search reporting feature. If this feature is not enabled then you can enable this feature from “Report Suite Manager”.

sitecatalyst paid search detection

I always use Omniture Discover because you can create custom segments on the fly and perform detailed analysis.

After creating custom segments, collect the traffic data from all channels and plot a multi channel line graph in excel using weekly buckets.

multi channel traffic

The line graph will reveal the exact cause of the traffic drop. In the sample graph above, notice how total website traffic is trending closely to the paid search visits and indicating that the paid search is the cause of the drop in traffic.

Once you identify the channel it will be easier to conduct further analysis for that particular channel.

Step#4. Channel Segment analysis

Here are the questions you need to ask in order to find and fix the traffic issue with respective channels.

Paid search

a. Has there been a change to the paid search budget?

b. Has there been a change to how the budget is allocated? (day parting, weekday/weekend parting, holiday budget)

c. Are you being penalized by Google? (quality score issues, ad copy, brand keywords)

d. Have you been aggressively outbid by your competitor? (bid adjustment, keyword level bidding)

Organic search

a. Has there been a big drop in the number of pages indexed in Google? ( find the exact number of pages by using command in Google search. You can contact Google for reconsideration of the dropped pages and in most cases they will accept your request)

b. Have you seen a big drop in ranking for your head keywords?

c. Have you recently made changes to your site linking structure?

d. Are currently involved in unethical link building practices? (please remove all the unethical and unnatural links to your website as soon as you can)

Referring Site

a. Identify the biggest traffic driving referral site and find out if they recently updated their site and missed your link.

b. Have you annoyed your social media follows by a negative comment?

Direct Traffic

a. Social media activities can influence your direct traffic.

b. Has there been a negative press about your company?

c. Do you see a drop in brand term keywords in paid and organic search?

Hope you enjoyed this post. Email me if you have any questions. Please share your thoughts, insights and comments because it really helps me write more effectively.

Leave a Reply

2 Comment threads
0 Thread replies
Most reacted comment
Hottest comment thread
0 Comment authors
Top 5 Power User Excel Tips & Tricks for Web Analytics ProductivityTop 10 Outrageous Web Analytics Myths..Debunked! Recent comment authors
newest oldest most voted
Notify of

[…] getting insights is the key to web analytics success. Focus on direction, build trends, and apply statistical significance to the trends to get meaningful […]


[…] have discussed control limits on this blog before, but I could not resist talking about it once again because it is extremely […]